Appendix Tabs 54-61

Tab
#
Description Pdf
Page
PART THREE (Continued)
MATERIALS ON LEGISLATIVE FACTS
54 U.S. Dept. of Health and Human Servs., Administration for Children & Families,
Office of Planning, Research & Evaluation, Distribution of Abuse and Neglect by
Family Characteristics, in FOURTH NATIONAL INCIDENCE STUDY OF CHILD
ABUSE AND NEGLECT (NIS-4)
2
55 Paul R. Amato, The Impact of Family Formation Change on the Cognitive, Social,
and Emotional Well-Being of the Next Generation, 15 THE FUTURE OF
CHILDREN 75-96 (2005).
56
56 Douglas W. Allen, High school graduation rates among children of same-sex
households, 11 Rev. of Econ. Of the Household (published on-line September 26,
2013).
71
57 Mark Regnerus, How different are the adult children of parents who have same-sex
relationships? Findings from the New Family Structures Study, 41 SOCIAL
SCIENCE RESEARCH 752-70 (2012).
96
58 Mark Regnerus, Parental same-sex relationships, family instability, and subsequent
life outcomes for adult children: Answering critics of the new family structures study
with additional analyses, 41 SOCIAL SCIENCE RESEARCH 1367-77 (2012).
116
59 Loren Marks, Same-sex parenting and childrens outcomes: A closer examination of
the American psychological associations brief on lesbian and gay parenting, 41
SOCIAL SCIENCE RESEARCH 735-51 (2012).
128
60 WILLIAM C. DUNCAN, MISPLACED RELIANCE ON SOCIAL SCIENCE
EVIDENCE IN THE PROPOSITION 8 CASE, Vol. 5, No. 6, an Institute for
Marriage and Public Policy Research Brief (2012).
146
61 JOHN R. SEARLE, THE CONSTRUCTION OF SOCIAL REALITY 4-5, 27-29, 31-
37, 55-57, 59-60, 76-104, 117-120, 227-28 (1995).
152

TAB 54

ACF OPRE: Fourth National Incidence Study of Child Abuse and Neglect (NIS-4)
http://www.acf.hhs.gov/programs/opre/abuse_neglect/natl_incid/reports/natl_incid/natl_incid_dist_family_char.html
2

hospital, or in jail).
60

5.1.1 Differences in Harm Standard Maltreatment related to
Parents Employment
Table 51 shows that parents employment was related to the incidence of most Harm Standard
maltreatment categories and severities.
61
Unless noted otherwise, any differences described in the
text are statistically significant. In all cases, children with employed parents had the lowest rate of
maltreatment.
Table 51. Incidence Rates per 1,000 Children for Harm Standard Maltreatment in the NIS4 (2005
2006) Related to Parents Employment
Harm Standard
Maltreatment Category
Parent(s)
Unemployed
Parent(s)
Employed
Parent(s)
Not in the
Labor Force
Significance
of
Differences
ALL MALTREATMENT 15.9 7.6 22.6
A
,
B

ABUSE:
All Abuse 4.8 3.9 9.6
B

Emotional
Abuse
2.3 1.2 2.9
B

NEGLECT:
ALL
NEGLECT
12.1 4.1 14.8
A
,
B

Physical
Neglect
5.6 1.4 6.1
A
,
B

Emotional
Neglect
2.7
1.4 4.9

a
,
B

Educational
Neglect

5.8
1.8 4.8

B

SEVERITY
OF HARM:
Serious 6.9 3.0 11.0
A
,
B

Moderate 8.2 4.2 9.3
B

Inferred 0.7
0.3
2.2

b

A
Difference between Unemployed and Employed" is significant at the p<.05 level.
a
Difference between Unemployed and Employed" is statistically marginal (i.e., .10>p>.05).
B
Difference between Employed and Not in the Labor Force" is significant at the p<.05 level.
b
Difference between Employed and Not in the Labor Force" is statistically marginal (i.e.,
.10>p>.05).
Educational neglect is identical under the Harm and Endangerment Standards. It is included in both
tables because it is in the summary categories in both standards: All Neglect and All Maltreatment.
This estimate is less reliable because it derives from fewer than 100 sample children.
Overall Harm Standard Maltreatment, Abuse, and Neglect
Children with no parent in the labor force and those with an unemployed parent were at significantly
higher risk of Harm Standard maltreatment compared to those whose parents were employed. An
estimated 15.9 per 1,000 children with an unemployed parent suffered some form of Harm Standard
maltreatment. This rate is more than 2 times the rate for children with an employed parent (7.6
children per 1,000). The rate for children with no parent in the labor force (22.6 children per 1,000) is
almost 3 times the rate for children an employed parent.
Children with no parent in the labor force were also more likely to suffer Harm Standard abuse than
those with employed parents. The rate for children with no parent in the labor force is more than 2
000893
3

times that of children with employed parents (9.6 per 1,000 versus 3.9 per 1,000).
The incidence of overall Harm Standard neglect is significantly higher among children with an
unemployed parent and those with no parent in the labor force. Children with an unemployed parent
had almost 3 times greater risk of suffering Harm Standard neglect compared to children with
employed parents (12.1 versus 4.1 children per 1,000), and children with no parent in the labor force
had more than 3 times greater risk than children with employed parents (14.8 versus 4.1 children per
1,000).
Specific Categories of Harm Standard Abuse
Rates of specific categories of Harm Standard physical and sexual abuse did not statistically differ in
relation to parents employment. However, children with no parent in the labor force had a higher rate
of emotional abuse than those with employed parents (2.9 versus 1.2 children per 1,000), a
statistically marginal difference.
Specific Categories of Harm Standard Neglect
All three specific categories of Harm Standard neglect evidenced significant differences related to
parents employment.
Physical neglect. Children with employed parents had a significantly lower rate of Harm Standard
physical neglect (1.4 children per 1,000) compared to those in the other groups. The rate for children
with no parent in the labor force is more than 4 times higher than the rate for children with employed
parents (6.1 versus 1.4 children per 1,000), while the rate for children with unemployed parents (5.6
per 1,000) is exactly 4 times higher.
Emotional neglect. Children with no parent in the labor force had the highest risk of Harm Standard
emotional neglect (4.9 per 1,000), almost 3.5 times higher than the rate for children with an
employed parent (1.4 per 1,000), a statistically significant difference. Children with an unemployed
parent had nearly 2 times the rate of Harm Standard emotional neglect compared to those with
employed parents (2.7 versus 1.4 children per 1,000), which is a statistically marginal difference.
Educational neglect. The incidence of educational neglect was nearly 2.7 times higher for children
with no parent in the labor force compared to those with working parents (4.8 versus 1.8 children per
1,000), a significant difference. No other differences in this category are statistically reliable.
Severity of Outcomes from Harm Standard Maltreatment
As Table 51 indicates, the incidence of children who were seriously or moderately harmed by Harm
Standard maltreatment or for whom harm could be inferred related to their parents employment
status.
Serious harm. Children with an unemployed parent and those with no parent in the labor force
suffered serious harm from Harm Standard maltreatment at significantly higher rates (6.9 and 11.0
per 1,000, respectively) compared to children with working parents (3.0 per 1,000.
Moderate harm. Children with no parent in the labor force had more than 2 times the risk of
suffering moderate harm from Harm Standard abuse or neglect compared to children whose parents
were steadily employed (9.3 versus 4.2 per 1,000). (Although the rate of moderate harm for children
with unemployed parents appears almost as large, that estimate is too unreliable for the difference to
be statistically meaningful.)
Inferred harm. The incidence of children with inferred harm due to maltreatment was statistically
000894
4

marginally higher for those with no parent in the labor force compared to children with working
parents. An estimated 2.2 children per 1,000 with no parent in the labor force experienced
maltreatment sufficiently severe to permit inferring that they were harmed. This was over 7 times the
rate of inferred harm for children with employed parents (0.3 children per 1,000, respectively).
5.1.2 Differences in Maltreatment under the Endangerment
Standard Related to Parents Employment Status
Table 52 indicates the statistically meaningful differences in incidence rates based on parental
employment in all categories of maltreatment and outcome severity.
Overall Endangerment Standard Maltreatment, Abuse, and Neglect
Chapter 3 reported that an estimated 39.5 children per 1,000 nationwide experienced some form of
Endangerment Standard maltreatment. That general result is qualified by significant differences in the
incidence rates for children depending on their parents employment status.
Children with no parent in the labor force had the highest rate of Endangerment Standard
maltreatment, an estimated 57.7 per 1,000 children. This rate is more than 3 times the rate for
children whose parents were working (17.1 per 1,000). Children with an unemployed parent
experienced Endangerment Standard maltreatment at more than 2 times that of children with
employed parents (39.9 versus 17.1 per 1,000). Both of these differences are statistically significant.
The Endangerment Standard maltreatment rate for children with no parent in the labor force was also
higher than for children with an unemployed parent (57.7 versus 39.9 per 1,000), a statistically
marginal difference.
Children with no parent in the labor force had the highest the incidence of Endangerment Standard
abuse (15.2 per 1,000), 2 or more times higher than the rates for children of working parents (5.8 per
1,000) or with an unemployed parent (7.5 per 1,000).
Rates of Endangerment Standard neglect were significantly higher for children whose parents did not
have steady work. The estimated incidence of Endangerment Standard neglect was 46.4 per 1,000
children with parent in the labor force, which is 3.6 times the rate of 12.8 per 1,000 children with
employed parents. The rate for children with an unemployed parent is 2.7 times the rate for children
whose parents were consistently employed (35.0 versus 12.8 per 1,000).

000895
5

Table 52. Incidence Rates per 1,000 Children for Endangerment Standard Maltreatment in the NIS
4 (20052006) Related to Parents Employment
Harm Standard
Parent(s)
Unemployed
Parent(s)
Employed
Parent(s)
Not in the
Labor
Force
Significance
of
Differences
ALL MALTREATMENT 39.9 17.1 57.7
A
,
B
,
c

ABUSE:
All Abuse 7.5 5.8 15.2
B
,
C

Physical
Abuse
3.5 3.4 7.3
B
,
c

Sexual
Abuse
0.9 1.1 3.7
b
,
c

Emotional
Abuse
4.1 2.3 7.1
B

NEGLECT:
ALL
NEGLECT
35.0 12.8 46.4
A
,
B

Physical
Neglect
23.0 6.0 25.5
A
,
B

Emotional
Neglect
19.1 7.4 25.3
A
,
B

Educational
Neglect

5.8
1.8 4.8

B

SEVERITY
OF HARM:
Serious 7.3 3.2 11.3
A
,
B

Moderate 12.1 6.3 15.7
A
,
B

Inferred 3.0
1.1 6.0

B

Endangered 17.5 6.5 24.8
A
,
B

A
Difference between Unemployed and Employed" is significant at the p<.05 level.
B
Difference between Employed and Not in the Labor Force" is significant at the p<.05 level.
b
Difference between Employed and Not in the Labor Force" is statistically marginal (i.e.,
.10>p>.05).
C
Difference between Unemployed and Not in the Labor Force" is significant at the p<.05 level.
c
Difference between Unemployed and Not in the Labor Force" is statistically marginal (i.e.,
.10>p>.05).
Specific Categories of Endangerment Standard Abuse
In the Endangerment Standard, all specific abuse categories revealed statistically meaningful
differences in incidence related to parents employment status.
Physical abuse. Children with no parent in the labor force had the highest rate of Endangerment
Standard physical abuse (7.3 per 1,000), more than 2 times the rates for other children. The
difference is statistically significant in comparison to children with employed parents (3.4 per 1,000),
but it is statistically marginal in comparison to children with an unemployed parent (3.5 per 1,000)
because that estimate is slightly less reliable.
000896
6

Sexual abuse. Children with no parent in the labor force have a notably higher rate of Endangerment
Standard sexual abuse (3.7 per 1,000) compared to those with an unemployed parent (0.9 per 1,000)
or steadily employed parents (1.1 per 1,000). Both these differences are statistically marginal.
Emotional abuse. Compared to children with employed parents, children with no parent in the labor
force had more than 3 times the rate of Endangerment Standard emotional abuse (2.3 versus 7.1 per
1,000, respectively).
Specific Categories of Endangerment Standard Neglect
Significant differences related to parents employment emerged in all specific categories of
Endangerment Standard neglect.
Physical neglect. Children with an unemployed parent were physically neglected at a rate of 23.0 per
1,000, which is almost 4 times the rate of physical neglect for children with employed parents (6.0
children per 1,000). Children who had no parent in the labor force (25.5 per 1,000) were physically
neglected at more than 4 times the rate of children with employed parents. Both of these differences
are statistically significant.
Emotional Neglect. The differences in rates of Endangerment Standard emotional neglect follow the
consistent pattern. Children with an unemployed parent and those with no parent in the labor force
suffered maltreatment at the highest rates (19.1 and 25.5 per 1,000, respectively). Their rates are 2.6
and 3.4 times the rate for children with employed parents, respectively, who had lowest rate (7.4 per
1,000). Again, both of these differences are significant.
Educational neglect. The subgroup differences in rates of educational neglect are identical to those
given earlier, so the discussion here does not reiterate them.
Severity of Outcomes from Endangerment Standard Maltreatment
As Table 52 indicates, parents employment significantly related to the incidence of children with all
levels of harm from Endangerment Standard maltreatment.
Serious harm. Children with steadily employed parents suffered serious injury or harm from
Endangerment Standard maltreatment at a significantly lower rate (3.2 per 1,000) compared to
children with an unemployed parent or with no parent in the labor force (7.3 and 11.3, respectively).
Compared to children with employed parents, the rate of serious harm for children with an
unemployed parent is 2.3 times higher, and the rate for children with no parent in the labor force is
3.5 times higher.
Moderate harm. The incidence of moderate harm from Endangerment Standard abuse or neglect was
almost 2 times higher among children with an unemployed parent compared to the incidence among
children whose parents were employed (12.1 versus 6.3 children per 1,000). Children with no parent
in the labor force were 2.5 times more likely to be moderately harmed by Endangerment Standard
maltreatment compared to those with employed parents (15.7 versus 6.3 children per 1,000). Again,
both differences are significant.
Inferred harm. The incidence of children with inferred harm from Endangerment Standard
maltreatment was significantly greater for those with no parent in the labor force than for those with
employed parents. An estimated 6.0 per 1,000 children with no parent in the labor force experienced
maltreatment of a type sufficiently severe that harm could be inferred. This was almost 4.5 times the
rate of 1.1 per 1,000 children with working parents.
Endangered. The estimated incidence of children endangered but not yet harmed by abuse or neglect
differed significantly in relation to their parents employment. The rates of endangerment for children
000897
7

with an unemployed parent (17.5 per 1,000) and for children with no parent in the labor force (24.8
per 1,000) were both significantly higher than the rate for children with employed parents (6.5
children per 1,000); both of these differences are statistically significant.

5.2 Differences in the Incidence of Maltreatment Related to
Socioeconomic Status (SES)
The NIS4 collected data on several indicators of economic wellbeing or socioeconomic status:
household income, household povertyrelated program participation,
62
and parents education.
However, initial analyses revealed that majorities of the countable NIS4 children were missing data
on these measures.
63
In order to minimize missing values, final analyses used a composite measure of
family socioeconomic status (SES) that integrated any available information across these three
measures.
64
The composite measure defined children to be in families of low economic status (low
SES) if they were in the bottom tier on any indicator: household income was below $15,000 a year,
parents highest education level was less than high school, or any household member participated in a
povertyrelated program.
65
,
66

Analyses of changes in this composite measure since the NIS3 were not possible because the NIS3
did not obtain information about household poverty program participation and parental education.
5.2.1 Differences in Harm Standard Maltreatment Related to
Socioeconomic Status (SES)
As Table 53 reveals, household SES was significantly related to incidence rates in all categories of
Harm Standard maltreatment and levels of outcome severity.
Children in families of low SES were at significantly greater risk of Harm Standard maltreatment
overall. An estimated 22.5 children per 1,000 children in lowSES families experienced Harm Standard
maltreatment, which is more than 5 times the rate of 4.4 per 1,000 children in families that were not
of low SES.
Children in families of low SES were also at significantly greater risk of Harm Standard abuse. An
estimated 7.7 children per 1,000 children in lowSES families experienced Harm Standard abuse
compared to 2.5 per 1,000 children not in lowSES families. The incidence rate for children in low-SES
families is more than 3 times the rate for children not in low-SES families.
Children in families of low SES had a significantly higher rate of Harm Standard neglect than those not
in low-SES families (16.1 versus 2.2 per 1,000 children, respectively). Thus, the risk of Harm
Standard neglect for children in lowSES families was over 7 times the risk for children not in families
of low SES.

000898
8

Table 53. Differences Related to Family Socioeconomic Status in Incidence Rates per 1,000 Children
for Harm Standard Maltreatment in the NIS4 (20052006)
*

Harm Standard Maltreatment
Category
Children not in low
SES Families
Children in Low
SES Families
ALL MALTREATMENT 4.4 22.5
ABUSE:
All Abuse 2.5 7.7
Physical Abuse 1.5 4.4
Sexual Abuse 0.6 1.7
Emotional Abuse 0.5 2.6
NEGLECT:
ALL NEGLECT 2.2 16.1
Physical Neglect 0.8 6.9
Emotional
Neglect
0.8 3.8
Educational
Neglect

1.0 7.1
SEVERITY OF
HARM:
Serious 1.7 9.9
Moderate 2.4 11.7
Inferred 0.2
0.9
*
All differences are significant at p<.05.
The NIS4 found significant socioeconomic status differences in all categories of Harm Standard
abuse: physical, sexual, and emotional.
Physical abuse. Children in families of low SES were at significantly higher risk of Harm Standard
physical abuse compared to children not in lowSES families (4.4 versus 1.5 per 1,000, respectively).
The incidence rate for children in lowSES families is almost 3 times the rate for children not in low
SES families.
Sexual abuse. Children in families of low socioeconomic status also experienced a significantly higher
risk of Harm Standard sexual abuse. The estimated incidence rate for children in lowSES families was
1.7 per 1,000 children, which is more than 2 times the rate of 0.6 children per 1,000 children not in
lowSES families.
Emotional abuse. The incidence of emotional abuse for children in lowSES families was more than 5
times the rate for children not in families of low SES (2.6 versus 0.5 children per 1,000, respectively).
The socioeconomic subgroups had significantly different incidence rates in all specific categories of
Harm Standard neglect.
Physical neglect. Children in lowSES families had a significantly higher rate of Harm Standard
physical neglect compared to those not in families of low SES. The risk of physical neglect for children
000899
9

in families of low SES is over 8 times the rate for children not in families of low SES (6.9 versus 0.8
per 1,000 children, respectively).
Emotional neglect. Children in families of low SES had a significantly higher risk of Harm Standard
emotional neglect. The estimated incidence rate in families of low SES was 3.8 children per 1,000
compared to 0.8 per 1,000 children not in families of low SES. The incidence rate for children in low-
SES families is more than 4 times the rate for children not in low-SES families.
Educational neglect. Children in families of low SES were over 7 times more likely to experience
educational neglect than children not in families of low SES. The incidence of educational neglect was
7.1 per 1,000 children in low-SES families, whereas the rate was 1.0 per 1,000 children not in low-
SES families.
Socioeconomic status was significantly related to incidence in three levels of outcomes due to Harm
Standard maltreatment: serious, moderate, and inferred harm.
Serious harm. Children in families of low SES had a significantly higher rate of serious injury or harm
from Harm Standard maltreatment compared to children not in lowSES families. The incidence rate
for children in families of low SES was 9.9 per 1,000 children, which is more than 5 times the rate of
1.7 per 1,000 children not in lowSES families.
Moderate harm. The incidence of children moderately harmed by Harm Standard maltreatment was
11.7 per 1,000 children in lowSES families, compared to 2.4 per 1,000 children not in lowSES
families. The incidence of moderate injury or harm for children in families of low SES is nearly 5 times
the rate for children not in families of low socioeconomic status.
Inferred harm. Children in families of low SES were more than 4 times more likely than those in
families not of low SES to experience maltreatment sufficiently severe that harm could be inferred
(0.9 versus 0.2 per 1,000 children, respectively).
5.2.2 Differences in Endangerment Standard Maltreatment
Related to Socioeconomic Status (SES)
The two socioeconomic subgroups differed significantly in their risk of maltreatment under the less
stringent Endangerment Standard, as Table 54 shows. Significant differences in incidence rates
emerged for all Endangerment Standard categories of abuse and neglect and the four levels of
outcome severity.

000900
10

Table 54. Differences Related to Family Socioeconomic Status in Incidence Rates per 1,000 Children
for Endangerment Standard Maltreatment in the NIS4 (20052006)
*

Endangerment Standard
Children not in Low-
SES Families
Children in Low-
SES Families
ALL MALTREATMENT 9.5 55.1
ABUSE:
All Abuse 3.6 12.1
Physical Abuse 2.2 6.5
Sexual Abuse 0.7 2.4
Emotional Abuse 1.0 5.5
NEGLECT:
ALL NEGLECT 6.7 46.5
Physical Neglect 3.1 27.0
Emotional
Neglect
4.0 23.5
Educational
Neglect

1.0 7.1
SEVERITY OF
HARM:
Serious 1.8 10.3
Moderate 3.5 18.7
Inferred 0.7 3.4
Endangered 3.5 22.6
*
All differences are significant at p<.05.

Children in lowSES families were at significantly higher risk of Endangerment Standard maltreatment
overall compared to children not in families of low SES. An estimated 55.1 children per 1,000 children
in families of low SES experienced one or more categories of Endangerment Standard maltreatment,
which is more than 5 times the rate of 9.5 per 1,000 children not in families of low SES.
The incidence rate of Endangerment Standard abuse for children in families of low SES is more than 3
times the rate for children not in families of low SES (12.1 versus 3.6 per 1,000 children,
respectively).
Children in families of low SES had a significantly higher rate of Endangerment Standard neglect than
children not in families of low SES. The estimated incidence rate for children in lowSES families was
46.5 children per 1,000 children, which is almost 7 times the rate of 6.7 per 1,000 children not in low
SES families.
The NIS4 found that children in families in the lowest socioeconomic tier had significantly higher risk
in all three categories of Endangerment Standard abuse: physical, sexual, and emotional.
000901
11

Physical abuse. Children in families of low SES had a significantly higher risk of Endangerment
Standard physical abuse. Their estimated incidence rate, 6.5 per 1,000 children, was 3 times the rate
of 2.2 per 1,000 for children not in lowSES families.
Sexual abuse. Children in families of low SES had a significantly higher rate of Endangerment
Standard sexual abuse (2.4 children per 1,000) compared to children not in families of low SES (0.7
per 1,000 children). Thus, the incidence rate for children in families of low SES is more than 3 times
the rate for children not in families of low SES.
Emotional abuse. The incidence of Endangerment Standard emotional abuse for children in lowSES
families was more than 5 times the rate for children not in families of low socioeconomic status (5.5
versus 1.0 per 1,000 children, respectively).
Differences between the socioeconomic groups were significant in all three categories of
Endangerment Standard neglect.
Physical neglect. Children in families of low SES were significantly more likely to experience
Endangerment Standard physical neglect than children not in lowSES families. The risk of physical
neglect for children in families of low SES was over 8 times that of children not in families of low
socioeconomic status (27.0 per 1,000 children compared to 3.1 per 1,000 children not in families of
low SES).
Emotional neglect. Children in families of low SES had a significantly higher rate of Endangerment
Standard emotional neglect, 23.5 children per 1,000, compared to 4.0 children per 1,000 children not
in families of low SES. Children in lowSES families were more than 5 times more likely to experience
emotional neglect than children not in families of low SES.
Educational neglect. The subgroup differences in rates of educational neglect are identical to those
given earlier, so information is not reiterated here.
The NIS4 revealed significant differences related to family socioeconomic status in the incidence of
four levels of outcomes from Endangerment Standard maltreatment: serious harm, moderate harm,
inferred harm, and endangered.
Serious harm. An estimated 10.3 per 1,000 children in families of low SES experienced serious harm
from Endangerment Standard maltreatment, which is more than 5 times the rate for children not in
families of low SES (1.8 children per 1,000).
Moderate harm. Children in families of low SES experienced a significantly higher risk of moderate
harm from Endangerment Standard maltreatment (18.7 versus 3.5 children per 1,000). Thus, children
in families of low SES were more than 5 times more likely to experience moderate harm from
Endangerment Standard maltreatment than children not in lowSES families.
Inferred harm. The incidence of Endangerment Standard maltreatment for children in families of low
SES was almost 5 times the rate for children not in families of low SES (3.4 versus 0.7 per 1,000
children, respectively).
Endangered. Children in families of low SES were significantly more likely to be endangered, but not
yet harmed, by maltreatment. The incidence of endangerment for children in families of low SES was
22.6 per 1,000 children, which is more than 6 times the rate of 3.5 per 1,000 children whose families
were not of low SES.
000902
12

5.3 Differences in the Incidence of Maltreatment Related to
Family Structure and Living Arrangement
This section presents differences in the incidence of child maltreatment related to the combination of
family structure and parents living arrangement, including the number of parents in the household,
their relationship to the child, and their marital or cohabitating status.
The definition of parent follows that used by the U.S. Census Bureau, which includes birth parents,
adoptive parents, and stepparents. Children may live with two parents, one parent, or neither parent.
In twoparent households, parents can be both biologically related to the child or one or both may
have another legal parental relationship to the child (e.g., adoptive parent, stepparent). A child may
have two unmarried cohabiting parents, biological or with other relationships to the child. A single
parent (of any relationship to the child) may or may not have a cohabiting partner. These variations in
family structure and living arrangement classified children into six categories: (1) living with two
married biological parents, (2) living with other married parents (not both biological but both having a
legal parental relationship to the child), (3) living with two unmarried parents (biological or other), (4)
living with one parent who had an unmarried partner (not the childs parent) in the household, (5)
living with one parent who had no partner in the household, and (6) living with no parent.
67
,
68

5.3.1 Differences in the Incidence of Harm Standard Maltreatment
Related to Family Structure and Living Arrangement
Figure 51 shows the incidence rates of overall Harm Standard maltreatment, abuse, and neglect for
children in the six conditions of family structure and living arrangement.
All maltreatment. Children living with two married biological parents had the lowest rate of overall
Harm Standard maltreatment, at 6.8 per 1,000 children. This rate differs significantly from the rates
for all other family structure and living arrangement circumstances. Children living with one parent
who had an unmarried partner in the household had the highest incidence of Harm Standard
maltreatment (57.2 per 1,000). Their rate is more than 8 times greater than the rate for children
living with two married biological parents.
The incidence of Harm Standard maltreatment also is significantly higher for children living with one
parent and that parents unmarried partner than for children in three other conditions: children living
with other married parents (24.4 children per 1,000), those living with two unmarried parents (23.5
children per 1,000), and those living with a single parent with no partner in the household (28.4
children per 1,000). The risk of Harm Standard maltreatment for children whose single parent has an
unmarried partner is more than 2 times greater than the risk for children living in these other living
arrangements.

000903
13

Figure 5-1. Incidence of Harm Standard Maltreatment by Family
Structure and Living Arrangement.

[D]
Abuse. The rate of Harm Standard abuse for children living with two married biological parents (2.9
children per 1,000) is significantly lower than the rate for children living in all other conditions of
family structure and living arrangement (10.2 or more children per 1,000). Again, the highest rate
was among children living with just one parent and that parents unmarried partner (33.6 per 1,000
children). The rates in the highest and lowest risk groups differ by more than a factor of 11.
The risk of Harm Standard abuse for children whose single parent has an unmarried partner is more
than twice that of children in three other circumstances. Their risk is significantly higher than the risks
for children living with a single parent who has no cohabiting partner (10.2 per 1,000 children), for
children living with two unmarried parents (12.1 per 1,000 children), and for those who live with
neither parent (15.3 per 1,000 children). The rate of Harm Standard abuse among children whose
single parent lives with an unmarried partner is almost twice the rate for children living with other
married parents (33.6 versus 17.4 children per 1,000), a statistically marginal difference.
Children living with other married parents experienced Harm Standard abuse at a significantly higher
rate than those living with a single parent with no partner (17.4 versus 10.2 children per 1,000).
Neglect. The pattern of group differences is somewhat different for the incidence of Harm Standard
neglect. Children living with just one parent, under any living arrangement, had significantly higher
rates of Harm Standard neglect (27.0 and 19.6 per 1,000 children) than those living with two married
biological parents (4.2 per 1,000 children). The estimated rates in the singleparent conditions are
more than 4 times the rate among children living with their married biological parents. Also, children
whose single parent had no partner had a significantly higher Harm Standard neglect rate than
children living with other married parents (19.6 versus 9.3 children per 1,000).
In addition, children living with their two married biological parents experienced Harm Standard
neglect at a lower rate than children with other married legal parents, children with unmarried
parents, and children living with neither parent (4.2 versus 9.3, 12.6, and 20.4 children per 1,000,
respectively), although these differences are statistically marginal.

000904
14

Physical abuse.
69
As Figure 52 shows, the incidence of Harm Standard physical abuse was
significantly lower for children living with two married biological parents compared to children living in
all other conditions. An estimated 1.9 per 1,000 children living with two married biological parents
suffered Harm Standard physical abuse, compared to 5.9 or more per 1,000 children in other
circumstances. In addition, children whose single parent had an unmarried, livein partner were at
significantly higher risk of Harm Standard physical abuse (19.5 children per 1,000) compared to
children in 4 other arrangements: children whose single parent had no partner in the home (5.9
children per 1,000), children with other married parents (9.8 children per 1,000), children with
unmarried parents (8.2 children per 1,000), and children living with no parent (6.8 children per
1,000).
Figure 5-2. Incidence of Harm Standard Abuse by Family Structure and
Living Arrangement.

[D]
The rate of Harm Standard physical abuse was also higher for children in homes with other married
parents than for children living with a single parent who had no cohabiting partner (9.8 versus 5.9
children per 1,000), a statistically marginal difference.
Sexual abuse.
70
Children living with two married biological parents were sexually abused at a
significantly lower rate (0.5 per 1,000) than children living in all but one of the other conditions. The
exception is children living with unmarried parents whose incidence rate of sexual abuse does not
differ from that of children living with married biological parents.
In addition, children living with a single parent who had no cohabiting partner had a lower sexual
abuse rate than children living with other married parents and than children living with a single parent
with a partner in the home (2.4 versus 4.3 and 9.9 children per 1,000, respectively), both statistically
marginal differences.
Emotional abuse.
71
Only three significant differences emerged in emotional abuse rates. Children
living with other married parents and those living with a single parent, whether with or without a
partner, were emotionally abused at significantly higher rates than those living with two married
biological parents (2.9 or more children per 1,000 versus 0.8 children per 1,000, respectively). The
rates differ by a factor of more than 3; the highest rate, for children whose single parent lived with a
000905
15

partner, is more than 10 times greater than the lowest rate, for children living with two married
biological parents.
Physical neglect.
72
Figure 53 indicates that the rate of Harm Standard physical neglect was 1.8 per
1,000 for children living with two married biological parents, compared to rates of 6.3 per 1,000 for
children living with two unmarried parents, 6.5 or more per 1,000 for children living with one parent,
and 9.1 per 1,000 for children with no parent. All these differences are statistically significant.
Emotional neglect.
73
Children living with other married parents and those living with one parent
(with or without a cohabiting partner) had significantly higher rates of emotional neglect than children
living with two married biological parents. An estimated 3.9 per 1,000 children living with other
married parents suffered Harm Standard emotional neglect, as did 10.9 per 1,000 children living with
one parent with an unmarried partner and 4.9 per 1,000 children living with one parent without a
partner, compared to just 0.9 per 1,000 children with two married biological parents. Compared to the
Harm Standard emotional neglect rate for children with two married biological parents, the rate for
children whose single parent had a cohabiting partner is 12 times higher, the rate for children whose
single parent had no partner is more than 5 times higher, and the rate for children with other married
parents is over 4 times higher.
Figure 5-3. Incidence of Harm Standard Neglect by Family Structure and
Living Arrangement.

[D]
Educational neglect.
74
The incidence of educational neglect for children living with one parent
without a partner was more than 6 times higher than the rate for children living with two married
biological parents (11.9 versus 1.9 children per 1,000), a significant difference.
Figure 54 shows significant differences in three levels of severity of harm attributable to Harm
Standard maltreatment.
000906
16

Figure 5-4. Incidence of Outcomes from Harm Standard Maltreatment by
Family Structure and Living Arrangement.

[D]
Serious harm. The incidence of children who suffered serious harm due to Harm Standard
maltreatment was significantly lower for children living with two married biological parents (2.6 per
1,000 children) compared to children living with parents under all other circumstances (9.1 or more
children per 1,000).
Children living with a single parent who had a cohabiting partner had more than 2 times the risk of
suffering serious harm compared to children living with other married parents (20.8 versus 9.1 per
1,000 children), a statistically marginal difference.
Moderate harm. Children living with their married biological parents had a significantly lower rate of
moderate harm from Harm Standard maltreatment compared to children in any other condition. The
highest risk of moderate harm was among children who lived with one parent who had an unmarried
partner (33.0 children per 1,000); their risk was 8 times higher than that of children who lived with
their married biological parents, a considerable rate differential. Additionally, the rate of moderate
harm for children whose single parent was cohabiting with a partner was significantly higher compared
to the rates for children with other married parents and for children with unmarried parents (33.0
versus 11.1 or more children per 1,000). Children living with one parent with a cohabiting partner
experienced moderate harm from Harm Standard maltreatment at a reliably higher rate than those
living with one parent without a partner (33.0 versus 14.8 children per 1,000), a statistically marginal
difference.
Inferred harm.
75
Although the rates of this outcome appear small, the relative differences across the
subgroups are considerable and statistically reliable. Children living in any circumstance other than
with their two married biological parents had significantly higher risks of inferred harm from Harm
Standard maltreatment compared to children living with their two married biological parents.
Consistent with the patterns for other categories of maltreatment and outcome, the highest risk was
for children whose single parent had an unmarried partner (3.3 per 1,000). These children
experienced maltreatment that warranted the inference they were harmed at 33 times the rate of
children living with two married biological parents.

000907
17

Changes since the NIS3 in the Distribution of Harm Standard
Maltreatment Related to Family Structure
The NIS3 did not obtain information about marital status or the presence of an unmarried partner, so
analyses could only assess changes in the incidence of Harm Standard maltreatment in relation to
whether one or two parents were present in the household. Further, because the NIS3 included too
few sample children who lived with neither parent to provide reliable estimates for most maltreatment
categories,
76
the betweenstudy analyses compared changes in maltreatment rates for two categories
of children: those living with two parents and those living with a single parent.
77

Harm Standard maltreatment. Figure 55 shows significant changes since the NIS3 in the overall
incidence of Harm Standard maltreatment in relation to this binary family structure classification.
Whereas incidence rates increased for children living with one parent, they decreased for children
living with two parents. For children living with a single parent, the rate of overall Harm Standard
maltreatment increased by 30%, abuse increased by 22%, and neglect increased by 36%. At the
same time, these rates decreased for children living with both parents, by 39%, 42%, and 33%,
respectively.
Figure 5-5. Percent Changes since NIS-3 in Rates of Harm Standard
Maltreatment, Abuse, and Neglect by Family Structure.

[D]
Abuse. Figure 56 presents the changes since the NIS3 in specific categories of Harm Standard
abuse that were related to family structure. Family structure was significantly related to the changes
in rates of Harm Standard sexual abuse and emotional abuse, while for physical abuse the relationship
is statistically marginal. Here again, the figure indicates increased incidence rates for children living
with one parent and decreased rates for those living with two parents. Rates of maltreatment for
children with one parent increased for Harm Standard physical abuse by 14%, for sexual abuse by
49%, and for emotional abuse by 43%. Rates in these same categories decreased for children living
with two parents, by 24% in physical abuse, by 62% in sexual abuse, and by 48% in emotional abuse.

000908
18

Figure 5-6. Percent Changes since NIS-3 in Rates of Specific Categories of
Harm Standard Abuse by Family Structure.

[D]
Neglect. Figure 57 shows that rates of two specific categories of Harm Standard neglect changed
since the NIS3 in different ways for children living in different family structures. The relationship
between family structure and the changes in incidence rates is significant for emotional neglect and
statistically marginal physical neglect. Children with single parents had a 42% higher risk of Harm
Standard physical neglect and a 48% higher risk of emotional neglect at the time of the NIS4
compared to their levels of risk during the NIS3. Rates in these same categories decreased for
children living with both parents, by 28% in physical neglect and by 44% in emotional neglect.
Severity of outcomes from Harm Standard maltreatment. Figure 58 displays the changes since
the NIS3 in the incidence of children who suffered serious harm, moderate harm, or inferred harm
that differed significantly by family structure. The incidence of children living with single parents who
were seriously harmed by Harm Standard maltreatment increased 34%, those moderately harmed
increased 25%,
Figure 5-7. Percent Changes since NIS-3 in Rates of Specific Categories of
Harm Standard Neglect by Family Structure.

[D]
and those who were maltreated in ways that warranted the inference of harm increased by 65%.
Opposite changes occurred for children living with both parents: the incidence of children seriously
harmed decreased 37%, those moderately harmed decreased 33%, and those whose maltreatment
000909
19

warranted inferred harm decreased 77%.
Figure 5-8. Percent Changes since NIS-3 in Severity of Outcomes from
Harm Standard Maltreatment by Family Structure.

[D]
5.3.2 Differences in the Incidence of Endangerment Standard
Maltreatment Related to Family Structure and Living Arrangement
All Maltreatment. Figure 59 shows the incidence rates of Endangerment Standard maltreatment for
the different family structure and living arrangement subgroups. The rate of overall Endangerment
Standard maltreatment for children living with two married biological parents (15.8 children per
1,000) is significantly lower than the rates for children in all other circumstances (51.5 or more
children per 1,000). Children living with one parent whose unmarried partner was in the household
had the highest incidence of Endangerment Standard maltreatment (136.1 children per 1,000). This is
equivalent to more than 13 per 100 children, or more than 1 child in 8 whose single parent has a
cohabiting partner in the general child population. Their risk of Endangerment Standard maltreatment
is more than 8 times higher than that of children living with two married biological parents.

000910
20

Figure 5-9. Incidence of Endangerment Standard Maltreatment by Family

[D]
Children whose single parent cohabited with a partner had significantly higher rates of Endangerment
Standard maltreatment compared to children living with other married parents, children with a single
parent who had no partner in the home, and children who lived with neither parent. When children
had a single parent living with an unmarried partner, their risk of experiencing Endangerment
Standard maltreatment was more than twice their risk in these other conditions. In addition, children
who lived with two unmarried parents had significantly higher risk of Endangerment Standard
maltreatment compared to those who lived with other married parents.
Abuse. Again, children living with two married biological parents had a significantly lower rate of this
maltreatment (4.3 per 1,000 children) than those in all other conditions of family structure and living
arrangement (15.9 or more per 1,000 children). The highest rate again occurred for children whose
single parent had a livein partner (45.8 per 1,000 children), which is more than 10 times higher than
the rate for children with married biological parents. The Endangerment Standard abuse rate for
children living with one parent who had a partner was also significantly higher than the rates for
children in all the other living arrangements. Moreover, children living with other married parents had
significantly higher risk of Endangerment Standard abuse than children living with a single parent who
had no partner.
Neglect. The incidence of Endangerment Standard neglect was significantly lower for children living
with two married biological parents (12.8 children per 1,000) compared to the risk for children living
in all other family structure and living arrangement circumstances (34.0 children or more per 1,000).
Children whose single parent had an unmarried partner experienced the highest risk of Endangerment
Standard neglect (100.8 children per 1,000), which is nearly 8 times the rate in the lowest risk group.
Children living with one parent with a cohabiting partner had a significantly higher rate of
Endangerment Standard neglect than children living with other married parents, children living with a
single parent with no partner, and children living with neither parent (100.8 versus 34.0, 51.7, and
48.7 children per 1,000, respectively). The incidence of Endangerment Standard neglect for children
living with two unmarried parents is significantly higher than the rate for children living with other
married parents (74.4 versus 34.0 children per 1,000).

000911
21

As Figure 510 shows, significant differences related to family structure and living arrangement
occurred in all specific categories of Endangerment Standard abuse.
Figure 5-10. Incidence of Endangerment Standard Abuse by Family

[D]
Physical abuse. The incidence of Endangerment Standard physical abuse was significantly lower for
children living with two married biological parents (2.5 children per 1,000) than for those living in all
the other family structures and living arrangements (9.0 or more children per 1,000). Children living
with a single parent with an unmarried partner had the highest incidence of physical abuse by far,
more than 10 times the lowest rate, and also significantly higher than the rates for children living with
other married parents, with unmarried parents, with a single parent without a partner, or with neither
parent. In addition, children living with other married parents experienced Endangerment Standard
physical abuse at a significantly higher rate than those whose single parent had no live-in partner
(15.4 versus 9.0 children per 1,000).
Sexual abuse.
78
Sexual abuse rates also differed significantly for children living with two married
biological parents compared to children living in all but one of the other conditions. The exception is
the comparison with children living with unmarried parents, whose rate of sexual abuse does not
statistically differ from the rate for children with married biological parents. Only 0.7 per 1,000
children living with two married biological parents were sexually abused, compared to 12.1 per 1,000
children living with a single parent who had an unmarried partner and at least 3.4 per 1,000 children
in the other living arrangements with different rates. In addition, children whose single parent lived
with a cohabiting partner were at significantly higher risk of Endangerment Standard sexual abuse
than those living with two unmarried parents and than those whose single parent had no live-in
partner (12.1 versus 3.2 and 3.4 children per 1,000, respectively). Children whose parent had a
cohabiting partner were also sexually abused at a higher rate than those with other married parents
(12.1 versus 5.5 children per 1,000), a statistically marginal difference.
Emotional abuse.
79
The subgroups exhibit a similar profile in their rates of Endangerment Standard
emotional abuse. This category of maltreatment occurred to 1.8 per 1,000 children who were living
000912
22

with two married biological parents, which is significantly lower than the rates for children living with
other married legal parents and for those living with just one parent under any arrangement. The rate
of 15.0 per 1,000 children living with a single parent with an unmarried partner is more than 8 times
higher than the rate for children with two married biological parents. Children whose single parent had
a cohabiting partner were also at significantly higher risk of emotional abuse than those whose single
parent had no partner (15.0 versus 5.9 children per 1,000).
Figure 511 displays the statistically meaningful differences in rates of specific categories of
Endangerment Standard neglect related to family structure and living arrangement.
Figure 5-11. Incidence of Endangerment Standard Neglect by Family

[D]
Physical neglect. The lowest incidence of Endangerment Standard physical neglect occurred for
children living with two married biological parents (6.5 children per 1,000), which is significantly lower
than the rates for children in all other living arrangements. The highest rate occurred for children
living with a single parent with a cohabiting partner (47.4 children 1,000), which is over 7 times
greater than the lowest rate. In addition, the rates of physical neglect for children living with two
unmarried parents and for those living with a single parent, with or without an unmarried partner, are
significantly higher than that for children living with other married parents.
Emotional neglect. The incidence of emotional neglect was 6.7 per 1,000 children with two married
biological parents, which is significantly lower than the rates for children in any other living
arrangements (20.3 or more per 1,000 children). Children whose single parent had an unmarried
partner again had the highest rate, at 68.2 per 1,000 children, which is a factor of more than 10 times
higher than the lowest rate. This rate is also significantly higher than the rates for children in all other
conditions. Children whose single parent had a livein partner also had a significantly higher rate than
those with other married parents, those whose single parent had no partner, and those who lived with
neither parent. Children living with two unmarried parents had the secondhighest rate of
Endangerment Standard emotional neglect (46.9 per 1,000 children), which is significantly higher than
the rates in all conditions except that in the highest-risk group (i.e., children whose single parent had
a live-in partner).
000913
23

Educational neglect. Rates of educational neglect are identical under the Harm and Endangerment
Standards, so the discussion here does not reiterate those findings.
Figure 512 shows the statistically meaningful differences related to the child's family structure and
living arrangement that emerged in the incidence of different outcomes attributable to Endangerment
Standard maltreatment.
Figure 5-12. Incidence of Outcomes from Endangerment Standard
Maltreatment by Family Structure and Living Arrangement.

[D]
Serious harm. The incidence of children who suffered serious harm from Endangerment Standard
maltreatment was significantly lower among those living with their married biological parents (2.8
children per 1,000), compared to the incidence rates for children living under any other conditions
(9.5 children or more per 1,000). Children living with a single parent who had a cohabiting partner
were at the highest risk of serious harm from Endangerment Standard maltreatment (21.5 children
per 1,000). Children with a cohabiting single parent were also at higher risk of serious harm from
Endangerment Standard maltreatment than children with other married parents (9.5 children per
1,000), but this difference is statistically marginal.
Moderate harm. The incidence of children moderately harmed by Endangerment Standard
maltreatment was significantly lower when they lived with married biological parents (6.0 children per
1,000) than under any other family arrangement (17.0 or more children per 1,000). Across these
subgroups, the highest rate, for children whose single parent cohabited with a partner (49.3 children
per 1,000) is more than 8 times higher than the lowest rate. Children whose single parent had a live
in partner were at significantly higher risk than those living with other married parents and than those
living with unmarried parents (17.0 and 17.8 children per 1,000, respectively). Also, children whose
single parent had a cohabiting partner had a higher rate of moderate harm from Endangerment
Standard maltreatment than those whose single parent had no partner, but this difference was
statistically marginal.
Inferred harm.
80
Risk of experiencing Endangerment Standard maltreatment that permitted the
inference of harm was lower for children who lived with two married biological parents (0.9 children
per 1,000) compared to those who lived in any other arrangement (5.2 or more children per 1,000).
000914
24

These differences are significant except for the comparison with children living with two unmarried
parents (7.8 children per 1,000), where the difference is statistically marginal.
In addition, children living with a single parent who cohabited with a partner had a significantly higher
rate of inferred harm from Endangerment Standard maltreatment than those with other married
parents, those with a single parent with no partner, and those living with neither parent (5.3, 5.2 and
5.5 children per 1,000, respectively).
Endangered. The rate of children who were endangered but not yet harmed by Endangerment
Standard maltreatment was significantly lower for children living with two married biological parents
compared to children living in all other family arrangements (6.1 versus 17.5 or more children per
1,000). The highest rate, 55.0 per 1,000 children who lived with a single parent with partner, is over 9
times the lowest rate. The endangerment rate for children whose single parent had a cohabiting
partner is significantly higher than the rate for all except those living with unmarried parents. Children
with unmarried parents had a significantly higher rate of endangerment than children in all other
conditions except those whose single parent had a cohabiting partner.
Changes since the NIS3 in the Distribution of Endangerment Standard
Maltreatment Related to Family Structure
Changes in Endangerment Standard maltreatment since the NIS3 significantly related to family
structure in all summary categories of maltreatment, all specific categories of abuse, two categories of
neglect, and four levels of outcome severity.
Figure 513. Percent Changes since NIS3 in Rates of Endangerment
Standard Maltreatment, Abuse, and Neglect by Family Structure.

[D]
Endangerment Standard maltreatment. As Figure 513 shows, the rates of overall maltreatment
and of neglect increased substantially for children living with one parent (by 56% and 61%,
respectively), whereas the rates for children who lived with both parents showed much smaller
changes in those categories (a 15% decrease and 2% increase, respectively). The pattern for changes
in rates of Endangerment Standard abuse is somewhat different, however. There, the rate for children
in single-parent homes showed negligible change (just a 6% increase), whereas the rate for children
living with two parents decreased substantially (by 46%).
Abuse. Figure 514, which gives the percent changes in rates for specific abuse categories,
demonstrates that the pattern of smaller changes in rates for single parent children and large rate
decreases for two-parent children carries through in all abuse categories. The rates for children in
one-parent households increased by 11% in Endangerment Standard physical abuse and by 21% in
000915
25

sexual abuse. The rate of emotional abuse for single-parent children actually decreased by 17%. By
contrast, the rates for children in two-parent homes decreased substantially in all abuse categories
physical abuse decreased by 34%, sexual abuse decreased by 60%, and emotional abuse decreased
by 55%.
Figure 514. Percent Changes since NIS3 in Rates of Specific Categories
of Endangerment Standard Abuse by Family Structure.

[D]
Neglect. Figure 515 shows that the overall neglect pattern, whereby single parent children showed a
large rate increase and two-parent children showed a lesser increase, only applied to emotional
neglect. The rate of Endangerment Standard emotional neglect increased 200% for single-parent
children, whereas it increased 58% for twoparent children. The relationship between family structure
and changes since NIS3 in rates of physical neglect was less dramatic, although still statistically
significant. Whereas the rate increased by 21% for singleparent children, it decreased by 18% for
children living with two parents.
Figure 5-15. Percent Changes since NIS3 in Rates of Specific Categories
of Endangerment Standard Neglect by Family Structure.

[D]
Severity of outcomes from Endangerment Standard maltreatment. As Figure 516 shows, the
changes in the incidence of children with serious harm, moderate harm, and inferred harm from
Endangerment Standard maltreatment, and of children who were endangered by their maltreatment
differed significantly by family structure. In all cases, the rates for children living with one parent
000916
26

increased. In contrast, the rates for children living with two parents decreased in three outcome
categories (serious, moderate, and inferred harm). Although the incidence of endangered twoparent
children increased, this rate increase was much smaller than the rate increase for singleparent
children in this outcome category (6% versus 50%, respectively).
5.4 Differences in the Incidence of Maltreatment Related to Grandparents
as Caregivers
This section describes the relationship between whether or not children had grandparents as
caregivers in their household and their incidence of abuse and neglect. The analyses classified children
on the basis of whether the information identified a grandparent as a caregiver for the children in the
household. The NIS4 only identified a grandparent as a childs caregiver under three conditions:
when the grandparent was the childs primary caregiver, when the primary caregiver did not have a
spouse or partner and the grandparent was the secondary caregiver, and when the grandparent was a
caregiver and maltreated the child.
81
Because the NIS4 did not exhaustively identify all cases where
a maltreated child had a grandparent as caregiver, these findings provide minimum estimates of the
rates of child maltreatment in grandparentcaregiver circumstances. Also note that grandparents who
are caregivers in these analyses are not necessarily the primary caregivers in the households.
Figure 5-16. Percent Changes since NIS3 in Severity of Outcomes from
Endangerment Standard Maltreatment by Family Structure.

[D]
5.4.1 Differences in Harm Standard Maltreatment Related to Grandparents
as Caregivers
Table 55 shows that two categories of Harm Standard maltreatmentoverall abuse and physical
abuserevealed statistically meaningful differences related to whether or not the child had
grandparents as caregivers in the household.
82

Harm Standard abuse. The incidence of overall Harm Standard abuse for children who had a
grandparent caregiver was lower than the rate for children with no identified grandparent caregiver
(6.1 versus 7.6 children per 1,000), a statistically marginal difference. Children with no identified
grandparent caregiver were 1.2 times more likely to experience Harm Standard abuse than children
who had a grandparent caregiver.

000917
27

Table 55. Differences in Incidence Rates per 1,000 Children for Harm Standard Maltreatment in the
NIS4 (20052006) Related to Grandparents as Caregivers
Harm Standard
Children With
Grandparents as
Caregivers
Children Without
Grandparents as
Caregivers
Significance
of Difference
ABUSE:
All Abuse 6.1 7.6
m

Physical
Abuse
3.0 4.5
*

*
The difference is significant at p.05.
m
The difference is statistically marginal (i.e., .10>p>.05).

Physical abuse. Children whose grandparent cared for them were at significantly lower risk of Harm
Standard physical abuse compared to children with no identified grandparent caregiver. An estimated
3.0 children per 1,000 who had a grandparent caregiver experienced Harm Standard physical abuse,
whereas the rate was 4.5 per 1,000 children with no identified grandparent caregiver. Thus, the risk of
Harm Standard physical abuse for children with no identified grandparent caregiver was 1.5 times the
risk for children cared for by a grandparent.
5.4.2 Differences in Endangerment Standard Maltreatment Related to
Grandparents as Caregivers
Table 56 shows the relationship between incidence rates for Endangerment Standard maltreatment
and the presence of a grandparent caregiver. Only three statistically meaningful differences emerged:
for physical abuse and for inferred and endangered outcomes resulting from Endangerment Standard
maltreatment.
Table 56. Differences in Incidence Rates per 1,000 Children for Endangerment Standard
Maltreatment in the NIS4 (20052006) Related to Grandparents as Caregivers
Endangerment Standard
Children With
Grandparents
as Caregivers
Children Without
Grandparents as
Caregivers
Significance
of Difference
ABUSE:
Physical
Abuse
5.2 6.6
m

SEVERITY
OF HARM:
Inferred 2.3
3.2
m

Endangered 12.2 15.8
*

*
The difference is significant at p.05.
m
The difference is statistically marginal (i.e., .10>p>.05).

Physical abuse. Children whose grandparent cared for them had a lower risk of Endangerment
Standard physical abuse; this is a statistically marginal difference. An estimated 5.2 children per
1,000 with a grandparent caregiver experienced Endangerment Standard physical abuse compared to
6.6 per 1,000 for children with no identified grandparent caregiver. Children with no identified
grandparent caregiver were 1.3 times as likely to experience Endangerment Standard physical abuse.
Inferred harm. Children with a grandparent caregiver had a lower risk of inferred harm from
Endangerment Standard maltreatment than children without a grandparent caregiver (2.3 versus 3.2
children per 1,000), a statistically marginal difference. Thus, children without a grandparent caregiver
experienced maltreatment that warranted the inference they were harmed at 1.4 times the rate of
children who lived with a grandparent caregiver.
83

000918
28

Endangerment. The incidence of children who were endangered, but not yet harmed, by
maltreatment was significantly lower for those who had a grandparent caregiver, at 12.2 children per
1,000, compared to children with no identified grandparent caregiver, at 15.8 children per 1,000.
Thus, children with no identified grandparent caregiver were 1.3 times more likely to experience
endangerment compared with children with a grandparent caregiver.
5.5 Family Size Differences in the Incidence of Maltreatment
Analyses examining the relationship between family size
84
and the incidence of child abuse and neglect
categorized children into one of four groups on the basis of the number of children in their family:
those in families where they were the only child, those in families with two children, three children,
and four or more children.
85
,
86

5.5.1 Family Size Differences in Harm Standard Maltreatment
The incidence of a few categories of Harm Standard maltreatment varied across the familysize
groups. Figure 517 graphs these categories, showing that they have a consistent pattern: incidence
rates are higher for children in the largest families, intermediate for only children, and lowest for
children in families with two or three children.
Harm Standard Maltreatment Categories
Overall Harm Standard maltreatment. An estimated 21.2 children in households with four or more
children suffered some form of Harm Standard maltreatment, which is equivalent to 2.1 children per
100, or 1 in 48 children in these larger families. This, the highest rate, is 1.8 times the lowest rate
(11.9 per 1,000 children in households with two children), a significant difference. In addition, the
incidence rate for only children (17.9 per 1,000 children) was 1.5 times the rate for children in
households with two children, a statistically marginal difference.
Neglect. The incidence of overall Harm Standard neglect is significantly higher among children living
in households with four or more children compared to those in households with two children. The rate
was 13.8 per 1,000 children in the larger households, which is more than twice the rate of 6.4 per
1,000 children in twochild households.
Physical abuse. The rate of physical abuse in the largest families (5.0 per 1,000 children) is higher
than the rate in twochild families (3.4 per 1,000 children), a statistically marginal difference.
Physical neglect. Children in households with four or more children suffered physical neglect at a
higher rate than those in households with three children (5.9 versus 2.6 children per 1,000,
respectively), a difference that is statistically marginal.

000919
29

Figure 517. Incidence of Harm Standard Maltreatment by Family Size.

[D]
The incidence of children who suffered serious harm or moderate harm from Harm Standard
maltreatment differed depending on the number of children in their households, as Figure 518
shows.

000920
30

Figure 518. Severity of Outcomes from Harm Standard Maltreatment by
Family Size.

[D]
Serious harm. Children in the largest households (four or more children) had a greater risk of
suffering serious harm as a result of their Harm Standard maltreatment than children in households
with just two children. An estimated 8.4 per 1,000 children in the largest families were seriously
harmed, which is nearly 1.8 times the rate for children in families with two children, where an
estimated 4.8 children per 1,000 experienced serious harm from Harm Standard maltreatment. This
difference is statistically marginal.
Moderate harm. An estimated 9.4 children per 1,000 only children had moderate injuries as a
result of their Harm Standard maltreatment, which is 1.5 times the rate of children who had moderate
injuries among those with in twochild families (6.2 per 1,000). Also, the rate of moderate harm in
the largest families (11.7 per 1,000 children) is almost twice the rate in families with two children (6.2
per 1,000). Both differences are statistically marginal.
000921
31

Maltreatment Related to Family Size
Changes in maltreatment rates since the NIS3 related to family size only for the incidence of children
with inferred harm from Harm Standard maltreatment.
87
Figure 519 displays this finding.
Inferred harm. The incidence of children with inferred harm as a result of their Harm Standard
maltreatment decreased differentially across the three family size groups in the graph, eradicating
their differences from the earlier study. The decline was greatest among children living in the largest
households, where the rate of inferred harm decreased by 82% (from 6.0 children per 1,000 in the
NIS3 to 1.1 child per 1,000 in the NIS4). Declines in the other subgroups, although noteworthy,
were less dramatic. Among only children, the incidence of inferred harm declined by 52% (from 2.5
to 1.1 children per 1,000), and among children in households with two to three children, the rate
declined by 47% (from 1.5 to 0.8 children per 1,000).

5.5.2 Family Size Differences in Endangerment Standard Maltreatment
Endangerment Standard incidence rates in several categories of maltreatment and levels of outcome
severity differed significantly by family size. Specifically, significant family size differences emerged in
the incidence of overall Endangerment Standard maltreatment, the main category of neglect, physical
neglect, emotional neglect, moderate harm, inferred harm, and endangerment. Statistically marginal
differences
Figure 519. Changes since NIS3 in the Incidence of Inferred Harm from
Harm Standard Maltreatment Related to Family Size.

[D]
related to family size occurred in the incidence of overall abuse, physical abuse, emotional abuse, and
serious harm.
000922
32

Chapter 3 indicated that an estimated 39.5 children per 1,000 nationwide experienced some form of
Endangerment Standard maltreatment. Significant differences among the incidence rates for children
living in families of different sizes qualify that general result. Figure 520 shows that the incidence is
much higher for children in the largest families and that rates are lower for children in the other
familysize groups.
Overall Endangerment Standard maltreatment. Children in families with four or more children
have a significantly higher rate of overall Endangerment Standard maltreatment compared to children
in all smaller family groups (62.9 versus 38.2 or fewer children per 1,000). The rate of Endangerment
Standard maltreatment for children in households with four or more children is 2.3 times the rate for
children in families with two children, 1.7 times the rate for only children, and 1.6 times the rate for
children in threechild households. The difference in the incidence rates for children in twochild
versus threechild households is statistically marginal.

Figure 520. Incidence of Endangerment Standard Overall Maltreatment,
Abuse, and Neglect by Family Size.

[D]
Abuse. Children in families with four or more children had a higher rate of Endangerment Standard
abuse compared to those in families with two children. The incidence rate for children in families with
four or more children is 1.5 times the rate for children in families with two children (13.9 versus 9.4
children per 1,000). This difference is statistically marginal.
Neglect. In contrast to the small and statistically marginal familysize differences in rates of
Endangerment Standard abuse, the differences in neglect rates are substantial and significant.
Children in families with four or more children had an incidence rate of 52.2 per 1,000 children, which
is significantly higher than the rate of neglect found among children in all smaller families. Children in
the largest families have an incidence rate of neglect that is almost 2.7 times the lowest rate of
neglect, an estimated 19.6 children in two-child families. The difference in rates of Endangerment
000923
33

Standard neglect for children in threechild families and those in twochild families is also significant.
Specific Categories of Endangerment Standard Abuse and Neglect
The incidence rates in four specific categories of Endangerment Standard maltreatment varied
significantly or marginally in relation to family size. Figure 521 displays these findings.

Figure 521. Incidence of Specific Categories of Endangerment Standard
Maltreatment by Family Size.

[D]
Physical abuse. The pattern of family size differences observed above in rates of overall
Endangerment Standard abuse emerged in the specific category of physical abuse. Children in larger
families had a significantly higher rate of Endangerment Standard physical abuse compared to those in
families with two children (7.8 versus 5.0 per 1,000 children). Children in families with four or more
children had a more than 1.5 times higher rate of physical abuse than those in twochild families.
Emotional abuse. The incidence of Endangerment Standard emotional abuse among children in
families with four or more children was significantly higher than among children who were the only
child. Children in the larger families had twice the rate of emotional abuse observed for only children
000924
34

(5.8 versus 2.8 per 1,000 children).
Physical neglect. The rate of Endangerment Standard physical neglect was 31.1 per 1,000 children
in families with four or more children, which is 2.3 times the rate for only children (13.3 children per
1,000), 3.1 times the rate for children in families with two children (10.0 children per 1,000), and 2.0
times the rate for those in families with three children (15.2 children per 1,000). All these differences
are statistically significant. In addition, children in families with three children had a higher rate than
those in twochild families, a statistically marginal difference.
Emotional neglect. The rate of Endangerment Standard emotional neglect was highest for children in
the largest families (27.4 children per 1,000). This rate differs significantly from the rates for children
in families with one child (13.9 children per 1,000) and families with two children (10.9 children per
1,000), and differs marginally from the rate for children in three-child families (17.7 children per
1,000). The rate of emotional neglect among children in the largest families is 2.0 times greater than
the rate for only children, 2.5 times greater than the rate for children in twochild families, and 1.5
times greater than the rate for children in threechild families. In addition, the midsize family groups
had significantly different rates of Endangerment Standard emotional neglect: children in threechild
families had 1.6 times the rate of emotional neglect compared to children in twochild families.
The incidence of children who suffered moderate harm, inferred harm, and endangerment from
Endangerment Standard maltreatment differed significantly depending on the number of children in
their households. Differences in relation to the incidence of serious harm are statistically marginal.
Figure 522 graphs these patterns.
Serious harm. The incidence of children who suffered serious harm from Endangerment Standard
maltreatment was higher in the largest families compared to the incidence of seriously harmed
children in families with two children (8.8 versus 5.0 children per 1,000). This is a statistically
Figure 522. Severity of Outcomes from Endangerment Standard
Maltreatment by Family Size.
000925
35

[D]
Moderate harm. The rates of children moderately harmed by abuse or neglect were higher in families
with four or more children (17.9 per 1,000 children) and in families with one child (13.8 per 1,000
children) than in families with two children (9.3 per 1,000 children). Compared to children in twochild
households, those in the largest families had 1.9 times the incidence of moderate harm from their
maltreatment (a significant difference) and only children had 1.3 times the rate of moderate harm (a
statistically marginal difference).
Inferred harm. The incidence of children with inferred harm as a result of Endangerment Standard
maltreatment was significantly greater in large families compared to the incidence in families with two
children. The estimated rate of 4.7 per 1,000 children in families with four or more children is more
than twice the rate of 2.3 per 1,000 children in twochild families.
Endangered. The estimated incidence of children who were endangered, but not yet harmed, by
abuse or neglect was significantly higher among children in the largest families compared to those in
families with three or fewer children. The incidence of endangered children in the largest families
(31.5 children per 1,000) is 2.6 times the rate among only children (12.0 children per 1,000), 3.0
times that in families with two children (10.5 children per 1,000), and almost 2.0 times the rate in
families with three children (16.1 children per 1,000). Moreover, children in the midsize families also
had significantly different rates: the incidence of children endangered by their maltreatment in three
child families is more than 1.5 times the incidence of endangered children in twochild families.
Maltreatment Related to Family Size
Changes since the NIS3 in the incidence of Endangerment Standard maltreatment did not differ by
family size.
5.6 Differences in the Incidence of Maltreatment Related to Metropolitan
Status (Metrostatus) of County of Residence
000926
36

The findings in this section apply a threeway classification of the metrostatus of childrens county of
residence: large (major) urban counties, other urban (including suburban) counties, and rural
counties.
88

Although the NIS3 used a similar classification, the Office of Management and Budget (OMB) recently
revised metrostatus definitions. These definitional changes, along with updated population data from
Census 2000, reclassified a number of counties in the United States.
89
The NIS4 analyses of
metrostatus differences in maltreatment rates used these revised definitions and county
classifications. However, the analyses regarding changes in metrostatus since the NIS3 used the
earlier metrostatus definitions.
5.6.1 Differences in Harm Standard Maltreatment Related to Metropolitan
Status (Metrostatus) of County of Residence
Significant and marginal differences related to county metrostatus emerged in several categories of
Harm Standard maltreatment.
90
As detailed below, incidence rates were consistently higher in rural
counties.
Harm Standard Abuse
Figure 523 displays the incidence rates for Harm Standard abuse by the metrostatus of the childs
county of residence.
Overall abuse. The incidence of overall Harm Standard abuse in rural counties was 1.7 times the rate
in major urban counties (10.8 versus 6.4 children per 1,000), a statistically significant difference.
Sexual abuse. The rate of Harm Standard sexual abuse in rural counties (2.8 per 1,000 children) was
twice the rate in urban counties (1.4 children per 1,000), a difference that is significant. The rural rate
is also 1.6 times the sexual abuse rate in major urban areas (1.8 children per 1,000), a difference that
is statistically marginal.

Figure 523. Incidence of Harm Standard Abuse by County Metrostatus
000927
37

[D]
Emotional abuse. The pattern of significantly higher maltreatment rates for children in rural areas
also applies to emotional abuse. The risk of emotional abuse for children in rural counties was 2.6
times that in major urban counties (3.4 versus 1.3 children per 1,000).
Harm Standard Neglect
Only the rate of emotional neglect differed by county metrostatus and this difference was statistically
marginal. Figure 524 displays the pattern, which conforms to the differences in other categories.
The incidence of emotional neglect among children living in rural counties (4.7 children per 1,000) is
higher than for children living in major urban counties (1.8 children per 1,000), a statistically marginal
difference. Thus, the risk of Harm Standard emotional neglect for children in rural counties was 2.6
times that of children in major urban counties.
Figure 524. Incidence of Harm Standard Neglect and of Serious Harm from
Harm Standard Maltreatment by County Metrostatus

[D]
000928
38

Serious harm. Figure 524 also shows that the incidence of children who were seriously harmed by
Harm Standard maltreatment in rural counties is 2.2 times the incidence in major urban counties
(11.5 versus 5.3 children per 1,000). This difference is statistically marginal.
Maltreatment Related to County Metrostatus
None of the changes since the NIS3 in the incidence of Harm Standard maltreatment categories or
outcomes differed by county metrostatus.
5.6.2 Differences in Endangerment Standard Maltreatment Related to
Metropolitan Status (Metrostatus) of County of Residence
Children in rural counties had higher incidence rates in all categories of Endangerment Standard
maltreatment. In nearly every category of maltreatment and level of severity, children living in rural
counties had higher rates than those in major urban counties. Also, in four maltreatment categories,
children in rural counties had higher incidence rates than those in urban counties. Across all
maltreatment categories and levels of outcome severity, the incidence of Endangerment Standard
maltreatment in major urban counties did not differ from the incidence in urban counties.
The rate of Endangerment Standard maltreatment overall was significantly higher in rural counties
than in major urban areas, with children in rural counties 2.2 times as likely to experience
Endangerment Standard maltreatment as children residing in major urban counties (68.1 versus 31.3
children per 1,000). Figure 525 shows this consistent pattern. The difference in incidence between
the rural areas and (nonmajor) urban areas (where the rate was 39.1 children per 1,000) is
statistically marginal.
Figure 525. Incidence of Endangerment Standard Maltreatment by County
Metrostatus

[D]

000929
39

The incidence of Endangerment Standard abuse among rural children is 1.7 times the incidence
among children in major urban areas. This difference (16.5 versus 9.5 per 1,000 children) is
statistically significant.
Endangerment Standard neglect revealed metrostatus differences similar to those observed for
Endangerment Standard maltreatment overall, with a rural rate of 57.4 children per 1,000, an urban
rate of 30.0 children per 1,000, and a major urban rate of 23.1 children per 1,000. Children in rural
counties were 2.5 times more likely to be neglected than children in major urban counties, a
significant difference, and they were 1.9 times as likely as children in urban counties, a statistically
The same general pattern of higher incidence rates in rural counties is evident in Figure 526, which
graphs the incidence of specific categories of Endangerment Standard abuse by county metrostatus.
Abuse by County Metrostatus.

[D]
Physical abuse. A statistically marginal difference emerged between the physical abuse rate for
children in rural counties (8.5 children per 1,000) and the rate for children in major urban counties
(5.6 children per 1,000). Children living in rural areas were 1.5 times more likely to be physically
abused than children living in the major metropolitan areas.
Sexual abuse. The differences between the rate of sexual abuse in rural areas and the rates in major
urban and urban areas are statistically marginal. Children in rural counties were about 1.5 times more
likely to be sexually abused than children living in the large metropolitan areas or in urban areas.
Emotional abuse. The incidence of emotional abuse in rural counties was significantly higher, at 6.9
children per 1,000, than the rate in major urban counties, at 3.0 children per 1,000. Thus, the children
in rural counties were 2.3 times as likely to be emotionally abused as the children in major
metropolitan areas.
Figure 527 depicts the statistically meaningful differences in specific categories of Endangerment
Standard neglect related to county metrostatus.
000930
40

Neglect by County Metrostatus.

[D]
Physical neglect. Children in rural counties were 2.8 times more likely to be physically neglected
than children in major urban counties and 2.2 times more likely than urban children. Both differences
are statistically significant.
Emotional neglect. Children who live in rural counties are also at significantly higher risk of
Endangerment Standard emotional neglect compared to children living in major urban counties. The
emotional neglect rate of 27.9 per 1,000 rural children is 2.3 times the rate of 11.9 per 1,000 children
living in major metropolitan counties.
Figure 528 shows the differences in incidence rates for outcomes from Endangerment Standard
maltreatment.

000931
41

Figure 528. Outcomes from Endangerment Standard Maltreatment by
County Metrostatus.

[D]
Figure 528. Outcomes from Endangerment Standard Maltreatment by County Metrostatus
Serious harm. The incidence of children who were seriously harmed by Endangerment Standard
maltreatment is higher in rural counties compared to major urban counties. In rural counties, children
were 2.1 times as likely to suffer serious harm from maltreatment compared to children in major
urban counties (11.8 versus 5.6 children per 1,000). This difference is statistically marginal.
Moderate harm. Children residing in rural counties had a 2.7 times greater rate of moderate harm
from maltreatment compared to children in major urban counties (27.9 versus 10.4 children per
1,000), a significant difference.
Inferred harm. The incidence of children who experienced Endangerment Standard maltreatment
sufficiently severe to permit the inference that they were harmed was significantly higher among
children in rural areas than among those in major urban counties. An estimated 4.7 children per 1,000
living in rural counties experienced maltreatment that allowed inferred harm, which is twice the rate of
2.4 per 1,000 children living in large urban counties.
Endangered. The estimated incidence of children who had been endangered, but not yet harmed, by
abuse and neglect was significantly higher for children living in rural areas compared to those living in
large metropolitan areas. The rate of endangered rural children (23.8 per 1,000) was 1.8 times the
rate for major urban area children (12.9 per 1,000).
No significant differences in severity of maltreatment outcomes occurred between children in rural and
(nonmajor) urban counties or between children in major urban counties and other urban counties.
Maltreatment Related to County Metrostatus
Similar to the findings for the Harm Standard, no significant or marginal differences emerged in the
analyses of changes in the rates of Endangerment Standard maltreatment by county metrostatus.
(back to top)

55 The measures available in the NIS3 allowed analyses to address this question for three characteristics: family
000932
42

structure, family size, and metropolitan status of the county of residence. (back)
56Appendices B and C detail all NIS4 estimates, including the estimated rates as well as totals, together with their
standard errors of estimate. (back)
57Appendix D provides the detailed results of all withinNIS4 statistical comparisons. Appendix E contains the details of
all betweenstudy comparisons. (back)
58These analyses excluded children who did not live with any parent (3% of children in the United States). (back)
59The classification applied employment information hierarchically: First, it assigned children who had any parent
unemployed during these timeframes to the Parent(s) Unemployed category. Next, it assigned children with any parent
employed to the Parent(s) Employed category. Finally, it assigned children with parent(s) not in the labor force to that
category. These analyses excluded 6% of the children countable under either definitional standard because they lived in
households with no parents present. They also excluded children who lived with parents but who were missing information
on parents labor force participation. Despite the effort to minimize the number of unknown cases by combining
information across timeframes, parents employment remained unknown for 29% of the children countable under the Harm
Standard and for 27% of those countable under the Endangerment Standard. (back)
60Incidence rate computations used the following population denominators, reflecting the number (in thousands) of
children in the general population: 8,986 children with any parent currently unemployed or unemployed in the past year,
58,218 children with parent(s) employed and none unemployed currently or during the past year, and 3,982 children with
no parent in the labor force during these time periods (U.S. Census Bureau, 2008a, 2008b, 2008c). (back)
61As noted in the previous chapter, analyses did not assess subgroup differences in fatality rates, because the total
number of sample children who died as a result of their maltreatment was fewer than 100, too few to permit reliable
subgroup estimates. (back)
62Study data forms asked whether anyone in the childs household was participating in any povertyrelated program.
Instructions defined povertyrelated programs to include subsidized school breakfasts or lunches, Temporary Assistance
for Needy Families (TANF), food stamps, public housing, energy assistance, and public assistance. (back)
63Household income was missing for 68% of the children who experienced Harm Standard maltreatment and for 66% of
those who experienced Endangerment Standard maltreatment. Household poverty program participation was missing for
58% of children who were countable under the Harm Standard and 54% of those countable under the Endangerment
Standard. Information about parents education was missing for 76% of children who were countable under the Harm
Standard and 77% of those whose maltreatment fit the Endangerment Standard. (back)
64This hybrid measure integrates information at the household and family levels. It is termed family socioeconomic
status for purposes here, since a family level measure (parents education) is the lowest level in the combined index. The
index assumes that if either household income or povertyprogram participation indicates low socioeconomic status then
this status applies to all families living in the household. (back)
65Although this strategy reduced the percentage of children with missing data to just under 50%, the composite low
socioeconomic status measure was still missing for those children who were missing data on all three component
measures: 48% of children who experienced Harm Standard maltreatment and 45% of those who suffered Endangerment
Standard maltreatment. Missing data can bias NIS findings if the cases that are missing information are predominantly in a
specific subgroup, since that would cause NIS to systematically underestimate the incidence of maltreatment in that
subgroup. To gauge the robustness of the findings on socioeconomic status, hypothetical analyses examined the worst
case scenario regarding the potential bias of the missing data, by allocating all the children still missing values on this
measure to the higher socioeconomic category (i.e., the subgroup with lower incidence rates across all maltreatment
categories). Statistically significant differences remained for half of the maltreatment categories under both the Harm and
Endangerment Standards. Specifically, the hypothetical subgroups still differed, under both definitional standards, in the
incidence of all maltreatment, emotional abuse, all neglect, physical neglect, emotional neglect, and serious harm. In
addition, the hypothetical subgroups differed significantly in their rates of experiencing moderate harm and endangerment
from Endangerment Standard maltreatment. (back)
66Calculations of incidence rates used the following population denominators, in thousands: 19,750 children in families of
low socioeconomic status and 53,885 children not in families of low socioeconomic status (U.S. Census Bureau, 2008a,
2008b, 2008c). (back)
67The incidence rate calculations used population denominators derived from the 2005 and 2006 Annual Estimates of the
Population (U.S. Census Bureau, 2008a) and the 2007 March Supplement of the Current Population Survey (U.S. Census
Bureau, 2008d). Combined information from these three sources indicated that, during the NIS4 reference periods, an
average of 44,799,000 children were living with two married biological parents; 5,152,000 children were living with other
married parents; 2,192,000 children were living with two unmarried parents; 2,081,000 children were living with one
parent who had an unmarried partner in the household, 16,962,000 children were living with one parent who had no
partner in the household, and 2,449,000 children were living with no parent. These groups represent 61%, 7%, 3% 3%,
000933
43

23% and 3% of the general child population in the United States, respectively. (back)
68In each category of maltreatment or level of harm, decisions about the significance of differences relied on the
Bonferroni critical values for t . This adjusted for the multiplicity of the comparisons involved. Appendix D gives the details
concerning the statistical tests for significance of differences in maltreatment incidence related to family structure and
living arrangement. (back)
69The estimate for children who lived with neither parent is less reliable, as it derives from fewer than 100 sample
children. (back)
70Estimates for children living with unmarried parents and for children living with neither parent are less reliable because
each derives from fewer than 100 sample children. (back)
71Estimates are less reliable for children living with unmarried parents, those whose single parent had a cohabiting
partner, and those who lived with neither parent, as each derives from fewer than 100 sample children. (back)
72Estimates are less reliable for children with other married parents, those whose single parent had no partner, and those
who lived with neither parent, since each relies on fewer than 100 sample children. (back)
73Estimates are less reliable for children with other married parents, those living with unmarried parents, those whose
single parent had a partner, and those who lived with neither parent, since each relies on fewer than 100 sample children.
(back)
74Estimated rates of educational neglect and of Harm Standard emotional neglect are unreliable for the same family
structure and living arrangement conditions. (back)
75Except for children who lived with a single parent who had no partner, estimates for children in all other conditions are
less reliable because each derives from fewer than 100 sample children. (back)
76NIS3 data include fewer than 100 sample children who meet the Harm Standard requirements. Endangerment
Standard estimates in the NIS4 derive from fewer than 100 sample children in all but two categories: overall
maltreatment and all abuse. (back)
77The definition of parent here includes stepparents and adoptive parents as well as all biological parents, regardless of
their marital status. Appendix E provides the actual estimates and results of the statistical tests. (back)
78The estimate for children living with unmarried parents is less reliable because it derives from fewer than 100 sample
children. (back)
children. (back)
children. (back)
81Incidence rate calculations used the following population denominators in thousands: 5,877 children with an identified
grandparent as a caregiver and 67,759 children without an identified grandparent (U.S. Census Bureau, 2006). (back)
82Appendix D gives the details concerning the statistical tests for the differences in incidence of maltreatment that relate
to whether a grandparent was a caregiver. (back)
83The estimate for children with a grandparent caregiver for Endangerment Standard inferred harm is less reliable
because there are fewer than 100 sample children. (back)
84As in previous NIS reports, family size reflects the number of children in the household rather than the number of
children within separate family units in a household. (back)
85Computations of incidence rates used the following population denominators, reflecting the number (in thousands) of
children in the general population: 16,791 children in onechild households, 28,919 children in twochild households,
17,413 children in threechild households, 10,511 children in fourchild households (U.S. Census Bureau, 2008a, 2008b,
2008c). (back)
86In each category of maltreatment or level of harm, decisions about the significance of differences relied on the
Bonferroni critical values for t. This adjusted for the multiplicity of the comparisons involved. Appendix D gives the details
concerning the statistical tests for significance of family size differences. (back)
87Analyses of changes since the NIS3 examined the three family size groups used in that earlier study: one child, 2 to 3
children, and 4 or more children. (back)
000934
44

88Computations of incidence rates used the following population denominators, reflecting the number of children (in
thousands) in the general population: 40,161 in major urban counties, 21,768 in urban counties and 11,706 in rural
counties. (U.S. Census Bureau, 2008a; U.S. Department of Agriculture, Economic Research Service, 2003, 2004). (back)
89In 2000, OMB published new standards for defining metropolitan areas. After applying these new standards to Census
2000 data, OMB announced the new area definitions for U.S. counties (U.S. Department of Agriculture, Economic Research
Service, 2003). The new system defines metropolitan areas for all urbanized areas regardless of total area population and
it includes outlying counties if they meet a commuting threshold of 25%, with no additional requirement. This affected the
NIS classification of urban versus rural. In addition, the Census Bureau, using updated population data, modified the
RuralUrban Continuum Codes, which distinguished among counties based on population size (Department of Agriculture,
Economic Research Service, 2004). This affected the NIS4 classification of major urban versus urban, the former being
counties in the top tier of this system (i.e., those in metro areas with populations of 1 million or more). (back)
90In each category of maltreatment or harm, decisions about the significance of differences relied on the Bonferroni
critical values for t. This adjusted for the multiplicity of the comparisons involved. Appendix D gives the details concerning
the statistical tests for significance of metrostatus differences. (back)
Table of Contents |Previous |Next
You may need the FREE Adobe Acrobat Reader or Microsoft PowerPoint Viewer to view PDF and PPT files located on this site.

OPRE Home|About OPRE|Research Conferences |Funding Announcements
Research Projects: Abuse, Neglect, Adoption & Foster Care Research|Child Care Research|Early Head Start Research|Family & Youth Services
Research|Head Start Research|Strengthening Families & Healthy Marriage Research|Welfare & Employment Research|Cross-Cutting & Other
Research|ACF Performance Plans & Reports
A-Z Document Index |Ask a Question

ACF Home|Questions?|Site Index|Contact Us|Accessibility|Privacy Policy|Freedom of Information Act| Disclaimers|Plain
Writing Act
Viewers & Players|Browser Downloads|Department of Health and Human Services|USA.gov: The U.S. government's official
web portal
Administration for Children and Families 370 L'Enfant Promenade, S.W. Washington, D.C. 20447
This is a Historical Document.

000935

TAB 55

The Impact of Family Formation Change
on the Cognitive, Social, and Emotional
Well-Being of the Next Generation
Paul R. Amato
Summary
How have recent changes in U.S. family structure affected the cognitive, social, and emotional
well-being of the nations children? Paul Amato examines the effects of family formation on
children and evaluates whether current marriage-promotion programs are likely to meet chil-
drens needs.
Amato begins by investigating how children in households with both biological parents differ
from children in households with only one biological parent. He shows that children growing
up with two continuously married parents are less likely to experience a wide range of cogni-
tive, emotional, and social problems, not only during childhood but also in adulthood. Although
it is not possible to demonstrate that family structure causes these differences, studies using a
variety of sophisticated statistical methods suggest that this is the case.
Amato then asks what accounts for the differences between these two groups of children. He
shows that compared with other children, those who grow up in stable, two-parent families
have a higher standard of living, receive more effective parenting, experience more cooperative
co-parenting, are emotionally closer to both parents, and are subjected to fewer stressful events
and circumstances.
Finally, Amato assesses how current marriage-promotion policies will affect the well-being of
children. He finds that interventions that increase the share of children who grow up with both
parents would improve the overall well-being of U.S. children only modestly, because chil-
drens social or emotional problems have many causes, of which family structure is but one. But
interventions that lower only modestly the overall share of U.S. children experiencing various
problems could nevertheless lower substantially the number of children experiencing them.
Even a small decline in percentages, when multiplied by the many children in the population,
is a substantial social benefit.
VOL . 15 / NO. 2 / FAL L 2005 75
www.futureofchildren.org
Paul R. Amato is professor of sociology at Pennsylvania State University.
05 FOC 15-2 fall05 Amato.qxp 8/4/2005 12:03 PM Page 75
000937
Case 2:13-cv-00217-RJS Document 45 Filed 10/11/13 Page 61 of 141
P
erhaps the most profound
change in the American family
over the past four decades has
been the decline in the share of
children growing up in house-
holds with both biological parents. Because
many social scientists, policymakers, and
members of the general public believe that a
two-parent household is the optimal setting
for childrens development, the decline in
such households has generated widespread
concern about the well-being of American
children. This concern has generated inter-
est among policymakers in programs and in-
terventions to increase the share of children
growing up in stable, two-parent families.
Not everyone, however, agrees with these
policies; many observers believe that it is
either inappropriate, or futile, for govern-
ment to attempt to affect childrens family
structures.
My goal in this article is to inform this debate
by addressing three questions. First, how do
children in households with only one biologi-
cal parent differ in terms of their cognitive,
social, and emotional well-being from chil-
dren in households with both biological par-
ents? Second, what accounts for the observed
differences between these two groups of chil-
dren? And finally, how might current policies
to strengthen marriage, decrease divorce,
and lower nonmarital fertility affect the well-
being of children in the United States?
Research on the Effects of
Family Structure on Children
The rise in the divorce rate during the 1960s
and 1970s prompted social scientists to inves-
tigate how differing family structures affect
children. Their research focus initially was on
children of divorced parents, but it expanded
to include out-of-wedlock children and those
in other nontraditional family structures.
Parental Divorce
Early studies generally supported the as-
sumption that children who experience
parental divorce are prone to a variety of aca-
demic, behavioral, and emotional problems.
1
In 1971, psychologists Judith Wallerstein and
Joan Kelly began an influential long-term
study of 60 divorced families and 131 chil-
dren. According to the authors, five years
after divorce, one-third of the children were
adjusting well and had good relationships
with both parents. Another group of children
(more than one-third of the sample) were
clinically depressed, were doing poorly in
school, had difficulty maintaining friend-
ships, experienced chronic problems such as
sleep disturbances, and continued to hope
that their parents would reconcile.
2
Despite these early findings, other studies in
the 1970s challenged the dominant view that
divorce is uniformly bad for children. For ex-
ample, Mavis Hetherington and her col-
leagues studied 144 preschool children, half
from recently divorced maternal-custody
families and half from continuously married
two-parent families. During the first year of
the study, the children with divorced parents
exhibited more behavioral and emotional
problems than did the children with continu-
ously married parents. Two years after di-
vorce, however, children with divorced par-
ents no longer exhibited an elevated number
of problems (although a few difficulties lin-
gered for boys). Despite this temporary im-
provement, a later wave of data collection re-
vealed that the remarriage of the custodial
mother was followed by additional problems
among the children, especially daughters.
3
Trying to make sense of this research litera-
ture can be frustrating, because the results of
individual studies vary considerably: some
suggest serious negative effects of divorce,
Pa ul R. Ama t o
76 T HE FUT URE OF CHI L DREN
000938
others suggest modest effects, and yet others
suggest no effects. Much of this inconsistency
is due to variations across studies in the types
of samples, the ages of the children, the out-
comes examined, and the methods of analy-
sis. To summarize general trends across such
a large and varied body of research, social
scientists use a technique known as meta-
analysis. By calculating an effect size for each
study (which reflects the difference between
two groups expressed in a common metric),
meta-analysis makes it possible to pool re-
sults across many studies and adjust for varia-
tions such as those noted.
4
In 1991, Bruce Keith and I published the
first meta-analysis dealing with the effects of
divorce on children.
5
Our analysis summa-
rized the results of ninety-three studies pub-
lished in the 1960s, 1970s, and 1980s and
confirmed that children with divorced par-
ents are worse off than those with continu-
ously married parents on measures of aca-
demic success (school grades, scores on
standardized achievement tests), conduct
(behavior problems, aggression), psychologi-
cal well-being (depression, distress symp-
toms), self-esteem (positive feelings about
oneself, perceptions of self-efficacy), and
peer relations (number of close friends, social
support from peers), on average. Moreover,
children in divorced families tend to have
weaker emotional bonds with mothers and
fathers than do their peers in two-parent
families. These results supported the conclu-
sion that the rise in divorce had lowered the
average level of child well-being.
Our meta-analysis also indicated, however,
that the estimated effects of parental divorce
on childrens well-being are modest rather
than strong. We concluded that these modest
differences reflect widely varying experiences
within both groups of children. Some children
growing up with continuously married parents
are exposed to stressful circumstances, such as
poverty, serious conflict between parents, vio-
lence, inept parenting, and mental illness or
substance abuse, that increase the risk of child
maladjustment. Correspondingly, some chil-
dren with divorced parents cope well, perhaps
because their parents are able to separate am-
icably and engage in cooperative co-parenting
following marital dissolution.
In a more recent meta-analysis, based on
sixty-seven studies conducted during the
1990s, I again found that children with di-
vorced parents, on average, scored signifi-
cantly lower on various measures of well-
being than did children with continuously
married parents.
6
As before, the differences
between the two groups were modest rather
than large. Nevertheless, the more recent
meta-analyses revealed that children with di-
vorced parents continued to have lower aver-
age levels of cognitive, social, and emotional
well-being, even in a decade in which divorce
had become common and widely accepted.
Other studies have shown that the differ-
ences in well-being between children with di-
vorced and children with continuously mar-
ried parents persist well into adulthood. For
example, adults who experience parental di-
vorce as a child have lower socioeconomic at-
The I mpa c t o f Fa mi l y Fo r ma t i o n Cha ng e o n t he We l l - Be i ng o f t he Ne x t Ge ne r a t i o n
VOL . 15 / NO. 2 / FAL L 2005 77
Children in divorced
families tend to have
weaker emotional bonds
with mothers and fathers
than do their peers in
two-parent families.
000939
tainment, an increased risk of having a non-
marital birth, weaker bonds with parents,
lower psychological well-being, poorer mari-
tal quality, and an elevated risk of seeing their
own marriage end in divorce.
7
Overall, the
evidence is consistent that parental divorce
during childhood is linked with a wide range
of problems in adulthood.
Children Born outside Marriage
Children born outside marriage have been
studied less frequently than have children of
divorce. Nevertheless, like children with di-
vorced parents, children who grow up with a
single parent because they were born out of
wedlock are more likely than children living
with continuously married parents to experi-
ence a variety of cognitive, emotional, and
behavioral problems. Specifically, compared
with children who grow up in stable, two-
parent families, children born outside mar-
riage reach adulthood with less education,
earn less income, have lower occupational
status, are more likely to be idle (that is, not
employed and not in school), are more likely
to have a nonmarital birth (among daugh-
ters), have more troubled marriages, experi-
ence higher rates of divorce, and report more
symptoms of depression.
8
A few studies have compared children of un-
married single parents and divorced single
parents. Despite some variation across studies,
this research generally shows that the long-
term risks for most problems are comparable
in these two groups. For example, Sara
McLanahan and Gary Sandefur, using the Na-
tional Survey of Families and Households,
found that 31 percent of youth with divorced
parents dropped out of high school, compared
with 37 percent of youth born outside mar-
riage (the corresponding figure for youth with
continuously married parents was 13 percent).
Similarly, 33 percent of daughters with di-
vorced parents had a teen birth, compared
with 37 percent of daughters born outside
marriage (the corresponding figure for daugh-
ters with continuously married parents was 11
percent).
9
Other studies that have compared
offspring in these two groups yield similar re-
sults with respect to occupational attainment,
earned income, depression, and the risk of
seeing ones own marriage end in divorce.
10
Although it is sometimes assumed that chil-
dren born to unwed mothers have little con-
tact with their fathers, about 40 percent of
unmarried mothers are living with the childs
father at the time of birth.
11
If one-third of all
children are born to unmarried parents, and
if 40 percent of these parents are cohabiting,
then about one out of every eight infants lives
with two biological but unmarried parents.
Structurally, these households are similar to
households with two married parents. And
young children are unlikely to be aware of
their parents marital status. Nevertheless,
cohabiting parents tend to be more disadvan-
taged than married parents. They have less
education, earn less income, report poorer
relationship quality, and experience more
mental health problems.
12
These considera-
tions suggest that children living with cohab-
iting biological parents may be worse off, in
some respects, than children living with two
married biological parents.
Consistent with this assumption, Susan L.
Brown found that children living with cohab-
iting biological parents, compared with chil-
dren living with continuously married par-
ents, had more behavioral problems, more
emotional problems, and lower levels of
school engagement (that is, caring about
school and doing homework).
13
Parents edu-
cation, income, psychological well-being, and
parenting stress explained mostbut not
allof these differences. In other words, un-
Pa ul R. Ama t o
000940
married cohabiting parents, compared with
married parents, had fewer years of educa-
tion, earned less income, had lower levels of
psychological well-being, and reported more
stress in parenting. These factors, in turn,
partly accounted for the elevated number of
problems among their children.
The risk of relationship dissolution also is
substantially higher for cohabiting couples
with children than for married couples with
children.
14
For example, the Fragile Families
Study indicates that about one-fourth of co-
habiting biological parents are no longer liv-
ing together one year after the childs birth.
15
Another study of first births found that 31
percent of cohabiting couples had broken up
after five years, as against 16 percent of mar-
ried couples.
16
Growing up with two continu-
ously cohabiting biological parents is rare.
Using the 1999 National Survey of American
Families, Brown found that only 1.5 percent
of all children lived with two cohabiting par-
ents at the time of the survey.
17
Similarly, an
analysis of the 1995 Adolescent Health Study
(Add Health) revealed that less than one-half
of 1 percent of adolescents aged sixteen to
eighteen had spent their entire childhoods
living with two continuously cohabiting bio-
logical parents.
18
Unresolved questions remain about children
born to cohabiting parents who later marry. If
cohabiting parents marry after the birth of a
child, is the child at any greater risk than if
the parents marry before having the child?
Correspondingly, do children benefit when
their cohabiting parents get married? To the
extent that marriage increases union stability
and binds fathers more strongly to their chil-
dren, marriage among cohabiting parents
may improve childrens long-term well-being.
Few studies, however, have addressed this
issue.
Death of a Parent
Some children live with a single parent not
because of divorce or because they were born
outside marriage but because their other par-
ent has died. Studies that compare children
who experienced the death of a parent with
children separated from a parent for other
reasons yield mixed results. The Amato and
Keith meta-analysis found that children who
experienced a parents death scored lower on
several forms of well-being than did children
living with continuously married parents.
Children who experienced a parents death,
however, scored significantly higher on sev-
eral measures of well-being than did children
with divorced parents.
19
McLanahan and
Sandefur found that children with a deceased
parent were no more likely than children
with continuously married parents to drop
out of high school. Daughters with a de-
ceased parent, however, were more likely
than teenagers living with both parents to
have a nonmarital birth.
20
Another study
found that although adults whose parents di-
vorced or never married during their child-
hood had lower levels of socioeconomic at-
tainment than did adults who grew up with
continuously married parents, adults who ex-
perienced the death of a parent as a child did
not differ from those with two continuously
married parents.
21
In contrast, Amato found
VOL . 15 / NO. 2 / FAL L 2005 79
The risk of relationship
dissolution also is
substantially higher for
cohabiting couples with
children than for married
couples with children.
000941
that all causes of separation from a parent
during childhood, including parental death,
were linked with increased symptoms of de-
pression in adulthood.
22
Although the re-
search findings are mixed, these studies sug-
gest that experiencing the death of a parent
during childhood puts children at risk for a
number of problems, but not as much as does
divorce or out-of-wedlock birth.
Discordant Two-Parent Families
Most studies in this literature have compared
children living with a single parent with a
broad group of children living with continu-
ously married parents. Some two-parent fam-
ilies, however, function better than others.
Marriages marked by chronic, overt conflict
and hostility are intact structurally but are
not necessarily good environments in which
to raise children. Some early studies com-
pared children living with divorced parents
and children living with two married but dis-
cordant parents. In general, these studies
found that children in high-conflict house-
holds experience many of the same problems
as do children with divorced parents. In fact,
some studies show that children with discor-
dant married parents are worse off than chil-
dren with divorced parents.
23
A more recent generation of long-term stud-
ies has shown that the effects of divorce vary
with the degree of marital discord that pre-
cedes divorce. When parents exhibit chronic
and overt conflict, children appear to be bet-
ter off, in the long run, if their parents split
up rather than stay together. But when par-
ents exhibit relatively little overt conflict,
children appear to be better off if their par-
ents stay together. In other words, children
are particularly at risk when low-conflict mar-
riages end in divorce.
24
In a twenty-year
study, Alan Booth and I found that the major-
ity of marriages that ended in divorce fell into
the low-conflict group. Spouses in these mar-
riages did not fight frequently or express hos-
tility toward their partners. Instead, they felt
emotionally estranged from their spouses,
and many ended their marriages to seek
greater happiness with new partners. Al-
though many parents saw this transition as
positive, their children often viewed it as un-
expected, inexplicable, and unwelcome. Chil-
dren and parents, it is clear, often have differ-
ent interpretations of family transitions.
25
Stepfamilies
Although rates of remarriage have declined in
recent years, most divorced parents eventu-
ally remarry. Similarly, many women who
have had a nonmarital birth eventually marry
men who are not the fathers of their children.
Adding a stepfather to the household usually
improves childrens standard of living. More-
over, in a stepfamily, two adults are available
to monitor childrens behavior, provide super-
vision, and assist children with everyday prob-
lems. For these reasons, one might assume
that children generally are better off in step-
families than in single-parent households.
Studies consistently indicate, however, that
children in stepfamilies exhibit more prob-
lems than do children with continuously mar-
ried parents and about the same number of
problems as do children with single parents.
26
In other words, the marriage of a single par-
ent (to someone other than the childs biolog-
ical parent) does not appear to improve the
functioning of most children.
Although the great majority of parents view
the formation of a stepfamily positively, chil-
dren tend to be less enthusiastic. Stepfamily
formation is stressful for many children be-
cause it often involves moving (generally to a
different neighborhood or town), adapting to
new people in the household, and learning
new rules and routines. Moreover, early rela-
Pa ul R. Ama t o
000942
tionships between stepparents and stepchil-
dren are often tense. Children, especially
adolescents, become accustomed to a sub-
stantial degree of autonomy in single-parent
households. They may resent the monitoring
and supervision by stepparents and react with
hostility when stepparents attempt to exert
authority. Some children experience loyalty
conflicts and fear that becoming emotionally
close to a stepparent implies betraying the
nonresident biological parent. Some become
jealous because they must share parental
time and attention with the stepparent. And
for some children, remarriage ends any lin-
gering hopes that the two biological parents
will one day reconcile.
27
Finally, stepchildren
are overrepresented in official reports of
child abuse.
28
Of course, the great majority of
stepparents are not abusive. Moreover, sur-
vey data have not supported the notion that
children in stepfamilies are more likely to be
abused than are children in two-parent fami-
lies.
29
Nevertheless, even a slight trend in
this direction would represent an additional
risk for children in stepfamilies.
Although relationships in many stepfamilies
are tense, stepparents are still able to make
positive contributions to their stepchildrens
lives. If stepfamilies survive the early crisis
stage, then close and supportive relationships
between stepparents and stepchildren often
develop. Research suggests that these rela-
tionships can serve as important resources for
childrens development and emotional well-
being.
30
The increase in nonmarital cohabitation has
focused attention on the distinction between
married-couple stepfamilies and cohabiting-
couple stepfamilies. Christine Buchanan,
Eleanor Maccoby, and Sanford Dornbusch
found that adolescents had fewer emotional
and behavior problems following divorce if
their mothers remarried than if they cohab-
ited with a partner.
31
Similarly, two studies of
African American families found that children
were better off in certain respects if they lived
with stepfathers than with their mothers co-
habiting partners.
32
In contrast, Susan Brown
found no significant differences between chil-
dren in married and cohabiting stepfamilies.
33
Although these data suggest that children may
be better off if single mothers marry their
partners rather than cohabit, the small num-
ber of studies on this topic makes it difficult
to draw firm conclusions.
Variations by Gender of Child
Several early influential studies found that
boys in divorced families had more adjust-
ment problems than did girls.
34
Given that
boys usually live with their mothers following
family disruption, the loss of contact with the
same-gender parent could account for such a
difference. In addition boys, compared with
girls, may be exposed to more conflict, re-
ceive less support from parents and others
(because they are believed to be tougher),
and be picked on more by custodial mothers
(because sons may resemble their fathers).
Subsequent studies, however, have failed to
find consistent gender differences in chil-
drens reactions to divorce.
The meta-analyses on children of divorce pro-
vide the most reliable evidence on this topic.
The Amato and Keith meta-analysis of studies
conducted before the 1990s revealed one sig-
nificant gender difference: the estimated neg-
ative effect of divorce on social adjustment
was stronger for boys than girls. In other
areas, however, such as academic achieve-
ment, conduct, and psychological adjustment,
no differences between boys and girls were
apparent.
35
In my meta-analysis of studies
conducted in the 1990s, the estimated effect
of divorce on childrens conduct problems was
VOL . 15 / NO. 2 / FAL L 2005 81
000943
stronger for boys than for girls, although no
other gender differences were apparent.
36
Why the earlier studies suggest a gender dif-
ference in social adjustment and the more re-
cent studies suggest a gender difference in
conduct problems is unclear. Nevertheless,
taken together, these meta-analyses provide
some limited support for the notion that boys
are more susceptible than girls to the detri-
mental consequences of divorce.
Variations by Race of Child
Compared with whites, African Americans
have a higher rate of marital disruption and a
substantially higher rate of nonmarital births.
Because relatively little research has focused
on this topic, however, it is difficult to reach
firm conclusions about racial differences in
childrens well-being in single-parent house-
holds. Some research suggests that the aca-
demic deficits associated with living with a
single mother are less pronounced for black
than for white children.
37
One study found
that growing up in a single-parent family pre-
dicted lower socioeconomic attainment
among white women, white men, and black
women, but not among black men.
38
McLana-
han and Sandefur found that white offspring
from single-parent families were more likely
to drop out of high school than were African
American offspring from single-parent fami-
lies.
39
African American children may thus
adjust better than white children to life in sin-
gle-parent families, although the explanation
for this difference is not clear. Other studies,
however, have found few racial differences in
the estimated effects of growing up with a sin-
gle parent on long-term outcomes.
40
Some studies suggest that stepfathers play a
particularly beneficial role in African Ameri-
can families. One study found that in African
American families (but not European Ameri-
can families), children who lived with stepfa-
thers were less likely to drop out of high
school or (among daughters) have a nonmari-
tal birth.
41
Similarly, a study of African Amer-
icans living in high-poverty neighborhoods
found that girls living with their mothers and
stepfathers were less likely than girls living
with single mothers to become sexually active
or pregnant. Interestingly, the protective ef-
fect of a stepfather held only when mothers
were married and not when they were cohab-
iting.
42
Another study yielded comparable re-
sults: among African Americans, adolescents
living with stepfathers were better off in
many respects than were adolescents living
with single mothers, but adolescents living
with cohabiting parents were worse off than
those living with single mothers.
43
The rea-
sons for these racial differences are not clear,
and future research is required to understand
how interpersonal dynamics differ in white
and African American stepfamilies.
Why Do Single-Parent Families
Put Children at Risk?
Researchers have several theories to explain
why children growing up with single parents
have an elevated risk of experiencing cogni-
tive, social, and emotional problems. Most
refer either to the economic and parental re-
sources available to children or to the stress-
ful events and circumstances to which these
children must adapt.
Economic Hardship
For a variety of reasons documented else-
where in this volume, most children living
with single parents are economically disad-
vantaged. It is difficult for poor single parents
to afford the books, home computers, and pri-
vate lessons that make it easier for their chil-
dren to succeed in school. Similarly, they can-
not afford clothes, shoes, cell phones, and
other consumer goods that give their children
status among their peers. Moreover, many live
Pa ul R. Ama t o
000944
in rundown neighborhoods with high crime
rates, low-quality schools, and few community
services. Consistent with these observations,
many studies have shown that economic re-
sources explain some of the differences in
well-being between children with single par-
ents and those with continuously married par-
ents.
44
Research showing that children do
better at school and exhibit fewer behavioral
problems when nonresident fathers pay child
support likewise suggests the importance of
income in facilitating childrens well-being in
single-parent households.
45
Quality of Parenting
Regardless of family structure, the quality of
parenting is one of the best predictors of chil-
drens emotional and social well-being. Many
single parents, however, find it difficult to
function effectively as parents. Compared
with continuously married parents, they are
less emotionally supportive of their children,
have fewer rules, dispense harsher discipline,
are more inconsistent in dispensing disci-
pline, provide less supervision, and engage in
more conflict with their children.
46
Many of
these deficits in parenting presumably result
from struggling to make ends meet with lim-
ited financial resources and trying to raise
children without the help of the other biolog-
ical parent. Many studies link inept parenting
by resident single parents with a variety of
negative outcomes among children, including
poor academic achievement, emotional prob-
lems, conduct problems, low self-esteem, and
problems forming and maintaining social re-
lationships. Other studies show that depres-
sion among custodial mothers, which usually
detracts from effective parenting, is related
to poor adjustment among offspring.
47
Although the role of the resident parent (usu-
ally the mother) in promoting childrens well-
being is clear, the nonresident parent (usually
the father) can also play an important role. In
a meta-analysis of sixty-three studies of non-
resident fathers and their children, Joan
Gilbreth and I found that children had
higher academic achievement and fewer
emotional and conduct problems when non-
resident fathers were closely involved in their
lives.
48
We also found that studies of nonresi-
dent fathers in the 1990s were more likely
than earlier studies to report positive effects
of father involvement. Nonresident fathers
may thus be enacting the parent role more
successfully now than in the past, with bene-
ficial consequences for children. Neverthe-
less, analysts consistently find that many non-
resident fathers are minimally engaged with
their children. Between one-fourth and one-
third of nonresident fathers maintain fre-
quent contact with their children, and a
roughly equal share of fathers maintains little
or no contact.
49
Interviews with children re-
veal that losing contact with fathers is one of
the most painful outcomes of divorce.
50
Children also thrive when their parents have a
cooperative co-parental relationship. When
parents agree on the rules and support one
anothers decisions, children learn that
parental authority is not arbitrary. Parental
agreement also means that children are not
subjected to inconsistent discipline when they
VOL . 15 / NO. 2 / FAL L 2005 83
Regardless of family
structure, the quality of
parenting is one of the
best predictors of childrens
emotional and social
well-being.
000945
misbehave. Consistency between parents
helps children to learn and internalize social
norms and moral values. Another benefit of a
positive co-parental relationship is the model-
ing of interpersonal skills, such as showing re-
spect, communicating clearly, and resolving
disputes through negotiation and compro-
mise. Children who learn these skills by ob-
serving their parents have positive relation-
ships with peers and, later, with intimate
partners. When childrens parents live in sep-
arate households, however, cooperative co-
parenting is not the norm. Although some
parents remain locked in conflict for many
years, especially if a divorce is involved, most
gradually disengage and communicate little
with one another. At best, most children living
with single parents experience parallel par-
enting rather than cooperative co-parenting.
51
Exposure to Stress
Children living with single parents are ex-
posed to more stressful experiences and cir-
cumstances than are children living with con-
tinuously married parents. Although scholars
define stress in somewhat different ways,
most assume that it occurs when external de-
mands exceed peoples coping resources.
This results in feelings of emotional distress,
a reduced capacity to function in school,
work, and family roles, and an increase in
physiological indicators of arousal.
52
Eco-
nomic hardship, inept parenting, and loss of
contact with a parent (as noted earlier) can
be stressful for children. Observing conflict
and hostility between resident and nonresi-
dent parents also is stressful.
53
Conflict be-
tween nonresident parents appears to be par-
ticularly harmful when children feel that they
are caught in the middle, as when one parent
denigrates the other parent in front of the
child, when children are asked to transmit
critical or emotionally negative messages
from one parent to the other, and when one
parent attempts to recruit the child as an ally
against the other.
54
Interparental conflict is a
direct stressor for children, and it can also in-
terfere with their attachments to parents, re-
sulting in feelings of emotional insecurity.
55
Moving is a difficult experience for many
children, especially when it involves losing
contact with neighborhood friends. More-
over, moves that require changing schools
can put children out of step with their class-
mates in terms of the curriculum. Children
with single parents move more frequently
than other children do, partly because of eco-
nomic hardship (which forces parents to seek
less expensive accommodation in other areas)
and partly because single parents form new
romantic attachments (as when a single
mother marries and moves in with her new
husband). Studies show that frequent moving
increases the risk of academic, behavioral,
and emotional problems for children with
single parents.
56
For many children, as noted,
the addition of a stepparent to the household
is a stressful change. And when remarriages
end in divorce, children are exposed to yet
more stressful transitions. Indeed, some
studies indicate that the number of transi-
tions that children experience while growing
up (including multiple parental divorces, co-
habitations, and remarriages) is a good pre-
dictor of their behavioral and emotional
problems as adolescents and young adults.
57
Pa ul R. Ama t o
Conflict between nonresident
parents appears to be
particularly harmful when
children feel that they are
caught in the middle.
000946
The Selection Perspective
Explanations that focus on economic hard-
ship, the quality of parenting, and exposure
to stress all assume that the circumstances as-
sociated with living in a single-parent house-
hold negatively affect childrens well-being. A
quite different explanationand the main al-
ternative to these viewsis that many poorly
adjusted individuals either never marry in the
first place or see their marriages end in di-
vorce. In other words, these people carry
traits that select them into single parent-
hood. Parents can transmit these problematic
traits to their children either through genetic
inheritance or inept parenting. For example,
a mother with an antisocial personality may
pass this genetic predisposition to her chil-
dren. Her personality also may contribute to
her marriages ending in divorce. Her chil-
dren will thus be at risk of exhibiting antiso-
cial behavior, but the risk has little to do with
the divorce. The discovery that concordance
(similarity between siblings) for divorce
among adults is higher among identical than
fraternal twins suggests that genes may pre-
dispose some people to engage in behaviors
that increase the risk of divorce.
58
If parents
personality traits and other genetically trans-
mitted predispositions are causes of single
parenthood as well as childhood problems,
then the apparent effects on children of
growing up with a single parent are spurious.
Because researchers cannot conduct a true
experiment and randomly allocate children
to live with single or married parents, it is
difficult to rule out the selection perspec-
tive. Nevertheless, many studies cast doubt
on it. For example, some have found signifi-
cant differences between children with di-
vorced and continuously married parents
even after controlling for personality traits
such as depression and antisocial behavior
in parents.
59
Others have found higher rates
of problems among children with single par-
ents, using statistical methods that adjust for
unmeasured variables that, in principle,
should include parents personality traits as
well as many genetic influences.
60
And a few
studies have found that the link between
parental divorce and childrens problems is
similar for adopted and biological chil-
drena finding that cannot be explained by
genetic transmission.
61
Another study, based
on a large sample of twins, found that grow-
ing up in a single-parent family predicted
depression in adulthood even with genetic
resemblance controlled statistically.
62
Al-
though some degree of selection still may be
operating, the weight of the evidence
strongly suggests that growing up without
two biological parents in the home increases
childrens risk of a variety of cognitive, emo-
tional, and social problems.
Implications of Policies to
Increase the Share of Children
in Two-Parent Families
Since social science research shows so clearly
the advantages enjoyed by children raised by
continuously married parents, it is no wonder
that policymakers and practitioners are inter-
ested in programs to strengthen marriage and
increase the proportion of children who grow
up in such families. Realistically speaking,
what could such programs accomplish? In
what follows, I present estimates of how they
could affect the share of children in the
United States who experience various types
of problems during adolescence.
Adolescent Family Structure and
Well-Being in the Add Health Study
To make these estimates, I used the Adoles-
cent Health Studya national long-term
sample of children in junior high and high
schoolsrelying on data from Wave I, con-
ducted in 1995. Table 1 is based on adoles-
VOL . 15 / NO. 2 / FAL L 2005 85
000947
cents responses to questions about behav-
ioral, emotional, and academic problems
specifically, whether they had repeated a
grade, been suspended from school, engaged
in delinquent behavior, engaged in a violent
altercation, received counseling or therapy
for an emotional problem, smoked cigarettes
regularly during the last month, thought
about suicide, or attempted suicide. Delin-
quency involved damaging property, shoplift-
ing, breaking into a house or building to steal
something, stealing something worth more
than $50, or taking a car without the owners
permission. Violence was defined as engaging
in a physical fight as a result of which the op-
ponent had received medical attention (in-
cluding bandaging a cut) or a fight involving
multiple people or using a weapon to
threaten someone. The results are based on
responses from more than 17,000 children
between the ages of twelve and eighteen, and
the data have been weighted to make them
nationally representative.
63
Responses are shown separately for adoles-
cents living with continuously married par-
ents and for those living with one parent only.
The results are striking. Adolescents living
with single parents consistently report en-
countering more problems than those living
with continuously married parents. Thirty
percent of the former reported that they had
repeated a grade, as against 19 percent of the
latter. Similarly, 40 percent of children living
with single parents reported having been sus-
pended from school, compared with 21 per-
cent of children living with continuously mar-
ried parents. Children in stable, two-parent
families also were less likely to have engaged
in delinquency or violence, seen a therapist
for an emotional problem, smoked during the
previous month, or thought about or at-
tempted suicide. These findings are consis-
tent with research demonstrating that chil-
dren living with continuously married
parents report fewer problems than do other
children. The increase in risk associated with
living without both parents ranged from
about 23 percent (for being involved in a vio-
lent altercation) to 127 percent (for receiving
emotional therapy).
To estimate the frequency of these problems
in the larger population, I relied on the Add
Pa ul R. Ama t o
Table 1. Family Structure and Adolescent Well-Being: Share of Adolescents Reporting
Problems in Various Scenarios
Percent
Family structure, 1995 Estimated share if family structure were the same as in
Problem Two parents One parent Combined 1980 1970 1960
Repeated grade 18.8 30.3 24.0 22.9 21.8 21.4
Suspended from school 21.2 39.8 29.6 27.9 26.0 25.4
Delinquency 36.4 44.7 40.1 39.4 38.5 38.3
Violence 36.0 44.1 39.6 38.9 38.1 37.8
Therapy 7.5 17.0 11.8 10.9 9.9 9.6
Smoked in last month 13.4 22.6 17.5 16.7 15.8 15.5
Thought of suicide 11.3 14.5 12.7 12.5 12.1 12.0
Attempted suicide 1.7 2.8 2.2 2.1 2.0 1.9
Source: National Study of Adolescent Health, 1995. See text for details.
000948
Health finding that 55 percent of adolescents
between the ages of twelve to eighteen lived
with both biological parents at the time of the
survey. Given that rates of divorce and non-
marital births have not changed much since
the mid-1990s, this figure is probably close to
the current figure, and it is nearly identical to
the estimate provided by Susan Brown from
the 1999 National Survey of American Fami-
lies. (Because most children in the sample
were younger than eighteen and could still
experience a parental divorce or death before
reaching adulthood, these results are consis-
tent with the projection that about half of all
children will live continuously with both bio-
logical parents until adulthood.) The third
column in table 1 shows the estimated share
of adolescents in the U.S. population who ex-
perience each problem, based on the data in
the first two columns.
64
How would increasing the share of children
growing up in stable, two-parent families af-
fect the overall levels of these problems in
the population? To provide estimates, I con-
sidered three levels of social change. The
fourth column in table 1 provides estimates
of adolescent outcomes if the share of adoles-
cents living with two biological parents were
the same as it was in 1980, the year in which
the share of marriages ending in divorce
reached its peak but before the large increase
in nonmarital births during the 1980s and
early 1990s. The fifth column provides esti-
mates of adolescent outcomes if the share of
adolescents living with continuously married
parents were the same as it was in 1970, the
year just before the massive increase in di-
vorce rates during the 1970s. The final col-
umn provides estimates of adolescent out-
comes if the share of adolescents living with
continuously married parents were the same
as it was in 1960, a period of relative family
stability in the United States.
65
Column four shows that if the share of ado-
lescents living with two biological parents
were the same today as it was in 1980, the
share of adolescents repeating a grade would
fall from 24 percent to about 23 percent.
Similarly, if the share of adolescents living
with two biological parents returned to its
1970 level, the share of adolescents repeating
a grade would fall to about 22 percent. Fi-
nally, if the share of adolescents living with
two biological parents increased to its 1960
level, the share of adolescents repeating a
grade would fall to 21 percent.
How is it that increasing the share of children
growing up with continuously married par-
ents has such a relatively small effect on the
share of children experiencing these prob-
lems? The explanation is that many children
living with continuously married parents also
experience these problems. In general, these
findings, which are likely to disappoint some
readers, are consistent with a broad, sociolog-
ical understanding of human behavior. Most
behaviors are determined by numerous so-
cial, cultural, individual, and biological fac-
tors. No single variable, such as family struc-
ture, has a monolithic effect on childrens
development and behavior. Although increas-
ing the share of children growing up in sta-
ble, two-parent families would lower the inci-
dence of all the problems shown in table 1,
clearly it is not a panacea for the problems
confronting our nations youth.
Individual versus Public Health
Perspectives
Whether one views the estimated changes in
table 1 as small or big depends in large part
on whether one adopts an individual perspec-
tive or a public health perspective. Attempts
during the past twenty years by public health
authorities to address cholesterol-related
health problems help to illustrate this distinc-
VOL . 15 / NO. 2 / FAL L 2005 87
000949
tion. Many epidemiological and clinical stud-
ies have shown that a high level of blood cho-
lesterol is a risk factor for cardiovascular dis-
ease. How large is the estimated effect of
cholesterol on cardiovascular disease? Con-
sider a group of male nonsmokers age fifty
with normal blood pressure. Men in this
group with high total cholesterol (defined as
250 mg/dL) have a 7 percent chance of suf-
fering a heart attack during the next decade.
In comparison, men in this group with low
total cholesterol (defined as 190 mg/dL) have
only a 4 percent chance. In other words, de-
creasing total cholesterol from a dangerous
level to a safe level would lower the risk of
having a heart attack for men in this group by
3 percentage points. Based on projections
like these, public health authorities have en-
couraged people with high cholesterol to
lower their cholesterol by eating fewer foods
high in saturated fat and cholesterol, losing
weight, and exercising more often. Physicians
often recommend supplementing these
lifestyle changes with cholesterol-lowering
medications, such as statin drugs.
66
Seen from a different perspective, however,
93 percent of men age fifty with high total
cholesterol will not suffer a heart attack in
the next decade. There are only 7 chances in
100 that a particular man will have a heart at-
tack, and even if he lowers his cholesterol, he
still has 4 chances in 100 of suffering a heart
attack. In other words, all the required
changes in lifestyle, plus the use of medica-
tions, will lower his chances of a heart attack
by only 3 chances out of 100. An individual
man with high cholesterol, therefore, may
well wonder if is worth the effort to change
his lifestyle and take medication. At the pop-
ulation level, however, with more than 9 mil-
lion men in the United States in their early
fifties, a 3 percentage point reduction in
heart attacks would be seen as a major public
health achievement, because it would mean a
quarter of a million fewer heart attacks in this
group over a decade.
67
The cholesterol example is relevant to under-
standing the effects of growing up without
both parents in the household. The increase
in the risk of cardiovascular disease associ-
ated with high blood cholesterol is compara-
ble in many respects to the increase in the
risk of behavioral, emotional, and academic
problems associated with growing up in a sin-
gle-parent household. For example, the in-
crease in heart attacks associated with high
blood cholesterol represents a 75 percent in-
crease in risk([7 4]/4) x 100a figure
comparable to the increased risk associated
with single parenthood and repeating a
grade, being suspended from school, receiv-
ing therapy, or attempting suicide. Adopting
a public health view and considering the
number rather than the percentage of adoles-
cents who might be affected helps put these
findings in perspective.
In 2002 there were about 29 million children
in the United States between the ages of
twelve and eighteenthe age range covered
in table 1.
68
Table 2 indicates that nearly 7
million children in this age group will have
repeated a grade. Increasing the share of
adolescents living with two biological parents
to the 1980 level, as illustrated in the second
column of the table, suggests that some
300,000 fewer children would repeat a grade.
Correspondingly, increasing the share of ado-
lescents living with two biological parents to
the 1970 level, as illustrated in the third col-
umn, would mean that 643,264 fewer chil-
dren would repeat a grade. Finally, increasing
the share of adolescents in two-parent fami-
lies to the 1960 level suggests that nearly
three-quarters of a million fewer children
would repeat a grade. Similarly, increasing
Pa ul R. Ama t o
000950
marital stability to its 1980 level would result
in nearly half a million fewer children sus-
pended from school, about 200,000 fewer
children engaging in delinquency or violence,
a quarter of a million fewer children receiv-
ing therapy, about a quarter of a million
fewer smokers, about 80,000 fewer children
thinking about suicide, and about 28,000
fewer children attempting suicide. Seen from
this perspective, restoring family stability to
levels of a few decades ago could dramati-
cally affect the lives of many children. More-
over, although the estimated decline in the
share of children encountering these prob-
lems in table 1 is modest, increasing the
number of children growing up with both
parents would simultaneously improve all
these outcomes, as well as many other out-
comes not considered in these tables.
General Conclusion
My goal in this paper has been to inform the
marriage debate by addressing three funda-
mental questions. First, how do children in
households with only one biological parent
differ from children in households with both
biological parents, in terms of their cognitive,
social, and emotional well-being? Research
clearly demonstrates that children growing
up with two continuously married parents are
less likely than other children to experience a
wide range of cognitive, emotional, and social
problems, not only during childhood, but also
in adulthood. Although it is not possible to
demonstrate that family structure is the
cause of these differences, studies that have
used a variety of sophisticated statistical
methods, including controls for genetic fac-
tors, suggest that this is the case. This distinc-
tion is even stronger if we focus on children
growing up with two happily married biolog-
ical parents.
Second, what accounts for the observed dif-
ferences between these two groups of chil-
dren? Compared with other children, those
who grow up in stable, two-parent families
have a higher standard of living, receive more
effective parenting, experience more cooper-
ative co-parenting, are emotionally closer to
both parents (especially fathers), and are sub-
jected to fewer stressful events and circum-
stances.
And third, how might current policies to
strengthen marriage, decrease the rate of di-
vorce, and lower nonmarital fertility affect
the overall well-being of American children?
VOL . 15 / NO. 2 / FAL L 2005 89
Table 2. Well-Being of Adolescents Aged Twelve to Eighteen, 2002 Estimates
Estimated change based on two-parent families in
Problem 2002 estimate 1980 1970 1960
Repeated grade 6,948,530 299,968 643,264 746,587
Suspended from school 8,570,096 485,165 1,040,410 1,207,523
Delinquency 11,632,086 216,498 464,269 538,841
Violence 11,490,072 211,282 453,082 525,857
Therapy 3,412,678 247,799 531,392 616,745
Smoked in last month 5,083,513 239,974 514,611 597,269
Thought of suicide 3,692,358 83,469 178,995 207,746
Attempted suicide 636,164 28,693 61,530 71,413
Source: Authors estimates based on data from the National Study of Adolescent Health, 1995. See text for details.
000951
The projections in tables 1 and 2 suggest that
increasing the share of children who grow up
with continuously married parents would
improve the overall well-being of U.S. chil-
dren only modestly. The improvements are
relatively small because problems such as
being suspended from school, engaging in
delinquent behavior, and attempting suicide
have many causes, with family structure
being but one.
What are the policy implications of these
findings? First, interventions that increase
the share of children growing up with two
continuously married biological parents will
have modest effects on the percentage of
U.S. children experiencing various problems,
but could have substantial effects on the
number of children experiencing them. From
a public health perspective, even a modest
decline in percentages, when multiplied by
the large number of children in the popula-
tion, represents a substantial social benefit.
That children living in stepfamilies do not
tend to have better outcomes, on average,
than children growing up in single-parent
families suggests that interventions to
strengthen marital quality and stability would
be most profitable if focused on parents in
first marriages. Similarly, interventions to
strengthen relationships and encourage mar-
riage among cohabiting couples with children
would be most profitable if focused on cou-
ples with a first child, rather than couples
with children from prior relationships.
U.S. policymakers also should acknowledge
that returning to substantially lower rates of
divorce and nonmarital childbearing, al-
though a worthwhile goal, is not realistic, at
least in the short term. Although policy inter-
ventions may lower the rate of divorce and
nonmarital childbearing, many children will
continue to grow up with a single parent.
This stubborn fact means that policies for im-
proving childrens well-being cannot focus
exclusively on promoting marriage and
strengthening marital stability. These policies
must be supplemented by others that im-
prove economic well-being, strengthen par-
ent-child bonds, and ease the stress experi-
enced by children in single-parent and
stepparent households. Such programs would
provide parent education classes for divorc-
ing parents, increase the minimum wage and
the earned income tax credit for poor work-
ing parents, establish paternity and increase
the payment of child support, and improve
the quantity and quality of time that nonresi-
dent parents, especially fathers, spend with
their children.
The importance of increasing the number of
children growing up with two happily and
continuously married parents and of improv-
ing the well-being of children now living in
other family structures is self-evident. Chil-
dren are the innocent victims of their par-
ents inability to maintain harmonious and
stable homes. The importance of effective
policies will become even clearer in the near
future, as the baby boom generation reaches
retirement age. As this happens, our society
will become increasingly dependent on the
emotional functioning, economic productiv-
ity, and leadership of a declining number of
young adults. Although it is a clich to say
that children are the future, it has never been
as true as it is today.
Pa ul R. Ama t o
000952
Endnotes
1. For examples, see Sheldon Glueck and Eleanor Glueck, Family Environment and Delinquency (Boston:
Houghton Mifflin, 1962); J. F. McDermott, Divorce and Its Psychiatric Sequelae in Children, Archives of
General Psychiatry 23 (1970): 42127.
2. Judith S. Wallerstein and Joan B. Kelly, Surviving the Breakup: How Children and Parents Cope with Di-
vorce (New York: Basic Books, 1980).
3. E. Mavis Hetherington, Divorce: A Childs Perspective, American Psychologist 34 (1979): 85158; E.
Mavis Hetherington, Martha Cox, and R. Cox, Effects of Divorce on Parents and Children, in Nontradi-
tional Families, edited by Michael Lamb (Hillsdale, N.J.: Lawrence Erlbaum, 1982), pp. 23388.
4. The effect size for a study is defined as the standardized mean difference on some outcome between two
groups of interest, that is, (
x
1
x
2
)/S
pooled
. For information on meta-analysis, see Harris M. Cooper and
Larry V. Hedges, eds., The Handbook of Research Synthesis (New York: Russell Sage, 1994).
5. Paul R. Amato and Bruce Keith, Consequences of Parental Divorce for Childrens Well-Being: A Meta-
Analysis, Psychological Bulletin 10 (1991): 2646.
6. Paul R. Amato, Children of Divorce in the 1990s: An Update of the Amato and Keith (1991) Meta-
Analysis, Journal of Family Psychology 15 (2001): 35570.
7. Paul R. Amato and Alan Booth, A Generation at Risk: Growing Up in an Era of Family Upheaval (Harvard
University Press, 1997); Paul R. Amato and Juliana M. Sobolewski, The Effects of Divorce and Marital
Discord on Adult Childrens Psychological Well-Being, American Sociological Review 66 (2001): 90021;
William S. Aquilino, Impact of Childhood Family Disruption on Young Adults Relationships with Par-
ents, Journal of Marriage and the Family 56 (1994): 295313; Alan Booth and Paul R. Amato, Parental
Predivorce Relations and Offspring Postdivorce Well-Being, Journal of Marriage and the Family 63
(2001): 197212; Larry L. Bumpass, Theresa C. Martin, and James A. Sweet, The Impact of Family Back-
ground and Early Marital Factors on Marital Disruption, Journal of Family Issues 12 (1991): 2242; An-
drew J. Cherlin, P. Lindsay Chase-Lansdale, and Christine McRae, Effects of Divorce on Mental Health
throughout the Life Course, American Sociological Review 63 (1998): 23949; Sara McLanahan and Gary
Sandefur, Growing Up with a Single Parent: What Hurts, What Helps (Harvard University Press, 1994);
Lawrence L. Wu and B. C. Martinson, Family Structure and the Risk of a Premarital Birth, American So-
ciological Review 58 (1993): 21032.
8. McLanahan and Sandefur, Growing Up with a Single Parent (see note 7); Paul R. Amato, Parental Ab-
sence during Childhood and Adult Depression, Sociological Quarterly 32 (1991): 54356; Paul R. Amato
and Bruce Keith, Separation from a Parent during Childhood and Adult Socioeconomic Attainment, So-
cial Forces 70 (1991): 187206; William Aquilino, The Life Course of Children Born to Unmarried Moth-
ers: Childhood Living Arrangements and Young Adult Outcomes, Journal of Marriage and the Family 58
(1996): 293310; Robert Haveman, Barbara Wolf, and Karen Pence, Intergenerational Effects of Non-
marital and Early Childbearing, in Out of Wedlock: Causes and Consequences of Nonmarital Fertility, ed-
ited by Lawrence L. Wu and Barbara Wolfe (New York: Russell Sage Foundation, 2001), pp. 287316; Jay
D. Teachman, Childhood Living Arrangements and the Intergenerational Transmission of Divorce, Jour-
nal of Marriage and Family 64 (2002): 71729; Jay D. Teachman, The Childhood Living Arrangements of
Children and the Characteristics of Their Marriages, Journal of Family Issues 25 (2004): 8696.
VOL . 15 / NO. 2 / FAL L 2005 91
000953
9. McLanahan and Sandefur, Growing Up with a Single Parent (see note 7).
10. Amato, Parental Absence during Childhood and Adult Depression (see note 8); Amato and Keith Sepa-
ration from a Parent (see note 8); Teachman, Childhood Living Arrangements (see note 8).
11. Larry L. Bumpass and Hsien-Hen Lu, Trends in Cohabitation and Implications for Childrens Family
Contexts in the United States, Population Studies 54 (2000): 2941; Sara McLanahan and others, Unwed
Parents or Fragile Families? Implications for Welfare and Child Support Policy, in Out of Wedlock:
Causes and Consequences of Nonmarital Fertility, edited by Lawrence L. Wu and Barbara Wolfe (New
York: Russell Sage Foundation, 2001), pp. 20228.
12. Susan Brown, The Effect of Union Type on Psychological Well-Being: Depression among Cohabitors ver-
sus Marrieds, Journal of Health and Social Behavior 41 (2000): 24155; Susan Brown and Alan Booth,
Cohabitation versus Marriage: A Comparison of Relationship Quality, Journal of Marriage and the Fam-
ily 58 (1996): 66878; Judith Seltzer, Families Formed outside of Marriage, Journal of Marriage and the
Family 62 (2000): 124768.
13. Susan Brown, Family Structure and Child Well-Being: The Significance of Parental Cohabitation, Jour-
nal of Marriage and the Family 66 (2004): 35167. For a general review of this literature, see Wendy Man-
ning, The Implications of Cohabitation for Childrens Well-Being, in Just Living Together: Implications of
Cohabitation for Families, Children, and Social Policy, edited by Alan Booth and Ann Crouter (Mahwah,
N.J.: Lawrence Erlbaum Associates, 2002), pp. 21152.
14. Nancy S. Landale and Susan M. Hauan, The Family Life Course of Puerto Rican Children, Journal of
Marriage and the Family 54 (1992): 91224; Wendy Manning, Pamela Smock, and Debarun Majumdar,
The Relative Stability of Marital and Cohabiting Unions for Children, Population Research and Policy
Review 23 (2004): 13559.
15. M. Carlson, Sara McLanahan, and Paula England, Union Formation and Dissolution in Fragile Families,
Fragile Families Research Brief, no. 4 (Bendheim-Thoman Center for Research on Child Wellbeing,
Princeton University, January 2003); see also Sara McLanahan, Diverging Destinies: How Children Are
Faring under the Second Demographic Transition, Demography 41 (2004): 60627.
16. Lawrence L. Wu, Larry L. Bumpass, and Kelly Musick, Historical and Life Course Trajectories of Non-
marital Childbearing, in Out of Wedlock: Causes and Consequences of Nonmarital Fertility, edited by
Lawrence L. Wu and Barbara Wolfe (New York: Russell Sage Foundation, 2001), pp. 348.
17. Brown, Family Structure and Child Well-Being (see note 13).
18. The Add Health study was designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris
and funded by grant R01-HD31921 from the NICHD, with cooperative funding from seventeen other
agencies. Ronald R. Rindfuss and Barbara Entwisle provided assistance in the original study design. The
analysis was conducted for this paper.
19. Amato and Keith, Consequences of Parental Divorce (see note 5).
21. Amato and Keith, Separation from a Parent (see note 8).
Pa ul R. Ama t o
000954
22. Amato, Parental Absence (see note 8).
23. David Mechanic and Stephen Hansell, Divorce, Family Conflict, and Adolescents Well-Being, Journal
of Health and Social Behavior 30 (1989): 10516; James L. Peterson and Nichola Zill, Marital Disruption,
Parent-Child Relationships, and Behavior Problems in Children, Journal of Marriage and the Family 49
(1986): 295307.
24. Amato and Booth, A Generation at Risk (see note 7); Booth and Amato, Parental Predivorce Relations
(see note 7); Susan M. Jekielek, Parental Conflict, Marital Disruption and Childrens Emotional Well-
Being, Social Forces 76 (1998): 90535; Thomas L. Hanson, Does Parental Conflict Explain Why Di-
vorce Is Negatively Associated with Child Welfare? Social Forces 77 (1999):1283316.
25. Amato and Booth, A Generation at Risk (see note 7); Booth and Amato, Parental Predivorce Relations
(see note 7); Paul R. Amato, Good Enough Marriages: Parental Discord, Divorce, and Childrens Well-
Being, Virginia Journal of Social Policy and the Law 9 (2002): 7194.
26. Paul R. Amato, The Implications of Research on Children in Stepfamilies, in Stepfamilies: Who Benefits?
Who Does Not? edited by Alan Booth and Judy Dunn (Hillsdale, N.J.: Lawrence Erlbaum, 1994); E. Mavis
Hetherington and W. Glenn Clingempeel, Coping with Marital Transitions, Monographs of the Society for
Research in Child Development, vol. 57, nos. 23 (University of Chicago Press, 1992); E. Mavis Hethering-
ton and K. M. Jodl, Stepfamilies as Settings for Child Development, in Stepfamilies: Who Benefits? Who
Does Not? edited by Alan Booth and Judy Dunn (Hillsdale, N.J.: Lawrence Erlbaum, 1994), pp. 5579.
27. For a discussion of how stepchildren view stepparents, see E. Mavis Hetherington and John Kelly, For Bet-
ter or for Worse: Divorce Reconsidered (New York: Norton, 2002).
28. Martin Daly and Margo Wilson, Child Abuse and Other Risks of Not Living with Both Biological Parents,
Ethology and Sociobiology 6 (1985): 197220; Leslie Margolin and John L. Craft, Child Sexual Abuse by
Caretakers, 38 (1989): 45055.
29. Richard Gelles and John W. Harrop, The Risk of Abusive Violence among Children with Nongenetic
Caretakers, Family Relations 40 (1991): 7883.
30. Phyllis Bronstein and others, Fathering after Separation or Divorce: Factors Predicting Childrens Adjust-
ment, Family Relations 43 (1994): 46979; Margaret Crosbie-Burnett and Jean Giles-Sims, Adolescent
Adjustment and Stepparenting Styles, Family Relations 43 (1994): 39499; Lynn White and Joan G.
Gilbreth, When Children Have Two Fathers: Effects of Relationships with Stepfathers and Noncustodial
Fathers on Adolescent Outcomes, Journal of Marriage and Family 63 (2001): 15567.
31. Christine M. Buchanan, Eleanor E. Maccoby, and Sanford M. Dornbusch, Adolescents after Divorce (Har-
vard University Press, 1996).
32. Mignon R. Moore and P. L. Chase-Lansdale, Sexual Intercourse and Pregnancy among African-American
Girls in High-Poverty Neighborhoods: The Role of Family and Perceived Community Environment, Jour-
nal of Marriage and the Family 63 (2001): 114657; Sandi Nelson, Rebecca L. Clark, and Gregory Acs, Be-
yond the Two-Parent Family: How Teenagers Fare in Cohabiting Couple and Blended Families, series B,
no. B-31 (Washington: Urban Institute, 2001).
33. Brown, The Effect of Union Type (see note 12).
VOL . 15 / NO. 2 / FAL L 2005 93
000955
34. Hetherington, Cox, and Cox, Effects of Divorce (see note 3).
35. Amato and Keith, Consequences of Parental Divorce (see note 5).
36. Amato, Children of Divorce (see note 6).
37. E. Mavis Hetherington, K. A. Camara, and David L. Featherman, Achievement and Intellectual Func-
tioning of Children in One-Parent Households, in Achievement and Achievement Motives, edited by J. T.
Spence (San Francisco: W. H. Freeman, 1983).
38. Amato and Keith, Separation from a Parent (see note 8).
40. Amato, Parental Absence (see note 8).
42. Moore and Chase-Lansdale, Sexual Intercourse and Pregnancy (see note 32).
43. Nelson, Clark, and Acs, Beyond the Two-Parent Family (see note 32).
44. McLanahan and Sandefur, Growing Up with a Single Parent (see note 7); Robert H. Aseltine, Pathways
Linking Parental Divorce with Adolescent Depression, Journal of Health and Social Behavior 37 (1996):
13348; Donna R. Morrison and Andrew J. Cherlin, The Divorce Process and Young Childrens Well-
Being: A Prospective Analysis, Journal of Marriage and the Family 57 (1995): 80012; Ronald L. Simons
and Associates, Understanding Differences between Divorced and Intact Families (Thousand Oaks, Calif.:
Sage, 1996).
45. Valarie King, Nonresident Father Involvement and Child Well-Being: Can Dads Make a Difference?
Journal of Family Issues 15 (1994): 7896; Sara McLanahan and others, Child Support Enforcement and
Child Well-Being: Greater Security or Greater Conflict? in Child Support and Child Well-Being, edited
by Irwin Garfinkel, Sara McLanahan, and Philip K. Robins (Washington: Urban Institute Press, 1996), pp.
23956.
46. Hetherington and Clingempeel, Coping with Marital Transitions (see note 26); Simons and Associates,
Understanding Differences (see note 44); Nan Astone and Sara S. McLanahan, Family Structure, Parental
Practices, and High School Completion, American Sociological Review 56 (1991): 30920; Elizabeth
Thomson and others, Family Structure, Gender, and Parental Socialization, Journal of Marriage and the
Family 54 (1992): 36878.
47. McLanahan and Sandefur, Growing Up with a Single Parent (see note 7); Hetherington and Clingempeel,
Coping with Marital Transition (see note 26); Buchanan, Maccoby, and Dornbusch, Adolescents after Di-
vorce (see note 31); Simons and Associates, Understanding Differences (see note 44).
48. Paul R. Amato and Joan Gilbreth, Nonresident Fathers and Childrens Well-Being: A Meta-Analysis,
Journal of Marriage and the Family 61 (1999): 55773.
49. Paul R. Amato and Juliana Sobolewski, The Effects of Divorce on Fathers and Children: Nonresidential
Fathers and Stepfathers, in The Role of the Father in Child Development, edited by Michael Lamb, 4th
ed. (Hillsdale, N.J.: Erlbaum, 2003), pp. 34167.
Pa ul R. Ama t o
000956
50. W. V. Fabricius, Listening to Children of Divorce: New Findings that Diverge from Wallerstein, Lewis,
and Blakeslee, Family Relations 52 (2003): 38594.
51. Frank F. Furstenberg and Andrew Cherlin, Divided Families: What Happens to Children When Parents
Part (Harvard University Press, 1991).
52. L. I. Pearlin and others, The Stress Process, Journal of Health and Social Behavior 22 (1981): 33756;
Peggy A. Thoits, Stress, Coping, and Social Support Processes: Where Are We? What Next? Journal of
Health and Social Behavior, extra issue (1995): 5379.
53. Buchanan, Maccoby, and Dornbusch, Adolescents after Divorce (see note 31); Jeanne M. Tschann and oth-
ers, Conflict, Loss, Change and Parent-Child Relationships: Predicting Childrens Adjustment during Di-
vorce, Journal of Divorce and Remarriage 13 (1999): 122; Elizabeth A. Vandewater and Jennifer E. Lans-
ford, Influences of Family Structure and Parental Conflict on Childrens Well-Being, Family Relations 47
(1998): 32330.
54. Buchanan, Maccoby, and Dornbusch, Adolescents after Divorce (see note 31).
55. Patrick T. Davies and E. Mark Cummings, Marital Conflict and Child Adjustment: An Emotional Security
Hypothesis, Psychological Bulletin 116 (1994): 387411.
56. McLanahan and Sandefur, Growing Up with a Single Parent (see note 7); Aseltine, Pathways (see note
44); Simons and Associates, Understanding Differences (see note 44); Buchanan, Maccoby, and Dorn-
busch, Adolescents after Divorce (see note 31); Michael S. Ellwood and Arnold L. Stolberg, The Effects of
Family Composition, Family Health, Parenting Behavior and Environmental Stress on Childrens Divorce
Adjustment, Journal of Child and Family Studies 2 (1993): 2336; Irwin Sandler and others, Stability and
Quality of Life Events and Psychological Symptomatology in Children of Divorce, American Journal of
Community Psychology 19 (1991): 50120; Jay D. Teachman, Kathleen Paasch, and Karen Carver, Social
Capital and Dropping Out of School Early, Journal of Marriage and the Family 58 (1996): 77383.
57. Wu and Martinson, Family Structure (see note 7); Paul R. Amato, Reconciling Divergent Perspectives:
Judith Wallerstein, Quantitative Family Research, and Children of Divorce, Family Relations 52 (2003):
33239; Bryan Rodgers and Jan Pryor, Divorce and Separation: The Outcomes for Children (York, Eng-
land: Joseph Rowntree Foundation, 1998).
58. M. McGue and D. T. Lykken, Genetic Influence on Risk of Divorce, Psychological Science 3 (1992):
36873; V. Jockin, M. McGue, and D. T. Lykken, Personality and Divorce: A Genetic Analysis, Journal of
Personality and Social Psychology 71 (1996): 28899. For a strong statement of this position, see Judith
Harris, The Nurture Assumption: Why Children Turn Out the Way They Do (New York: Touchstone,
1999).
59. Simons and Associates, Understanding Differences (see note 44); Amato, Reconciling Divergent Perspec-
tives (see note 57).
60. See note 7. Cherlin and others used a fixed-effects model, which eliminates all unmeasured variables that
do not change over time. McLanahan and Sandefur relied on biprobit analysis, a method that makes it pos-
sible to correlate error terms across equations, which is equivalent to adjusting for unmeasured variables
that could affect family structure as well as childrens outcomes.
VOL . 15 / NO. 2 / FAL L 2005 95
000957
61. David Brodzinsky, Jennifer C. Hitt, and Daniel Smith, Impact of Parental Separation and Divorce on
Adopted and Nonadopted Children, American Journal of Orthopsychiatry 63 (1993): 45161; Thomas G.
OConnor and others, Are Associations between Parental Divorce and Childrens Adjustment Genetically
Mediated? An Adoption Study, Developmental Psychology 36 (2000): 42937.
62. K. S. Kendler and others, Childhood Parental Loss and Adult Psychopathology in Women, Archives of
General Psychiatry 49 (1992): 10916.
63. In this analysis, I considered adoptive parents to be the same as biological parents. The one parent cate-
gory includes adolescents living with one biological parent and a stepparent (or a cohabiting partner of the
parent). This category also includes a small percentage of children living with neither parent at the time of
the interview. I used logistic regression analysis to adjust the percentages in table 1 for variables that could
be associated with family structure as well as child outcomes: mothers education, fathers education, childs
race (white, black, Latino, or other), childs age, childs gender, and whether the child was born in the
United States. All of the differences reported in table 1 were statistically significant at p < .01.
64. The margin of error for these estimates, based on a 95 percent confidence interval, is about 1 percent.
65. To estimate the percentage of adolescents between the ages of thirteen and eighteen living with two bio-
logical parents in 1980, 1970, and 1960, I relied on retrospective data from the 1988 wave of the National
Survey of Families and Households. The resulting figures are 64 percent, 74 percent, and 77 percent, re-
spectively. The margin of error for these estimates, based on a 95 percent confidence interval, is about 2
percent. For details on the National Survey of Families and Households, see James Sweet, Larry Bumpass,
and Vaughn Call, The Design and Content of the National Survey of Families and Households, NSFH
Working Paper 1 (Center for Demography and Ecology, University of Wisconsin-Madison, 1988). These
estimates should not be equated with levels of program effectiveness, because it is naive to assume that
specific, short-term interventions could reverse family trends so strongly. It is possible, however, that a
range of interventions, combined with a shift in the larger culture, could result in substantial changes in
family structure over a decade or longer. Moreover, the figures used in table 1 are not completely unrealis-
tic, because they correspond to levels of family stability that actually existed in recent U.S. history. Note
also that these estimates are based only on changes in family structure and assume no changes in marital
quality in two-parent families. If future policies also are capable of improving the quality of existing mar-
riages, then the figures in tables 1 and 2 are underestimates of the total benefit to children.
66. The National Heart, Lung, and Blood Institute, High Blood Cholesterol: What You Need to Know, NIH
Publication 01-3290 (May 2001). See also F. B. Hu, J. E. Manson, and W. C. Willett, Types of Dietary Fat
and Risk of Coronary Heart Disease: A Critical Review, Journal of the American College of Nutrition 20
(2001): 519; S. Lewington and S. MacMahon, Blood Pressure, Cholesterol, and Common Causes of
Death: A Review, American Journal of Hypertension 12 (1999): 96S98S.
67. The estimates for ten-year risk of a heart attack vary with age and gender. The link between cholesterol and
cardiovascular disease is stronger for men than for women, and stronger for older individuals than for
younger individuals. The margin of error for these estimates, based on a 95 percent confidence interval, is
about 1 percent.
68. U.S. Bureau of the Census, Statistical Abstract of the United States (Government Printing Office, 2003),
table 11.
Pa ul R. Ama t o
000958

TAB 56

Rev Econ Household
DO! 10. 1007/sl I 150-013-9220-y
High school graduation rates among children of same-
sex households
Douglas W. Allen
Received: 7 November 2012/ Accepted: 23 August 2013
Springer Science+Business Media New York 2013
Abstract Almost all studies of same-sex parenting have concluded there is "no
difference" in a range of outcome measures for children who live in a household
with same-sex parents compared to children living with married opposite-sex par-
ents. Recently, some work based on the US census has suggested otherwise, but
those studies have considerable drawbacks. Here, a 20 % sample of the 2006
Canada census is used to identify self-reported children living with same-sex par-
ents, and to examine the association of household type with children's high school
graduation rates. This large random sample allows for control of parental marital
status, distinguishes between gay and lesbian families, and is large enough to
evaluate differences in gender between parents and children. Children living with
gay and lesbian families in 2006 were about 65 % as likely to graduate compared to
children living in opposite sex marriage families. Daughters of same-sex parents do
considerably worse than sons.
Keywords Same sex households Same sex parents High school
graduation
JEL Classification 121 Jl2 Jl6
Children raised by gay or lesbian parents are as likely as children raised by
heterosexual parents to be healthy, successful and well-adjusted. The research
supporting this conclusion is accepted beyond serious debate in the field of
developmental psychology.
[Justice Vaughn Walker, section 70, Perry v. Schwarzenegger]
D. W. Allen (1'81)
Department of Economics, Simon Fraser University, Burnaby, BC V5A 186, Canada
e-mail: allen @sfu.ca
Published online: 26 September 2013
<1 Springer
000959
D. W. Allen
1 Introduction
The matter of same-sex marriage is perhaps the most significant policy issue in
family law since the introduction of no-fault divorce in the late 1960s and 1970s.
Much of the debate is focused on the question of equality, although issues of
relationship stability, consequences for opposite sex marriages, aud marriage culture
are often brought up. One aspect that is seldom argued is the effect a same-sex union
might have on the children within that union. The absence of any discussion on
children no doubt reflects the unanimous consensus in the child development
literature on this question-it makes no difference.
Within the last 15 years there have been over fifty empirical studies considering
the effects on children of growing up within a same-sex household-' Despite the
various differences in each study, all but a couple have the same conclusion:
children of same-sex parents perform at least as well as children from heterosexual
families. This conclusion, that there is no difference in child outcomes based on
family structure, has played a major role in legal cases, legislation, popular culture,
and professional opinions on gay family rights-including rights to adoption aud
maniage
2
As Justice Walker claimed, to suggest otherwise is to risk not being taken
seriously.
Unfortunately, the literature. on child development in same-sex households is
lacking on several grounds? First, the research is characterized by levels of
advocacy, policy endorsement, and awareness of political consequences, that is
disproportionate with the strength and substance of the preliminary empirical
findings. Second, the literature generally utilizes measures of child and family
perfonnance that are not easily verifiable by third party replication, which vary from
one study to another in ways that make comparisons difficult, and which differ
substantially from measures standardly used in other family studies.
4
But most
important, almost all of the literature on same-sex parenting (which almost always
1
Table I lists the studies. See Allen (2012) or Marks (2012) for surveys of this literature. Throughout the
paper the term "same-sex household" is used to mean gay or lesbian headed households.
2
For example, it forms the basis for the American Psychological Association's position supporting gay
marriage.
3
Economists have written a considerable amount on gay and lesbian issues outside of child development,
and generally find differences in behavior. For example, Negrusa and Oreffice (2011) on savings rates,
Oreffice (2010) on labor supply, Black et al. (2007) on labor markets, Jepsen and Jepsen (2009) on home
ownership, and Carpenter and Gates (2008) on family fonnation. Indeed, The Review of Economics of the
Household devoted its 2008 December issue to gay and lesbian households. Those papers examined wage
differentials [Zavodny (2008), Booth and Frank (2008)], household formation [Badgett et al. (2008)], and
bank deposits [Klawitter (2008)]. This is the first paper in economics to examine differences in child
perfonnance.
4
This is often a characrerisric of a nascent field. These measures include self reports on attitudes,
awareness, and adjustments [e.g., McNeill and Rienzi (1998)]; self reports on parenting quality and socio-
emotional child development [e.g., Golombok et al. (1997)]; self reports on psychological well-being,
identity, and relationships [e.g., Tasker et al. (1995)]; self reports on family closeness, parental
legitimacy, child bonding [e.g., Gartrell et al. (1999)]; and self reports on stigma and self-esteem [e.g.,
Gershon et al. (1999)].
<21 Springer
000960
High school graduation rates
Table 1 Summaries of gay parenting studies
Study
Bailey et a!. ( 1995)
Flaks et a!. (1995)
Patterson ( 1995)
Tasker et a!. (1995)
Golombok et a!. (1996)
Sarantakos (1996)
Brewaeys et a!. (1997)
Golombok eta!. (1997)
Chan et a!. (1998a)
Chan eta!. (1998b)
McNeill and Rienzi (1998)
Patterson et a!. (1998)
Gershon et a!. (1999)
Gartrell et a!. (1999)
Dundas et a!. (2000)
Barrett eta!. (2001)
Chrisp (200 I)
Patterson (200 I)
Fulcher et a!. 2002)
Vanfraussen et al. (2002)
Golombok et a!. (2003)
Bos et a!. (2004)
Patterson et a!. (2004)
Stacey (2004)
MacCallum and
Golombok (2004)
Wainright eta!. (2004)
Leung et a!. (2005)
Scheib et a!. (2005)
Stacey (2005)
Wainright et a!. (2006)
Wright et a!. (2006)
Bos et a!. (2007)
Goldberg (2007)
Balsam et a!. (2008)
Bos et a!. (2008)
Bos et a!. (2008)
Fairlough (2008)
Random
sample
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
Yes
No
No
No
No
Yes
No
No
No
No
No
No
No
Gay Contenta Comparison
sample group size
size
55 Hard None
30 Soft 30
26 Soft None
25 Soft 21
25 Hard 21
58 Soft 116
30 Soft 68
30 Soft 83
30 Soft 16
55 Soft 25
24 Soft 35
37 Soft None
76 Soft None
84 Soft None
27 Soft None
84 Soft None
10 I Soft None
8 Soft None
37 Soft None
55 Soft 25
24 Soft 24
39 Soft 134
100 Soft None
33 Hard 33
50 Soft None
25 Soft 76
44 Hard 44
84 Soft None
47 Soft Ill
12 Soft 17
50 Soft None
44 Hard 44
156 Soft None
99 Soft 100
46 Soft None
281 Soft 55
63 Soft None
152 Soft None
67 Soft None
Time
series
data
No
No
No
No
No
Yes
No
Yes
No
No
No
No
No
Yes
No
Yes
No
No
No
No
No
No
No
No
No
No
No
Yes
No
No
No
No
No
No
No
No
No
No
No
Gay or
lesbian
study
G
L
L
L
L
G&L
L
L
L
L
L
L
L
L
L
L
G
L
L
L
L
L
L
L
G
L
L
L
G&L
L
G
L
G
L
G&L
G&L
L
L
G&L
<l Springer
000961
D. W. Allen
Table 1 continued
Study Random Gay Contenta Comparison Time Gay or
sample sample group size series lesbian
size data study
Fulcher et al. (2008) No 33 Soft 33 No L
Goldberg et al. (2008) No 30 Soft None No L
Oswald et al. (2008) No 190 Hard None No G&L
Rothblum et al. (2008) Pop. 475 Hard None No G,L& T
Rivers et al. (2008) Yes 18 Soft 18 No L
Sutfin et al. (2008) No 29 Soft 28 No L
Wainright and Patterson (2008) Yes 44 Soft 44 No L
Bas (2010) No 36 Soft 36 No G
Gartrell and Bas (2010) No 84 Hard 93 Yes L
Lehmiller (2010) No 68 Soft 86 No G
Power et al. (20 10) No 455 Hard None No G&L
Rosenfeld (2010) Yes 3,502 Hard >700,000 No G&L
Regnerus (2012) Yes 248 Hard 2,988 No G&L
Allen et al. (2013) Yes 8,632 Hard I ,189,833 No G&L
G gay, L lesbian, and T transgendered
"
Hard implies the questions asked were potentially verifiable, quantifiable, and had observable answers.
Soft implies the opposite. Some studies included both and were classified as hard
means lesbian parenting) is based on some combination of weak empirical designs,
small biased convenience samples, "snowballing," and low powered tests.
5
This paper addresses these shortcomings by nsing the 2006 Canada census to
study high school graduation probabilities of children who lived with both gay and
lesbian parents in 2006, and to compare them with four other family types: married,
common law, single mothers, and single fathers. Currently, the 2006 Canada census
has several strengths compared to any other data set. First, it uses information from
a country where same-sex couples have enjoyed all taxation and government
benefits since 1997, and legal same-sex marriage since 2005
6
As Biblarz and Savci
note, such legalization reduces the stress and stigma of homosexuality, and
encourages honest participation in census questions.
7
Second, not only does the
census provide a large random sample, but married and common law same-sex
5
"Snowballing" is the practice of asking individuals within a study to recruit their friends and associates
to join the study.
6
The first Canadian same-sex maniages took place on January 14, 2001 at the Toronto Metropolitan
Community Church. These became the basis of a successful legal challenge which ended at the court of
appeal on June 10, 2003. On July 20, 2005, the Federal government passed the Civil Marriage Act that
made Canada the fourth country in the world to legalize same-sex marriage. Thus, different people date
the arrival of same-sex marriage in Canada as 2001, 2003, or 2005.
7
Biblarz and Savci, p. 490, 2010 .
.:1 Springer
000962
couples and their children are self identified-" This is an important advantage over
the US census. Third, because the child and parent records are linked together, the
marital status and educational levels of the parents can be controlled for when
analyzing child performance. Finally, because of the relatively large sample size,
there is enough power to not only separate gay from lesbian households, but also
enough to examine the gender mix of same-sex households.
9
The point estimates for high school graduation show that there is a significant
reduction in the odds of children living in same-sex homes completing high school.
In the case of gay parents, children are estimated to be 69 % as likely to graduate
compared to children from opposite sex married homes.
1
For lesbian households
the children are 60 % as likely to graduate from high school. A breakdown of
performance by the sex of the child shows a more dramatic result. Daughters of gay
parents are only 15 % as likely to graduate, while daughters of lesbian parents are
45 % as likely to graduate. Both sets of results are estimated with precision. On the
other hand, sons of lesbian parents are 76 % as likely to graduate, while sons of gay
parents are 61 % more likely to graduate. However, neither of these results are
statistically significant. In general, the results for gays and lesbians respond
differently to different controls, and differ from the results for the other family
types. This, and the different graduation rates for sons and daughters, suggest that
the two types of same-sex couples are much different and should not be categorized
together in empirical work.
These results survive several robustness checks. Graduation rates may be
different because school attendance rates are different, yet no statistical difference
in the probability of attending school across the different family types is found. In
fact, the point estimates indicate children of opposite-sex married parents are less
likely to attend school. Various changes in sample restrictions and controls also
leave the results unchanged.
2 Context within the child development literature
Since most economists are unfamiliar with the literature on child performance in
same-sex households, a brief review of its empirical problems is warranted.
Generally speaking the literature is characterized by several different types of data
bias and small samples that lack any power. Table I reports some information on
the relevant fifty three studies conducted the past 15 years. With the exception of
8
Unfortunately, it also lumps married and common law same-sex couples together, and I am unable to
separate them.
9
The census is not a panel, and provides only a snap shot of the population. As a result, this paper does
not study the effect of growing up in a same-sex household, but rather examines the association of school
performance for those children who lived with same-sex parents in 2006.
10
Rosenfeld (2610) stressed the importance of contro11ing for a child's home life stability. He restricted
the sample to households that remained in one place for the past 5 years. Here mobility is controlled for
with a fixed effect on whether or not the child has remained in the home for I year. Results reported in the
text all refer to this mobility control. The "Appendix" shows the results of the alternative control: did the
child move residences in the past 5 years.
<! Springer
000963
D. W. Allen
two of the last three, the others have serious empirical problems that render them
exploratory in nature.
2.1 Random samples
Although a proper probability sample is a necessary condition for making any claim
about an unknown population, within the same-sex parenting literature researchers
have studied only those community members who are convenient to study. This
point has been raised by others regarding the literature on gay parenting, including
many within the literature.'.! Of the fifty-three studies reviewed here, only seven
used probability samples.
12
All of the other studies arrived at their samples through
means that introduced various levels of bias. Some studies recruited individuals
from sperm bank data sources or other types of reproduction technology providers.
13
Other studies used Internet surveys where the respondents were recruited by various
methods: parent forums, gay and lesbian web-sites, and online advocacy organi-
zations.
14
Many studies recruited through LGBT events, bookstore and newspaper
advertisements, word of mouth, networking, and youth groups.
15
A common
method of recruitment was to use a combination of the above methods to form a
sample base, and then recruit friends of the base.
16
Still other studies failed to even
mention how their samples were arrived at.
17
Each different procedure has a
different and unknown source of bias.
Of the studies before 2010, there are only four that use a random sample, and
each has a trivial sample size. For example, consider the three studies by Wainright
and Patterson.
18
These are not three independent studies, but rather three separate
publications utilizing the same data source: the National Longitudinal Survey of
Adolescent Health. Even though the survey contains 12, I 05 households, Wainright
and Patterson were only able to identify 6 gay households and 44 lesbian ones. The
11
Andersson et al. (2006) note:
The lack of representative samples is the most fundamental problem in quantitative studies on gays
and lesbians, which commonly rely on self-recruited samples from an unknown population. [p. 81]
See also Sweet (2009) or Stacey and Biblarz (2001).
12
These were Allen eta!. (2013), Regnerus (2012), Rosenfeld (2010), Wainright et a!. (2004), Wainright
and Patterson (2006, 2008), and Golombok et a!. (2003). One study used a population: Rothblum et a!.
(2008).
13
For example: Bos eta!. (2007), Bos and Van Balen (2008), Chan eta!. (1998a), Brewaeys eta!. (1997)
and Chan eta!. (1998b).
14
For example: Lehrniller (2010), Bos (2010), or Power eta!. (2010).
15
For example: Wright and Perry (2006), Oswald et a!. (2008), Lehmiller (2010), Goldberg (2007),
Bailey eta!. (1995), Flaks et a!. (1995), Fairtlough (2008), Dundas and Kaufman (2000), Power eta!. or
Fulcher et a!. (2008).
16
For example: Balsam eta!. (2008), Golombok eta!. (2003).
17
For example: Stacey (2004, 2005) or Chrisp (2001).
18
Wainright and Patterson (2006, 2008) and Wainright eta!. (2006).
42\ Springer
000964
other study by Rivers et a!. (2008) used a similar British survey, and ended up with
a sample of 18 lesbian households.
19
The only study with a large random sample in the entire literature is Rosenfeld
(2010), that used the 2000 US Census 5 %Public-use Micro-sample to examine the
association between same-sex parenting and normal progress through schooi.
20
His
study confirmed the findings of most earlier research, and he concluded that in terms
of school grade progression children raised by same-sex couples "cannot be
distinguished with statistical certainty from children of heterosexual married
couples." Rosenfeld's study was the first to use a large random sample to support
the finding that children of same-sex households were no different in a performance
measure from children of married opposite sex couples.
However, a follow up study by Allen eta!. (2013), found that Rosenfeld's
conclusion was questionable. His estimates were so imprecise that the outcomes of
children in same-sex households could not be distinguished with any statistical
certainty from almost any other family type-not just opposite-sex married families.
The imprecision came from Rosenfeld's decision to exclude from the sample any
family who had moved within the past 5 years. Same-sex households turned out to
be strongly correlated with mobility, and the result was a large reduction in the
same-sex household sample, which led to an inability to statistically distinguish the
children from these households with any others-including ones known to be poor
environments for children. By controlling for mobility and restoring the sample to
its full size, Allen et a!. (2013) found that children from same-sex homes were about
35 % more likely to fail a grade compared to children from intact opposite sex
married homes. About on par with children from single parent homes.
21
2.2 Small sample sizes
Aside from the problem of non-random samples, most of the ex1stmg parenting
studies contain tiny sample sizes.
22
Of the fifty-three studies examined here, only
19
Golombok et aL (2003) uses a random sample from the Avon Longitudinal Study of Parents and
Children-a local British study-and comes up with 18 lesbians. They then use snowball methods to
bring their numbers up to 39 lesbians.
20
The 2000 US Census does not directly identify same-sex couples. Rosenfeld, like others, did the best
he could by indirectly identifying them. He did this by selecting those couples who indicated they were a
couple and who identified their sex as being the same. This procedure requires the correct answer of three
questions, and a small chance of error on the part of heterosexuals can lead to a large measurement error
for the same-sex couple sample, given the large size of the fonner and the small size of the latter. Black
et al. (2006) suggest a procedure for correcting this statistical problem; however, there is no indication in
the Rosenfeld paper that he followed it.
21
The Regnerus study (2012) also used a random sample; however, it was still too small to identify a
sufficient sample of same-sex parents. To increase his sample size he decided to use a broader definition
of same-sex parent.
22
Of the fifty-three studies examined here, only a few dealt with gay male parents. Almost all of the
studies are done on lesbians. This is another source of bias that warrants caution in drawing any
conclusions about non-lesbian families.
<l Springer
000965
D. W. Allen
two had sample sizes larger than 500
23
Much more common were sample sizes
between 30-60.'
4
The problem with such small sample sizes is that the data cannot
generate any power for statistical testing, and low power means there is a small
chance of rejecting a false null hypothesis.
Hence, the very small sample sizes found in many of these studies creates a bias
towards accepting a null hypothesis of "no effect" in child outcomes between same-
and opposite-sex households. This is well recognized, but it is exacerbated in the
context of gay parenting because avenues through which these households are
formed are many and complicated. As noted by Stacey and Biblarz (2001 ), Biblarz
and Stacey (2010), these families often have experienced a prior divorce, previous
heterosexual marriages, intentional pregnancies, co-parenting, donor insemination,
adoption, and surragacy. These channels may have different effects on boys or girls,
and may differ in gay or lesbian homes. Empirical work needs to control for the
various selection effects that arise from the number of parents, sexual identity,
marital status, gender, and biological relationships with children. That is, child
performance is affected by all these channels and they need to be statistically
identified, but this requires large sample sizes.
A review of the same-sex parenting literature inevitably leads to the conclusion
that it is a collection of exploratory studies. Even the two most recent studies by
Rosenfeld (2010) and Allen eta!. (2013) suffer from several drawbacks. First, they
have to rely on indirect identification of same-sex couples within an environment
where same-sex marriage was illegal in all fifty states. Second, neither paper
distinguishes between gay and lesbian households, and there is no reason to think
their parental performance should be the same. Third, both papers fail to control for
the marital history of the parents. The increased chance of failing a grade-
especially when the correlation magnitude is so close to that of single parents-
could likely be the result of a previous divorce or separation since many children in
same-sex households were initially born into opposite sex families that later broke
apart. The "same-sex" aspect of these parents may have nothing to do with slower
grade progress.
And so, within the context of this uniform literature based on small biased
samples, this study intends to examine high school graduation rates of children who
lived within either gay or lesbian households in 2006, using a large random data set
that links parent and child records.
23
These were Rosenfeld (2010) and Allen et al. (2013). According to Nock (p. 37, 2001), to properly test
any hypothesis regarding gay parenting, a sample size of 800 is required.
24
Often the problem of small sample size comes from low response rates. Many of the fifty-two studies
are silent on the question of response rates to their surveys, but when information is provided it often
shows that response rates are very low. For example, in Bos (201 0) the gay males were recruited from an
Internet mail list for gay parents. Although the list had 1,000 names, only 36 replied and participated in
the study. This amounts to a 3.6 %response rate. Other studies (e.g., Chan et al. and Fulcher et al.) have
reductions in their samples similar in relative size to Rosenfeld. Response rates lower than 60 % are
usually taken to mean the presence of a strong selection bias--even when the initial list is random.
ttl Springer
000966
3 Data
Data come from the 2006 Canada census 20 % restricted master file
25
From this file
all children living with a parent within the home were selected?
6
It is important to
note that the census identifies children living with their parents, and not just adults.
Hence, children of same sex parents are those who respond affirmative to the
question: "Are you a child of a male (female) same-sex manied or common law
couple?" This implies that the results below address the association of having two
same-sex parents with a given sexual orientation, rather than just the association of
having two parents of the same-sex. That is, the two parents are not same-sex
roommates, friends, or other relatives
27
Restricting the sample to children living with parents allowed a matching of the
child files with the parent files. Children over the age of 22 were dropped because of
a likely selection bias in children who live at home well into adulthood?
8
Although
the Census identifies children living with two same-sex parents, it does not identify
children living with a gay or lesbian single parent. These families are inadvertently
included with the single mothers and fathers.
29
Table 2 defines the variables used in the analysis, and Tables 3 and 4 report some
unconditional means for children between the ages 17-22 across the six family
types. Table 3 reports graduation rates for the different family types, not just for the
full sample, but also on three sub samples. In tenns of the full sample three things
stand out: children of manied opposite-sex families have a high graduation rate
compared to the others; children of lesbian families have a very low graduation rate
compared to the others; and the other four types are similar to each other and lie in
between the married/lesbian extremes. The three sub samples (both parents are high
school graduates; the family never changed dwellings in the previous 5 years; and
the family did change dwellings in the previous 5 years) show that even though the
25
This file is not a public use data set. To use the data, a proposal is screened by the Social Sciences
Research Council of Canada, an RCMP criminal check is conducted, and the researcher becomes a
deemed employee of Statistics Canada subject to the penalties of the Statistics Act. Empirical work was
conducted at the SFU Research Data Center, and all results were screened by Statistics Canada before
release. Statistics Canada does not allow any unweighted observations or descriptives to be released, nor
any maximums or minimums of weighted estimates, nor sample sizes for the weighted regressions.
26
Because the procedure starts by selecting the children, and then matches the parents of the child to the
file, the problem of having a non-biological parent not report a child in the household who is biologically
related to their spouse is avoided.
27
Statistics Canada does not allow the sample sizes to be released; however, there are approximately ten
million children in Canada, and so the sample has close to two million children in it.
28
There's no reason to believe this selection bias would be correlated with family type, however. All
regressions were run with various restrictions on the child's age within the sample, including keeping
everyone, and none of the gay or lesbian family results in the paper change, in terms of magnitudes or
levels of significance, in an important way.
29
Many children in Canada who live with a gay or lesbian parent are actually living with a single parent.
About 64 % of children in gay homes have a single father, and about 46 % of children in lesbian homes
have a single mother (see Allen and Lu, "Marriage and children: differences across sexual orientations,"
(unpublished, 2013). The number of gay and lesbian single parent homes is so small compared to all other
single parent homes, however, that it likely causes little bias. In any event, the children analyzed here are
a distinct subset of all children raised by a gay or lesbian parent.
.:l Springer
000967
Table 2 Definitions of variables
Variable name
Human capital variables
Highschool grad
Province
Visible minority
Disabled
Moved within I year
Moved within 5 years
Urban
Age
Family size
Female
Family income
Same race
Father HSG
Mother HSG
Father grad degree
Mother grad degree
Father attended
Mother attended
Parent's current marital status
Father married
Father divorced
Father separated
Father never married
Father widowed
Mother married
Mother divorced
Mother separated
Mother never manied
Mother widowed
Family type
Common law
Gay parents
Lesbian parents
Single mother
Single father
D. W. Allen
Definition
I if child has graduated from highschool
10 if located in Newfoundland, 12 if PEI...62 if Nunavut
1 if a member of a visible minority
I if physica1ly or mentally disabled
1 if family changed dwelling within past year
I if family changed dwelling within the past 5 years
1 if child lives in urban area
Age of child in years
Number of family members
1 if child is a girl
Before tax income in dollars
1 if child is the same race as resident parent(s)
1 if father graduated from highschool
I if mother graduated from highschool
1 if father has a graduate degree
I if mother has a graduate degree
I if father ever attended school
I if mother ever attened school
1 if father is legally married
I if father is divorced
I if father is separated
1 if father has never legally married and single
I if father is widowed
1 if mother is legally married
1 if mother is divorced
1 if mother is separated
1 if mother has never legally married and single
1 if mother is widowed
1 if couple is living common law
1 if couple is two gay men
1 if couple is two lesbian women
1 if only parent in the home is the mother
1 if only parent in the home is the father
level of graduation rates may change, the relationship between the groups remains
the same.
Some of the results from Table 4 are fascinating. In terms of sample sizes, it is
striking how few same-sex couples with children (between ages 17-22) there are.
The country estimates for gay families is just 423, and for lesbian families 969;
<Zl Springer
000968
which together make up just over half of 1 % of all couples with children in this age
group. There are a higher number of visible minority children for gay households
(28 % compared to 13 % for common law couples), and a higher number of
disabled children (13 % compared to 6 % for opposite sex married parents). This
may imply a high number of adopted children in gay households, but interestingly
there are no cases of inter-racial same-sex families within the 20 % sample.
30
Both
lesbian and gay parents are well educated with well over 19 % of them graduating
from high school. Finally, lesbians are much more likely to have moved dwellings,
with 60 % having moved within the past 5 years.
The next section estimates the association of family type on high school
graduation rates, controlling for individual and family characteristics. One
contribution of this paper is to control for parental marital status
31
However, the
census, of course, is not a panel or even a retrospective data set. All it records is the
current marital status of the parents. Unfortunately, this introduces measurement
error into the marital status control for married individuals because the census only
identifies if a spouse is currently married, common law, never married, divorced,
separated, or widowed. Hence a married spouse may have previously been divorced,
but is recorded as married; that is, the married category contains couples who have
been divorced. This is not a problem for those currently cohabitating, since they are
accurately coded as divorced, separated, never married (single), or widowed. Since
the marriage rate is lower for gays and lesbians, this measurement error is likely to
bias the opposite-sex family type effect on child school performance downwards?
2
4 Estimation
4.1 School graduation
Table 5 reports on three logit regressions, where the dependent variable equals 1 if
the child has graduated from high school.
33
All of the regressions in this table
control for whether or not the family moved with the past year. Table 8 in the
appendix reports on another three logit regressions with the same dependent
variable and the same right hand side variables, except for the variable used to
control for family mobility-it uses the mobility measure "did child move within
30
The census identifies many visible minorities, but only has a broad based question on race. Hence, the
same race variable likely contains significant measurement error.
31
This control is lacking in other large sample studies on same-sex parents. It is important because a
previous marriage disruption is likely to have a negative impact on high school performance. This is
particularly important with same-sex couples given the evidence that their relationships are less stable
[see Andersson et al. (2006)].
32
Using current parental marital status is a decent control for family history (as used here), but the
coefficients estimated are biased measures of the correlation of parental marital history on child school
performance. For this reason, and to keep the tables to one page, these coefficients are not reported.
33
Rosenfeld (2010), and Allen et al. (2013) use normal progress through school as their measure of child
performance. The Canada census does not identify the grade of the student in 2006, and therefore, this
measure is not possible. It does, however, identify if the child has graduated from high school or not.
411 Springer
000969
D. W. Allen
Table 3 Estimated population averages for child high school graduation (weighted observations, chil-
dren ages 17-22)
Opposite Opposite sex Gay Lesbian Single Single
sex married common law parents parents father mother
parents parents
Full sample
Highschool grad 0.72 0.59 0.60 0.52 0.62 0.61
Both parent high school grad
Highschool grad 0.75 0.68 0.64 0.55 0.67 0.65
Never moved
Highschool grad 0.73 0.62 0.59 0.58 0.65 0.64
Did move
Highschool grad 0.69 0.55 0.60 0.48 0.58 0.57
past 5 years. "
34
There is no qualitative difference in the estimates when using the
different mobility controls.
Table 5 only reports the Jogit coefficient, its standard error, the odds ratio, and
marginal effects, for the household type variables in order to keep the tables a
reasonable size. The Jog of tbe odds ratio is the Jogit, which is a linear combination
of the parameters and exogenous variables. The odds ratio is found by taking the
exponential of both sides of the logit equation. The odds ratio has the nice property
of an easy interpretation.
35
The marginal effect equals (ily/ilx), where y is the
graduation rate and x is one of the right hand side variables.
The different columns result from different types of controls. Column (I)
includes controls for child characteristics, and these include: province, visible
minority, disabled, mobility, urban, age, family size, family income, female, and
same race. Column (2) adds the parental education controls: did the mother/father
graduate from high school, and did tbe mother/father have a graduate degree.
Column (3) adds tbe parental marital status variables found in Table 2.
Before examining the results of Tables 5, some comment on the unreported
results is warranted. Among the child characteristics, being disabled or having
moved in the recent past reduces tbe odds of graduation (on average, to about 50 and
75 % respectively). Living in an urban area, being female, and having all family
members the same race raises the odds of graduation (on average, by 30, 60, and
35 % respectively). Parental education matters a great deal: if tbe parents have
34
Two mobility measures are used because of the important role mobility played in Rosenfeld (2010).
He decided to restrict his sample by removing households that moved within the past 5 years. This
procedure was also performed here. No qualitative difference was made in terms of the point estimates.
Rather than controlling for whether or not the child had moved residences over the past one or 5 years,
the regressions were also run controlling for whether or not the child changed census metropolitan areas
over the past 1 or 5 years. No qualitative difference in the point estimates on type of household resulted,
although they were estimated with more precision. AJJ regressions cluster by province to provide robust
standard errors.
35
The key to interpreting the odds ratio is to compare it to the odds of 1 (equally likely). Hence, an odds
ratio of 1.2 means that a unit change in an independent variable, others held constant, increases the chance
of a positive outcome by 20 %.

000970
Table 4 Estimated population averages of other variables for children (weighted observations, children
ages 17-22)
Married Common Gay Lesbian Single Single
parents law parents parents parents father mother
Child characteristics
% of total pop. 71.16 6.60 0.02 0.04 5.08 17.07
School attendance 0.76 0.67 0.73 0.68 0.64 0.69
Province 36.58 32.2 29.32 36.9 35.5 35.9
Visible minority 0.23 0.13 0.28 0.17 0.19 0.28
Disabled 0.06 0.06 0.13 0.08 0.08 0.09
Moved within I year 0.07 0.13 0.08 0.19 0.15 0.15
Moved within 5 years 0.24 0.38 0.39 0.60 0.42 0.44
Urban 0.78 0.74 0.72 0.91 0.79 0.88
Age 19.26 18.91 18.96 18.79 19.20 19.15
Family size 4.30 4.05 3.34 3.77 2.72 2.98
Female 0.48 0.45 0.43 0.54 0.41 0.47
Family income 119,172 95,656 91,357 88,600 68,473 49,874
Same race 0.99 0.98
Father HSG 0.81 0.70 0.94 0.93 0.73 na
Mother HSG 0.84 0.77 0.96 0.93 na 0.79
Father grad degree 0.08 0.03 0.05 0.08 0.05 na
Mother grad degree 0.04 0.02 0.00 0.04 na 0.03
Parent's current legal marital status
Father married 1.00 na 0.45 0.20 0.03 na
Father divorced 0.00 0.38 0.40 0.37 0.44 na
Father separated 0.00 0.08 0.01 0.11 0.27 na
Father never married 0.00 0.51 0.13 0.32 0.14 na
Father widowed 0.00 0.03 0.01 0.00 0.12 na
Mother married 1.00 na 0.45 0.20 na 0.03
Mother divorced 0.00 0.40 0.24 0.22 na 0.43
Mother separated 0.00 0.09 O.Dl 0.05 na 0.24
Mother never married 0.00 0.47 0.29 0.52 na 0.17
Mother widowed 0.00 0.04 0.01 0.01 na 0.13
Estimated N 1,400,074 129,991 423 969 99,978 336,036
For gay and lesbian households the "father" is the survey respondent who self-identified as the household
head
graduated from high school, the child is almost twice as likely to do so. Finally,
marital history has the expected effects. Any marital disruption reduces the odds of
a child graduating from high school.
36
For any given household type variable the
odds ratio and level of statistical significance is generally robust to the different
36
The odds are reduced to around 70-80 %, but keep in mind this variable contains measurement error.
<.1 Springer
000971
D. W. Allen
Table 5 Odds ratios of high school graduation (weighted observations, children ages 17-22, dependent
variable: I if child graduated from high school controlling for moved within past year)
(I) (2) (3)
Gay parellts
Coefficient -0.446 -0.618 -0.374
Std. error (0.255) (0.234)* (0.191)*
Odds ratio 0.64 0.54 0.69
Marginal effect -0.08 -0.13 -0.06
Lesbian parents
Coefficient -0.816 -0.925 -0.511
Std. error (0.221)* (0.240)* (0.336)
Odds ratio 0.44 0.41 0.60
Common law
Coefficient -0.466 -0.338 0.124
Std. error (0.079)* (0.066)* (.111)
Odds ratio 0.63 0.71 1.13
Marginal effect -0.09 -0.06 0.03
Single mother
Coefficient -0.661 -0.672 -0.471
Std. error (0.041)* (0.038)* (0.123)*
Odds ratio 0.51 0.51 0.62
Single father
Coefficient -0.633 -0.685 -0.454
Std. error (0.039)* (0.039)* (0.161)*
Odds ratio 0.53 0.50 0.63
Child controls Yes Yes Yes
Parent education controls No Yes Yes
Marital status controls No No Yes
Pseudo R
2
.21 .23 .23
The variables of interest are highlighted in bold
* Significant at the 5 % level
specifications. That is, changing the controls does not change the parameter
estimates for the association of graduation rates and household type in large ways.
Table 5 shows the associations between family type and child graduation. In all
cases, the odds of a child with gay or lesbian parents completing high school are
lower, by a considerable margin, compared to children of married opposite sex
parents. For gay and lesbian households, adding the parental education controls to
the base controls lowers the odds of a child graduating for same-sex families. This is
because gay and lesbian homes are characterized by high levels of parent education
which contributes to child graduation, and so conditional on this the odds of a child
1\ Springer
000972
graduating are even lower. When all controls are used, including those for parental
marital status, the conditional graduation rate odds ratios are reasonably similar
between the two types of same-sex couples: 0.69 for gays and 0.60 for lesbians
using the I year mobility measure
37
The difference between the two point estimates
for gay and lesbian parents in column (3) is not significant. To put this in another
context, the marginal effect on the probability of graduating for children of same-
sex homes is a reduction of approximately 6-9 % points
38
The point estimates for
gay households are always statistically significant at the 5 %level, but the estimate
for lesbian households in column (3) is not.
39
Table 6 repeats the column (2) and (3) regressions of Table 5, but this time
separating girls and boys. This table shows that the particular gender mix of a same
sex household has a dramatic difference in the association with child graduation.
Consider the case of girls first. Regardless of the controls and whether or not girls
are currently living in a gay or lesbian household, the odds of graduating from high
school are considerably lower than any other household type. Indeed, girls living in
gay households are only 15 % as likely to graduate compared to girls from opposite
sex married homes. In all cases for girls the estimates are measured with precision.
The point estimates for boys are considerably different. Looking at equation (4) in
Table 6, boys in lesbian homes are 76 % as likely to graduate, while boys in gay
homes are 61 % more likely to graduate compared to boys in opposite sex married
homes. However, none of these estimates are statistically significant. the results from
Table 5 mask this gender difference, and the significant effect found in Table 5
column (3) for gay households is clearly being driven by the strong daughter effect.
The different results for the household gender mix are fascinating, especially since
this difference is not found in single parent households. Table 6 shows that boys do
better than girls in single parent homes, but the difference is not nearly as pronounced
as in same sex households. Looking at the unconditional graduation rates (with
standard errors in parentheses) for gay households, sons achieve 0.72 (0.074), while
daughters achieve 0.43 (0.090). For lesbian households, son's graduation rates are 0.48
(0.060), and daughter's have 0.55 (0.055). Based simply on these unconditional
measures, sons do better with fathers, and daughters do better with mothers.
37
They are also reasonably close to the unconditional estimated average graduation rates found in
Table 3. The odds ratios are .71 and .64 for the 5 year mobility measure.
38
The reported odds ratios are relative to children from opposite sex married parents. Compared to
children of opposite sex cohabitating parents, the children of same-sex parents do even worse. This can be
seen indirectly from Table 5. If cohabitating parents are the left out category, the odds ratio (standard
error) for high school graduation from a gay home is 0.61 (0.132), and 0.53 (0.138) from a lesbian
home-when a11 controls are used.
39
In order to further test the idea that lower graduation rates for children of gay and lesbian parents may
be the result of a negative environment, more controls were used for location. Rather than just control for
the province of residence, in an alternative specification the census metropolitan area was also controlled
for. For gay parents the odds ratio changes from 0.69 to 0.68 if the 1 year mobility control is used with all
other controls, and remains unchanged if the 5 year mobility control is used. For lesbian parents the odd
ration changes from 0.60 to 0.57, and from 0.64 to 0.58 depending on the mobility controL These
estimates have slightly lower standard errors.
<l Springer
000973
D. W. Allen
variable: I if child graduated from high school controlling for moved within past year)
Gir1s Boys
(I) (2) (3) (4)
Gay parents
Coefficient -1.939 -1.860 0.225 0.476
Std error (0.109)* (0.244)* (0.510) (0.491)
Odds ratio 0.14 0.15 1.25 1.61
Marginal effect -0.42 -0.40 0.04 0.08
Lesbian parents
Coefficient -0.913 -0.796 -0.883 -0.269
Std error (0.165)* (0.365)* (0.441)* (0.519)
Odds ratio 0.40 0.45 0.41 0.76
Marginal effect -0.17 -0.14 -0.20 -0.06
Common law
Coefficient -0.180 -0.072 -0.450 0.214
Std error (0.053)* (0.313) (0.079)* (0.109)
Odds ratio 0.83 0.93 0.64 1.23
Marginal effect -0.027 -0.010 -0.099 0.042
Single mother
Coefficient -0.663 -0.672 -0.723 -0.371
Std error (0.031)* (0.350) (0.049)* (0.064)*
Odds ratio 0.55 0.51 0.48 0.68
Marginal effect -0.099 -0.112 -0.160 -0.079
Single father
Coefficient -0.615 -0.761 -0.721 -0.274
Std error (0.048)* (0.379)* (0.042)* (0.092)*
Odds ratio 0.54 0.46 0.48 0.76
Marginal effect -0.106 -0.113 -0.164 -0.058
Child controls Yes Yes Yes Yes
Parent education controls Yes Yes Yes Yes
Marital status controls No Yes No Yes
Pseudo R
2
.26 .26 .21 .21
At this state, such a result is an interesting empirical finding, and one worthy of
further investigation. On the one hand, it seems this result is inconsistent with any
type of discrimination theory for the lower graduation rates among children of
same-sex households. Or, a discrimination theory would have to be modified to
include the household gender mix. Within the child development literature and pop
culture, there is the belief that mothers and fathers provide different parenting inputs
<l Springer
000974
that are not perfectly substitutable.4(' These results would be consistent with this
notion, but further research is necessary to show any causality.
4.2 School attendance
Any difference found in graduation rates may be the result of a selection bias-
children of same-sex families may be Jess likely to attend school. In this section,
school attendance is investigated. Table 7 reports on four Iogit regressions, where
the dependent variable equals I if the child attended school between September
2005 to May 2006; that is, if the child was in school during the previous year.
School attendance is mandatory in Canada until age I 5, thus children between
17-22 who do not attend have either quit, been expelled, or graduated already.
The structure of the first three regressions of Table 7 is the same as Table 5
41
The last regression in column ( 4) controls for the gender mix of the household.
In terms of the odds ratios results for unreported controls, being disabled or
moving residences both lead to a reduced chance of attending school, while being a
visible minority, older, and urban increase the odds of attending
42
Having a parent
who graduated school significantly increases the odds of a child attending.
In terms of the odds ratios reported in the table, once all controls are in place,
column (3) shows that each family type is more likely to have their children in school
compared to married parents (the omitted category). Indeed, lesbian households are
23 %more likely, while gays are about 16 %more likely. None of the column (3)
coefficients are statistically significant; that is, there is little statistical confidence in
the difference between married opposite sex and other family types when it comes to
child school attendance. Indeed, none of the odds ratios for any family type are
statistically different from each other. The bottom line from Table 7 is that in terms of
school attendance, a conclusion of "no difference," between children of gay, lesbian,
and married families is reasonable, and therefore, any difference in graduation rates is
unlikely caused by a selection bias based on attendance.
40
Within the literature, see Chrisp (2001), which addresses sons in lesbian homes. Within the popular
culture, see Modern Family, Season 4, Episode 19, where the gay couple Cam and Mitchell decide their
daughter Lily needs the input of aunt Gloria to discuss "girl issues."
41
For school attendance only the results for the I year mobility control are reported. The results for
controlling for 5 year mobility were virtua11y identical. An unreported regression on primary school
attendance found no difference between the different household types.
42
It might seem odd that the effect of Age is positive. However, the dependent variable is 1 if the child
ever attended school, or is now attending. Given that some students start school later than age 5, and that
many children are home schooled in primary divisions, a positive effect of Age is expected. If the
regression is run restricting the sample to students older than 12, the age effect is greatly removed. When
the sample is restricted to various age ranges (e.g., starting at ages 6-12, or ending at ages 17-22, the odds
ratios for the family type variables barely change at all and remain statistically indistinguishable.
<1 Springer
000975
D. W. Allen
Table 7 Odds ratios of school attendance (weighted observations, children ages 17-22, dependent
variable: 1 if child graduated from high school controlling for moved within past 5 years)
(I) (2) (3) (4)
Gay parents
Coefficient -0.019 0.006 0.149
Std error (0.259) (0.260) (0.311)
Odds ratio 0.980 1.05 1.16
Marginal effect 0.004 0.009 0.034
Daughter of gay parents
Coefficient 0.166
Std error (0.437)
Odds ratio 1.18
Marginal effect 0.038
Son of gay parents
Coefficient 0.737
Std error (0.287)*
Odds ratio 1.14
Lesbian parents
Coefficient -0.020 -0.013 0.208
Std error (0.084) (0.083) (0.137)
Odds ratio 0.979 0.986 1.23
Daughter of lesbian parents
Coefficient 0.277
Std error (0.144)
Odds ratio 1.31
Son of lesbian parents
Coefficient 0.143
Std error (0.250)
Odds ratio 1.15
Common law
Coefficient -0.179 -0.186 0.093 0.092
Std error (0.042)* (0.040)* (0.076) (0.076)
Odds ratio 0.835 0.829 1.09 1.09
Marginal effect -0.043 -0.044 0.022 0.021
Single mother
Coefficient -0.106 -0.100 0.091 0.091
Std error (0.025)* (0.277)* (0.076) (0.076)
Odds ratio 0.898 0.904 1.090 1.090
Marginal effect -0.025 -0.0249 0.021 0.021
fl Springer
000976
Table 7 continued
(1) (2) (3) (4)
Single father
Coefficient -0.213 -0.192 0.112 0.112
Std error (0.036)* (0.037) (0.082) (0.082)
Odds ratio 0.806 0.825 1.120 1.120
Marginal effect -0.041 -0.046 0.026 0.026
Child controls Yes Yes Yes Yes
Parent education controls No Yes Yes Yes
Marital status controls No No Yes Yes
Pseudo R
2
.II .II .12 .12
5 Conclusion
A casual reading of the literature on child performance would suggest that no-
difference in child outcomes exists between children in same-sex or opposite-sex
households. Indeed, the unanimous opinion of so many studies would appear
conclusive-as noted by Justice Walker. However, a closer inspection reveals that
there are really fifty-plus "preliminary" studies, and no general conclusion about
child performance differences is warranted based on the literature. The samples
used in these studies are often biased in some way, and the sample sizes are often
very small. The one study that did use a large random sample and address a reliable
performance measure (Rosenfeld 2010), turned out to draw the wrong conclusion,
did not compare gay versus lesbian homes, did not examine the gender mix of the
household, and did not control for parental marital status. As a result, there is little
hard evidence to support the general popular consensus of "no difference."
I have argued that the 2006 Canada census-though not perfect-is able to address
most of these issues, and the results on high school graduation rates suggest that
children living in both gay and lesbian households struggle compared to children from
opposite sex married households. In general, it appears that these children are only
about 65 % as likely to graduate from high school compared to the control group-a
difference that holds whether conditioned on controls or not.
43
When the households
are broken down by child gender it appears that daughters are struggling more than
sons, and that daughters of gay parents have strikingly low graduation rates.
This paper confirms the findings of Allen et a!. (2013), and taken together they cast
doubt on the ubiquitous claim that no difference exists in child outcomes for children
43
As mentioned, the census data has an imperfect measure of marital status. Those "currently married"
could be divorced from an earlier marriage. Given the higher marriage rate for opposite sex couples; the
estimated odds ratio on graduation rates for children of same-sex families may be biased upwards. The
true effect may be larger and more troubling.
<l Springer
000977
D. W. Allen
raised by same-sex parents compared to married opposite sex parents. That is, both the
US census and Canada census show that children living with same-sex parents perform
poorer in school when compared to children from married opposite sex families.
The question is: why? This study suggests further work is necessary to narrow
down the source of this difference. This will require an exceptional data set that not
only identifies sexual orientation of parents, but also has a retrospective or panel
design to completely control for marital history. A better data set would also be able
to test for the reasons behind any difference. An economist may be inclined to think
that fathers and mothers are not perfect substitutes and that there must be some
gains from a sexual division of labor in parenting. Others may suspect that children
of same-sex parents are more likely to be harassed at school, and therefore, less
likely to graduate. In any event, it is time to investigate the difference and reject the
conventional wisdom of "no difference."
Acknowledgments Thanks to Sonia Oreffice, Krishna Pendakur, and three journal referees for their
comments. This project was funded by the Social Sciences Research Council of Canada.
Appendix
See Table 8.
variable: I if child graduated from high school controlling for moved within past 5 years)
(I) (2) (3)
Gay parents 0.66 0.55 0.71
(0.208) (0.132)* (0.135)
Lesbian parents 0.47 0.42 0.64
(0.085)* (O.Q95)* (0.205)
Common law 0.65 0.73 1.17
(0.208)* (0.048)* (0.131)*
Single mother 0.53 0.52 0.65
(0.007)* (0.021)* (0.082)*
Single father 0.55 0.51 0.66
(0.012)* (0.021)* (0.106)*
Child controls Yes Yes Yes
Parent education controls No Yes Yes
Marital status controls No No Yes
Pseudo R
2
.21 .23 .23
* Significant at the 5 % Jevel. z scores in parentheses
ifl Springer
000978
References
Allen, D. W. (2012). More heat than light: A critical assessment of the same-sex parenting literature,
1995-2012. Working paper, Simon Fraser University
Allen, D. W .. , Pakaluk, C., & Price, J. (2013). Nontraditional families and childhood progress through
school: A comment on Rosenfeld. Demography, 50(3), 955-961.
Andersson, G., Noack, T., Seierstad, A., & Weedon-Fekjaer, H. (2006). The demographics of same-sex
marriages in Norway and Sweden. Demography, 43(1), 79-98.
Badgett, M. V. L, Gates, G., & Maisel, N. (2008). Registered domestic partnerships among gay men and
lesbians: The role of economic factors. Review of Economics of the Household, 6(4), 327-346.
Bailey, J. M., Bobrow, D., Wolfe, M., & Mikach, S. (1995). Sexual orientation of adult sons of gay
fathers. Developmental Psychology, 31(1), 124-129.
Balsam, K., Beauchaine, T., Rothblum, E., & Solomon, S. (2008). Three-year follow-up of same-sex
couples who had civil unions in Vermont, same-sex couples not in civil unions, and heterosexual
married couples. Developmental Psychology, 44(1), 102-116.
Barrett, H., & Tasker, F. (2001). Growing up with a gay parent: Views of 101 gay fathers on their sons'
and daughters' experiences. Educational and Child Psychology, I8(1), 62-77.
Becker, G. (1981). A treatise on the family. Cambridge: Harvard University Press.
Biblarz, T., & Stacey, J. (2010). How does the gender of parents matter?. Journal of Marriage and
Family, 72, 3-22.
Biblarz, T., & Savci, E. (2010). Lesbian, gay, bisexual, and trans gender families. Journal of Marriage and
Family, 72, 480-497.
Black, D., Sanders, S., & Taylor, L (2007). The economics of lesbian and gay families. Journal of
Economic Perspectives., 21(2), 53-70.
Black, D., Sanders, S., & Taylor, L (2006). The measurement of same-sex unmarried partner couples in
the 2000 US Census. California Center for Population Research Working Paper, 2007b.
Booth, A., & Frank, J. (2008). Marriage, partnership and sexual orientation: A study of British university
academics and administrators. Review of Economics of the Household, 6(4), 409-422.
Bos, H. (2010). Planned gay father families in kinship arrangements. Australia and New Zealand Journal
of Family Therapy, 31(4), 356-371.
Bos, H., & van Balen, F. (2008). Children in planned lesbian families: Stigmatisation, psychological
adjustment and protective factors. Culture, Health, and Sexuality, 10(3), 221-236.
Bos, H., van Balen, F., & van den Boom, D. (2007). Child adjustment and parenting in planned lesbian
parent families. American Journal of Orthopsychiatry, 77(1 ), 38-48.
Bos, H., van Balen, F., van den Boom, D., & Sandfort, Th. (2004). Minority stress, experience of
parenthood and child adjustment in lesbian families. Journal of Reproductive and Infant
Psychology, 22(4), 1-14.
Bos, H., van Balen, F., Gartrell, N., Peyser, H., & Sandfort, T. (2008). Children in planned lesbian
families: A cross-cultural comparison between the United States and the Netherlands. American
Journal of Orthopsychiatry, 78(2), 211-219.
Brewaeys, A., Ponjaert, I., Van Hall, E. V., & Golombok, S. (1997). Donor insemination: Child
development and family functioning in lesbian mother families. Human Reproduction, 12(6),
1349-1359.
Carpenter, C., & Gates, G. (2008). Gay and lesbian partnership: Evidence from California. Demography,
45(3), 573-590.
Chan, R., Brooks, R., Raboy, B., & Patterson, C. (1998a). Division of labor among lesbian and
heterosexual parents: Associations with children's adjustment. Journal of Family Psychology, 12(3),
402-419.
Chan, R., Raboy, B., & Patterson, C. (1998b). Psychosocial adjustment among children conceived via
donor insemination by lesbian and heterosexual mothers. Child Development, 69(2), 443-457.
Chrisp, J. (2001). That four letter word-sons: Lesbian mothers and adolescent son. Journal of Lesbian
Studies, 5(1-2), 195-209.
Clarke, V., Kitzinger, C., & Potter, J. (2004). Kids are just cruel anyway: Lesbian and gay parents talk
about homophobic bullying. British Journal of Social Psychology, 43, 531-550.
Cloughessy, K. (2010). Sorry, but you're not a mother: An examination of the validity of the de facto
threshold in determining motherhood for the non-birth mother in lesbian-parented families. Gay&
Lesbian Issues and Psychology Review, 6(2), 82-90.
421 Springer
000979
D. W. Allen
Dundas, S., & Kaufman, M. (2000). The Toronto lesbian family study. Journal of Homosexuality, 40,
65-79.
Fairtlough, A. (2008). Growing up with a lesbian or gay parent: Young people's perspectives. Health and
Social Care in the Community, 16(5), 521-528.
Flaks, D., Fieber, I., Masterpasqua, F., & Joseph, G. (1995). Lesbians choosing motherhood: A
comparative study of lesbian and heterosexual parents and their children. Developmental
Psychology, 31(1), 105-114.
Fulcher, M., Sutfin, E., & Patterson, C. (2008). Individual differences in gender development:
Associations with parental sexual orientation, attitudes, and division of labor. Sex Roles, 58,
330-341.
Fulcher, M., Chan, R., Raboy, B., & Patterson, C. (2002). Contact with grandparents among children
conceived via donor insemination by lesbian and heterosexual mothers. Parenting: Science and
Practice, 2(1), 61-76.
Gartrell, N., Bos, H. (2010). US national longitudinal lesbian family study: Psychological adjustment of
17 year old adolescents. Pediatrics, doi: 10.1542/peds.2009-3153.
Gartrell, N., Deck, A., Rodas, C., & Peyser, H. (2005). The national lesbian family study 4: Interviews
with the 10-year old children. American Journal of Orthopsychiatry, 75(4), 518-524.
Gartrell, N., Banks, A., Reed, N., Hamilton, J., Rodas, C., & Deck, A. (2000). The national lesbian family
study 3: Interviews with mothers of five-year olds. American Journal of Orthopsychiatry, 70(4),
542-548.
Gartrell, N., Banks, A., Reed, N., Hamilton, J., Rodas, C., & Biship, H. (1999). The national lesbian
family study: 2. Interviews with mothers of toddlers. American Journal of Orthopsychiatry, 69(3),
362-369.
Gershon, T., Tschann, J., & Jemerin, J. (1999). Stigmatization, self-esteem, and coping among the
adolescent children of lesbian mothers. Journal of Adolescent Health, 24, 437-445.
Goldberg, A. (2010). Lesbian and gay parents and their children: Research on the family life cycle.
Washington, DC: American Psychological Association.
Goldberg, A. (2007). (How) does it make a difference? Perspectives of adults with lesbian, gay, and
bisexual parents. American Journal of Orthopsychiatry, 77(4), 550--562.
Goldberg, A., Downing, J., & Sauck, C. (2008). Perceptions of children's parental preferences in lesbian
two-mother households. Journal of Marriage and Family, 70, 419-434.
Golombok, S., Perry, B., Burston, A., Murray, C., Mooney-Somers, J., Stevens, M., Golding, J.et al.
(2003). Children with lesbian parents: A community study. Developmental Psychology, 39(1 ),
20-33.
Golombok, S., & Tasker, F. (1996). Do parents influence the sexual orientation of their children? Findings
from a longitudinal study of lesbian families. Developmental Psychology, 32(1), 3-11.
Golombok, S., Tasker, F., & Murray, C. (1997). Children raised in fatherless families from Infancy:
Family relationships and the socioemotional development of children of lesbian and single
heterosexual mothers. Journal of Child Psychology and Psychiatry, 38, 783-791.
Jepsen, C., & Jepsen, L. (2009). Does home ownership vary by sexual orientation?. Regional Science and
Urban Economics, 39(3), 307-315.
Klawitter, M. (2008). The effects of sexual orientation and marital status on how couples hold their
money. Review of Economics and the Household, 6(4), 423-446.
Lehmiller, J. (2010). Differences in relationship investments between gay and heterosexual men.
Personal Relationships, 17(1),
Leung, P., Erich, S., & Kanenberg, H. (2005). A comparison of family functions in gay/lesbian,
heterosexual and special needs adoptions. Children and Youth Services Review, 27,
Lev, A. (2008). More than surface tension: Femmes in families. Journal of Lesbian Studies, 12(2-3),
126-143.
Lewin, E. (2006). Book review: 'The family of woman: Lesbian mothers, their children and the undoing
of gender' by Maureen Sullivan. Theory and Society, 35, 601-605.
MacCallum, F., & Golombok, S. (2004). Children raised in fatherless families from infancy: A follow-up
of children of lesbian and single heterosexual mothers at early adolescence. Journal of Child
Psychology and Psychiatry, 45(8), 1407-1419.
Marks, L. (2012). Same-sex parenting and children's outcomes: A closer examination of the American
Psychological Association's brief on lesbian and gay parenting. Social Science Research, 41(4),
735-751.
.:1 Springer
000980
McNeill, K., & Rienzi, B. ( 1998). Families and parenting: A comparison of lesbian and heterosexual
mothers. Psychological Reports, 82, 59-62.
Negrusa, B., & Oreffice, S. (2011). Sexual orientation and household financial decisions: Evidence from
couples in the United States. Review of Economics of the Household, I 8(2), 445-463.
Nock, S. L. (2001). Sworn affidavit of Stephen Lowell Nock. Ontario Superior Court of Justice. Between
Hedy Halpern et al. and the Attorney General of Canada et al.: Court File No. 684/00.
Oreffice, S. (2010). Sexual orientation and household decision making. Same-sex couples balance of
power and labor supply choices. Labour Economics, 18(2), 145-158.
Oswald, R., Goldberg, A, Kuvalanka, K., & Clausell, E. (2008). Structural and moral commitment
among same-sex couples: Relationship duration, religiosity, and parental status. Journal of Family
Psychology, 22(3), 411-419.
Patterson, C. (1995). Families of the lesbian baby boom: Parent's division of labor and children's
adjustment. Developmental Psychology, 31(1), 411-419.
Patterson, C. (2001). Families of the lesbian baby boom: Maternal mental health and child adjustment.
Journal of Gay & Lesbian Psychotherapy, 4(314), 91-107.
Patterson, C., Hurt, S., & Mason, C. (1998). Families of the lesbian baby boom: Children's contact with
grandparents and other adults. American Journal of Orthopsychiatry, 68(3), 390--399.
Patterson, C., Sutfin, E., & Fulcher, M. (2004). Division of labor among lesbian and heterosexual
parenting couples: Correlates of specialized versus shared patterns. Journal of Adult Development,
11(3), 179-189.
Power, J., Perlesz, A., Brown, R., Schofield, M., Pitts, M., Mcnair, R., Bickerdike, Aet al. (2010).
Diversity, tradition and family: Australian same sex attracted parents and their families. Gay&
Lesbian Issues and Psychology Review, 6(2), 66-81.
Regnerus, M. (2012). How different are the adult children of parents who have same-sex relationships?
Findings from the New Family Structures Study. Social Science Research, 41(4), 752-770.
Rivers, L, Poteat, V. P., & Noret, N. (2008). Victimization, social support, and psychosocial functioning
among children of same-sex and opposite-sex couples in the United Kingdom. Developmental
Psychology, 44(1 ), 127-134.
Rothblum, E., Balsam, K., & Solomon, S. (2008). Comparison of same-sex couples who were married in
Massachusetts, had domestic partnerships in California, or had civil unions in Vermont. Journal of
Family Issues, 29(1), 48-78.
Rosenfeld, M. (20 1 0). Nontraditional families and childhood progress through school. Demography,
47(3), 755-775.
Sarantkos, S. (1996). Children in three contexts: Family, education, and social development. Children
Australia, 21, 23-31.
Scheib, J., Riordan, M., & Rubin, S. (2005). Adolescents with open-identity sperm donors: Reports from
12-17 year olds. Human Reproduction, 20(1), 239-252.
Stacey, J. (2005). The families of man: Gay male intimacy and kinship in a global metropolis. Journal of
Women in Culture and Society, 30(3), 1911-1937.
Stacey, J. (2004). Cruising to family land: Gay hypergamy and rainbow kinship. Current Sociology, 52(2),
181-197.
Stacey, J., & Biblarz, T. (2001). (How) does the sexual orientation of parents matter?. American
Sociological Review, 66, 159-183.
Sutfin, E. L., Fulcher, M., Bowles, R. P., & Patterson, C. J. (2008). How lesbian and heterosexual parents
convey attitudes about gender to their children: The role of gendered environments. Sex Roles, 58,
501-513.
Sweet, M. (2009). The science of unisex parenting: A review of published studies. (unpublished
manuscript).
Tasker, F., & Golombok, S. (1995). Adults raised as children in lesbian families. American Journal of
Orthopsychiatry, 65(2), 203-215.
Vanfraussen, K., Ponjaert-Kristoffersen, I., & Brewaeys, A. (2002). What does it mean for youngsters to
grow up in a lesbian family created by means of donor insemination?. Journal of Reproductive and
Infant Psychology, 20(4), 237-252.
Wainright, J., & Patterson, C. (2006). Delinquency, victimization, and substance use among adolescents
with female same sex parents. Journal of Family Psychology, 20(3), 526-530.
Wainright, J., & Patterson, C. (2008). Peer relations among adolescents with female same-sex parents.
Developmental Psychology, 44(1), 117-126.
<\ Springer
000981
D. W. Allen
Wainright, J., Russell, S., & Patterson, C. (2004). Psychosocial adjustment, school outcomes, and
romantic relationships of adolescents with same sex parents. Child Development, 75(6), 1886--1898.
Wright, E., & Perry, B. (2006). Sexual identity distress, social support, and the health of gay, lesbian, and
bisexual youth. Journal of Homosexuality, 51(1), 81-110.
Zavodny, M. (2008). Is there a 'maniage premium' for gay men?. Review of Economics of the Household,
6(4), 369-389.
<l Springer
000982

TAB 57

In their 2001 American Sociological Review article reviewing ndings on sexual orientation and parenting, however, soci-
ologists Judith Stacey and Tim Biblarz began noting that while there are some differences in outcomes between children in
same-sex and heterosexual unions, there were not as many as family sociologists might expect, and differences need not
necessarily be perceived as decits. Since that time the conventional wisdom emerging from comparative studies of
same-sex parenting is that there are very few differences of note in the child outcomes of gay and lesbian parents (Tasker,
2005; Wainright and Patterson, 2006; Rosenfeld, 2010). Moreover, a variety of possible advantages of having a lesbian couple
as parents have emerged in recent studies (Crowl et al., 2008; Biblarz and Stacey, 2010; Gartrell and Bos, 2010; MacCallum
and Golombok, 2004). The scholarly discourse concerning gay and lesbian parenting, then, has increasingly posed a challenge
to previous assumptions about the supposed benets of being raised in biologically-intact, two-parent heterosexual
households.
1.1. Sampling concerns in previous surveys
Concern has arisen, however, about the methodological quality of many studies focusing on same-sex parents. In partic-
ular, most are based on non-random, non-representative data often employing small samples that do not allow for gener-
alization to the larger population of gay and lesbian families (Nock, 2001; Perrin and Committee on Psychosocial Aspects
of Child and Family Health, 2002; Redding, 2008). For instance, many published studies on the children of same-sex parents
collect data from snowball or convenience samples (e.g., Bos et al., 2007; Brewaeys et al., 1997; Fulcher et al., 2008; Sirota,
2009; Vanfraussen et al., 2003). One notable example of this is the National Longitudinal Lesbian Family Study, analyses of
which were prominently featured in the media in 2011 (e.g., Hufngton Post, 2011). The NLLFS employs a convenience sam-
ple, recruited entirely by self-selection from announcements posted at lesbian events, in womens bookstores, and in les-
bian newspapers in Boston, Washington, and San Francisco. While I do not wish to downplay the signicance of such a
longitudinal studyit is itself quite a featthis sampling approach is a problem when the goal (or in this case, the practical
result and conventional use of its ndings) is to generalize to a population. All such samples are biased, often in unknown
ways. As a formal sampling method, snowball sampling is known to have some serious problems, one expert asserts (Snij-
ders, 1992, p. 59). Indeed, such samples are likely biased toward inclusion of those who have many interrelationships with,
or are coupled to, a large number of other individuals (Berg, 1988, p. 531). But apart from the knowledge of individuals
inclusion probability, unbiased estimation is not possible.
Further, as Nock (2001) entreated, consider the convenience sample recruited from within organizations devoted to
seeking rights for gays and lesbians, like the NLLFS sampling strategy. Suppose, for example, that the respondents have
higher levels of education than comparable lesbians who do not frequent such events or bookstores, or who live else-
where. If such a sample is used for research purposes, then anything that is correlated with educational attainmentlike
better health, more deliberative parenting, and greater access to social capital and educational opportunities for children
will be biased. Any claims about a population based on a group that does not represent it will be distorted, since its sam-
ple of lesbian parents is less diverse (given what is known about it) than a representative sample would reveal (Baumle
et al., 2009).
To compound the problem, results from nonprobability samplesfrom which meaningful statistics cannot be generated
are regularly compared with population-level samples of heterosexual parents, which no doubt are comprised of a blend of
higher and lower quality parents. For example, Gartrell et al. (2011a,b) inquired about the sexual orientation and behavior of
adolescents by comparing data from the National Survey of Family Growth (NSFG) with those in the snowball sample of
youth in the NLLFS. Comparing a population-based sample (the NSFG) to a select sample of youth from same-sex parents
does not provide the statistical condence demanded of good social science. Until now, this has been a primary way in which
scholars have collected and evaluated data on same-sex parents. This is not to suggest that snowball samples are inherently
problematic as data-collection techniques, only that they are not adequate for making useful comparisons with samples that
are entirely different with regard to selection characteristics. Snowball and various other types of convenience sampling are
simply not widely generalizable or comparable to the population of interest as a whole. While researchers themselves com-
monly note this important limitation, it is often entirely lost in the translation and transmission of ndings by the media to
the public.
1.2. Are there notable differences?
The no differences paradigm suggests that children from same-sex families display no notable disadvantages when
compared to children from other family forms. This suggestion has increasingly come to include even comparisons with
intact biological, two-parent families, the form most associated with stability and developmental benets for children
(McLanahan and Sandefur, 1994; Moore et al., 2002).
Answering questions about notable between-group differences has nevertheless typically depended on with whom com-
parisons are being made, what outcomes the researchers explored, and whether the outcomes evaluated are considered sub-
stantial or supercial, or portents of future risk. Some outcomeslike sexual behavior, gender roles, and democratic
parenting, for examplehave come to be valued differently in American society over time.
For the sake of brevityand to give ample space here to describing the NFSSI will avoid spending too much time
describing previous studies, many of whose methodological challenges are addressed by the NFSS. Several review articles,
M. Regnerus / Social Science Research 41 (2012) 752770 753
000984
and at least one book, have sought to provide a more thorough assessment of the literature (Anderssen et al., 2002; Biblarz
and Stacey, 2010; Goldberg, 2010; Patterson, 2000; Stacey and Biblarz, 2001a). Sufce it to say that versions of the phrase
no differences have been employed in a wide variety of studies, reports, depositions, books, and articles since 2000 (e.g.,
Crowl et al., 2008; Movement Advancement Project, 2011; Rosenfeld, 2010; Tasker, 2005; Stacey and Biblarz, 2001a,b;
Veldorale-Brogan and Cooley, 2011; Wainright et al., 2004).
Much early research on gay parents typically compared the child development outcomes of divorced lesbian mothers
with those of divorced heterosexual mothers (Patterson, 1997). This was also the strategy employed by psychologist Fiona
Tasker (2005), who compared lesbian mothers with single, divorced heterosexual mothers and found no systematic differ-
ences between the quality of family relationships therein. Wainright et al. (2004), using 44 cases in the nationally-repre-
sentative Add Health data, reported that teenagers living with female same-sex parents displayed comparable self-
esteem, psychological adjustment, academic achievement, delinquency, substance use, and family relationship quality to
44 demographically matched cases of adolescents with opposite-sex parents, suggesting that here too the comparisons
were not likely made with respondents from stable, biologically-intact, married families.
However, small sample sizes can contribute to no differences conclusions. It is not surprising that statistically-signi-
cant differences would not emerge in studies employing as few as 18 or 33 or 44 cases of respondents with same-sex parents,
respectively (Fulcher et al., 2008; Golombok et al., 2003; Wainright and Patterson, 2006). Even analyzing matched samples,
as a variety of studies have done, fails to mitigate the challenge of locating statistically-signicant differences when the sam-
ple size is small. This is a concern in all of social science, but one that is doubly important when there may be motivation to
conrm the null hypothesis (that is, that there are in fact no statistically-signicant differences between groups). Therefore,
one important issue in such studies is the simple matter of if there is enough statistical power to detect meaningful differ-
ences should they exist. Rosenfeld (2010) is the rst scholar to employ a large, random sample of the population in order to
compare outcomes among children of same-sex parents with those of heterosexual married parents. He concludedafter
controlling for parents education and income and electing to limit the sample to households exhibiting at least 5 years of
co-residential stabilitythat there were no statistically-signicant differences between the two groups in a pair of measures
assessing childrens progress through primary school.
Sex-related outcomes have more consistently revealed distinctions, although the tone of concern about them has dimin-
ished over time. For example, while the daughters of lesbian mothers are now widely understood to be more apt to explore
same-sex sexual identity and behavior, concern about this nding has faded as scholars and the general public have become
more accepting of GLB identities (Goldberg, 2010). Tasker and Golombok (1997) noted that girls raised by lesbian mothers
reported a higher number of sexual partners in young adulthood than daughters of heterosexual mothers. Boys with lesbian
mothers, on the other hand, appear to display the opposite trendfewer partners than the sons of heterosexual mothers.
More recently, however, the tone about no differences has shifted some toward the assertion of differences, and that
same-sex parents appear to be more competent than heterosexual parents (Biblarz and Stacey, 2010; Crowl et al., 2008).
Even their romantic relationships may be better: a comparative study of Vermont gay civil unions and heterosexual mar-
riages revealed that same-sex couples report higher relationship quality, compatibility, and intimacy, and less conict than
did married heterosexual couples (Balsam et al., 2008). Biblarz and Staceys (2010) review article on gender and parenting
asserts that,
based strictly on the published science, one could argue that two women parent better on average than a woman and a
man, or at least than a woman and man with a traditional division of labor. Lesbian coparents seem to outperform com-
parable married heterosexual, biological parents on several measures, even while being denied the substantial privileges
of marriage (p. 17).
Even here, however, the authors note that lesbian parents face a somewhat greater risk of splitting up, due, they sug-
gest, to their asymmetrical biological and legal statuses and their high standards of equality (2010, p. 17).
Another meta-analysis asserts that non-heterosexual parents, on average, enjoy signicantly better relationships with
their children than do heterosexual parents, together with no differences in the domains of cognitive development, psycho-
logical adjustment, gender identity, and sexual partner preference (Crowl et al., 2008).
However, the meta-analysis reinforces the profound importance of who is doing the reportingnearly always volunteers
for small studies on a group whose claims about documentable parenting successes are very relevant in recent legislative
and judicial debates over rights and legal statuses. Tasker (2010, p. 36) suggests caution:
Parental self-report, of course, may be biased. It is plausible to argue that, in a prejudiced social climate, lesbian and gay
parents may have more at stake in presenting a positive picture. . ..Future studies need to consider using additional
sophisticated measures to rule out potential biases. . .
Sufce it to say that the pace at which the overall academic discourse surrounding gay and lesbian parents comparative
competence has shiftedfrom slightly-less adept to virtually identical to more adeptis notable, and rapid. By comparison,
studies of adoptiona common method by which many same-sex couples (but more heterosexual ones) become parents
have repeatedly and consistently revealed important and wide-ranging differences, on average, between adopted children
and biological ones. In fact, these differences have been so pervasive and consistent that adoption experts now emphasize
that acknowledgement of difference is critical for both parents and clinicians when working with adopted children and
754 M. Regnerus / Social Science Research 41 (2012) 752770
000985
teens (Miller et al., 2000). This ought to give social scientists studying gay parenting outcomes pause, especially in light of
concerns noted above about small sample sizes and the absence of a comparable recent, documented improvement in out-
comes from youth in adopted families and stepfamilies.
Far more, too, is known about the children of lesbian mothers than about those of gay fathers (Biblarz and Stacey, 2010;
Patterson, 2006; Veldorale-Brogan and Cooley, 2011). Biblarz and Stacey (2010, p. 17) note that while gay-male families re-
main understudied, their daunting routes to parenthood seem likely to select more for strengths than limitations. Others
are not so optimistic. One veteran of a study of the daughters of gay fathers warns scholars to avoid overlooking the family
dynamics of emergent gay parents, who likely outnumber planned ones: Children born into heterosexually organized
marriages where fathers come out as gay or bisexual also face having to deal with maternal bitterness, marital conict, pos-
sible divorce, custody issues, and fathers absence (Sirota, 2009, p. 291).
Regardless of sampling strategy, scholars also know much less about the lives of young-adult children of gay and lesbian
parents, or how their experiences and accomplishments as adults compare with others who experienced different sorts of
household arrangements during their youth. Most contemporary studies of gay parenting processes have focused on the
presentwhat is going on inside the household when children are still under parental care (Tasker, 2005; Bos and Sandfort,
2010; Brewaeys et al., 1997). Moreover, such research tends to emphasize parent-reported outcomes like parental divisions of
labor, parentchild closeness, daily interaction patterns, gender roles, and disciplinary habits. While such information is
important to learn, it means we know far more about the current experience of parents in households with children than
we do about young adults who have already moved through their childhood and nowspeak for themselves. Studies on family
structure, however, serve scholars and family practitioners best when they span into adulthood. Do the children of gay and
lesbian parents look comparable to those of their heterosexual counterparts? The NFSS is poised to address this question
about the lives of young adults between the ages of 18 and 39, but not about children or adolescents. While the NFSS is
not the answer to all of this domains methodological challenges, it is a notable contribution in important ways.
1.3. The New Family Structures Study
Besides being brand-new data, several other aspects about the NFSS are novel and noteworthy. First, it is a study of young
adults rather than children or adolescents, with particular attention paid to reaching ample numbers of respondents who
were raised by parents that had a same-sex relationship. Second, it is a much larger study than nearly all of its peers. The
NFSS interviewed just under 3000 respondents, including 175 who reported their mother having had a same-sex romantic
relationship and 73 who said the same about their father. Third, it is a weighted probability sample, from which meaningful
statistical inferences and interpretations can be drawn. While the 2000 (and presumably, the 2010) US Census Integrated
Public Use Microdata Series (IPUMS) offers the largest nationally-representative sample-based information about youth in
same-sex households, the Census collects much less outcome information of interest. The NFSS, however, asked numerous
questions about respondents social behaviors, health behaviors, and relationships. This manuscript provides the rst
glimpse into those outcomes by offering statistical comparisons of them among eight different family structures/experiences
of origin. Accordingly, there is much that the NFSS offers, and not just about the particular research questions of this study.
There are several things the NFSS is not. The NFSS is not a longitudinal study, and therefore cannot attempt to broach
questions of causation. It is a cross-sectional study, and collected data from respondents at only one point in time, when they
were between the ages of 18 and 39. It does not evaluate the offspring of gay marriages, since the vast majority of its respon-
dents came of age prior to the legalization of gay marriage in several states. This study cannot answer political questions
about same-sex relationships and their legal legitimacy. Nevertheless, social science is a resource that offers insight to polit-
ical and legal decision-makers, and there have been enough competing claims about what the data says about the children
of same-sex parentsincluding legal depositions of social scientists in important casesthat a study with the methodolog-
ical strengths of this one deserves scholarly attention and scrutiny.
2. Data collection, measures, and analytic approach
The NFSS data collection project is based at the University of Texas at Austins Population Research Center. A survey de-
sign team consisting of several leading family researchers in sociology, demography, and human developmentfrom Penn
State University, Brigham Young University, San Diego State University, the University of Virginia, and several from the
University of Texas at Austinmet over 2 days in January 2011 to discuss the projects sampling strategy and scope, and con-
tinued to offer advice as questions arose over the course of the data collection process. The team was designed to merge
scholars across disciplines and ideological lines in a spirit of civility and reasoned inquiry. Several additional external con-
sultants also gave close scrutiny to the survey instrument, and advised on how best to measure diverse topics. Both the study
protocol and the questionnaire were approved by the University of Texas at Austins Institutional Review Board. The NFSS
data is intended to be publicly accessible and will thus be made so with minimal requirements by mid-late 2012. The NFSS
was supported in part by grants from the Witherspoon Institute and the Bradley Foundation. While both of these are com-
monly known for their support of conservative causesjust as other private foundations are known for supporting more
liberal causesthe funding sources played no role at all in the design or conduct of the study, the analyses, the interpreta-
tions of the data, or in the preparation of this manuscript.
000986
2.1. The data collection process
The data collection was conducted by Knowledge Networks (or KN), a research rm with a very strong record of gener-
ating high-quality data for academic projects. Knowledge Networks recruited the rst online research panel, dubbed the
KnowledgePanel
, that is representative of the US population. Members of the KnowledgePanel
are randomly recruited

by telephone and mail surveys, and households are provided with access to the Internet and computer hardware if needed.
Unlike other Internet research panels sampling only individuals with Internet access who volunteer for research, the Knowl-
edgePanel
is based on a sampling frame which includes both listed and unlisted numbers, those without a landline tele-
phone and is not limited to current Internet users or computer owners, and does not accept self-selected volunteers. As a
result, it is a random, nationally-representative sample of the American population. At last count, over 350 working papers,
conference presentations, published articles, and books have used Knowledge Networks panels, including the 2009 National
Survey of Sexual Health and Behavior, whose extensive results were featured in an entire volume of the Journal of Sexual
Medicineand prominently in the mediain 2010 (Herbenick et al., 2010). More information about KN and the Knowledge-
Panel
, including panel recruitment, connection, retention, completion, and total response rates, are available from KN. The
typical within survey response rate for a KnowledgePanel
survey is 65%. Appendix A presents a comparison of age-appro-

priate summary statistics from a variety of socio-demographic variables in the NFSS, alongside the most recent iterations of
the Current Population Survey, the National Longitudinal Study of Adolescent Health (Add Health), the National Survey of
Family Growth, and the National Study of Youth and Religionall recent nationally-representative survey efforts. The esti-
mates reported there suggest the NFSS compares very favorably with other nationally-representative datasets.
2.2. The screening process
Particularly relevant for the NFSS is the fact that key populationsgay and lesbian parents, as well as heterosexual adop-
tive parentscan be challenging to identify and locate. The National Center for Marriage and Family Research (2010) esti-
mates that there are approximately 580,000 same-sex households in the United States. Among them, about 17%or
98,600are thought to have children present. While that may seem like a substantial number, in population-based sampling
strategies it is not. Locating minority populations requires a search for a probability sample of the general population, typ-
ically by way of screening the general population to identify members of rarer groups. Thus in order to boost the number of
respondents who reported being adopted or whose parent had a same-sex romantic relationship, the screener survey (which
distinguished such respondents) was left in the eld for several months between July 2011 and February 2012, enabling
existing panelists more time to be screened and new panelists to be added. Additionally, in late Fall 2011, former members
of the KnowledgePanel
were re-contacted by mail, phone, and email to encourage their screening. A total of 15,058 current
and former members of KNs KnowledgePanel
were screened and asked, among several other questions, From when you
were born until age 18 (or until you left home to be on your own), did either of your parents ever have a romantic relation-
ship with someone of the same sex? Response choices were Yes, my mother had a romantic relationship with another wo-
man, Yes, my father had a romantic relationship with another man, or no. (Respondents were also able to select both of
the rst two choices.) If they selected either of the rst two, they were asked about whether they had ever lived with that
parent while they were in a same-sex romantic relationship. The NFSS completed full surveys with 2988 Americans between
the ages of 18 and 39. The screener and full survey instrument is available at the NFSS homepage, located at: www.prc.utex-
as.edu/nfss.
2.3. What does a representative sample of gay and lesbian parents (of young adults) look like?
The weighted screener dataa nationally-representative samplereveal that 1.7% of all Americans between the ages of
18 and 39 report that their father or mother has had a same-sex relationship, a gure comparable to other estimates of chil-
dren in gay and lesbian households (e.g., Stacey and Biblarz (2001a,b) report a plausible range from 1% to 12%). Over twice as
many respondents report that their mother has had a lesbian relationship as report that their fathers have had a gay rela-
tionship. (A total of 58% of the 15,058 persons screened report spending their entire youthup until they turned 18 or left
the housewith their biological mother and father.)
While gay and lesbian Americans typically become parents today in four waysthrough one partners previous partici-
pation in a heterosexual union, through adoption, in-vitro fertilization, or by a surrogatethe NFSS is more likely to be com-
prised of respondents from the rst two of these arrangements than from the last two. Todays children of gay men and
lesbian women are more apt to be planned (that is, by using adoption, IVF, or surrogacy) than as little as 1520 years
ago, when such children were more typically the products of heterosexual unions. The youngest NFSS respondents turned
18 in 2011, while the oldest did so in 1990. Given that unintended pregnancy is impossible among gay men and a rarity
among lesbian couples, it stands to reason that gay and lesbian parents today are far more selective about parenting than
the heterosexual population, among whom unintended pregnancies remain very common, around 50% of total (Finer and
Henshaw, 2006). The share of all same-sex parenting arrangements that is planned, however, remains unknown. Although
the NFSS did not directly ask those respondents whose parent has had a same-sex romantic relationship about the manner of
000987
their own birth, a failed heterosexual union is clearly the modal method: just under half of such respondents reported that
their biological parents were once married. This distinguishes the NFSS from numerous studies that have been entirely con-
cerned with planned gay and lesbian families, like the NLLFS.
Among those who said their mother had a same-sex relationship, 91% reported living with their mother while she was in
the romantic relationship, and 57% said they had lived with their mother and her partner for at least 4 months at some point
prior to age 18. A smaller share (23%) said they had spent at least 3 years living in the same household with a romantic part-
ner of their mothers.
Among those who said their father had a same-sex relationship, however, 42% reported living with him while he was in a
same-sex romantic relationship, and 23% reported living with him and his partner for at least 4 months (but less than 2% said
they had spent at least 3 years together in the same household), a trend similarly noted in Taskers (2005) review article on
gay and lesbian parenting.
Fifty-eight (58) percent of those whose biological mothers had a same-sex relationship also reported that their biological
mother exited the respondents household at some point during their youth, and just under 14% of them reported spending
time in the foster care system, indicating greater-than-average household instability. Ancillary analyses of the NFSS suggests
a likely planned lesbian origin of between 17% and 26% of such respondents, a range estimated from the share of such
respondents who claimed that (1) their biological parents were never married or lived together, and that (2) they never lived
with a parental opposite-sex partner or with their biological father. The share of respondents (whose fathers had a same-sex
relationship) that likely came from planned gay families in the NFSS is under 1%.
These distinctions between the NFSSa population-based sampleand small studies of planned gay and lesbian families
nevertheless raise again the question of just how unrepresentative convenience samples of gay and lesbian parents actually
are. The use of a probability sample reveals that the young-adult children of parents who have had same-sex relationships
(in the NFSS) look less like the children of todays stereotypic gay and lesbian coupleswhite, uppermiddle class, well-edu-
cated, employed, and prosperousthan many studies have tacitly or explicitly portrayed. Goldberg (2010, pp. 1213) aptly
notes that existing studies of lesbian and gay couples and their families have largely included white, middle-class persons
who are relatively out in the gay community and who are living in urban areas, while working-class sexual minorities,
racial or ethnic sexual minorities, sexual minorities who live in rural or isolated geographical areas have been overlooked,
understudied, and difcult to reach. Rosenfelds (2010) analysis of Census data suggests that 37% of children in lesbian
cohabiting households are Black or Hispanic. Among respondents in the NFSS who said their mother had a same-sex rela-
tionship, 43% are Black or Hispanic. In the NLLFS, by contrast, only 6% are Black or Hispanic.
This is an important oversight: demographic indicators of where gay parents live today point less toward stereotypic
places like New York and San Francisco and increasingly toward locales where families are more numerous and overall fer-
tility is higher, like San Antonio and Memphis. In their comprehensive demographic look at the American gay and lesbian
population, Gates and Ost (2004, p. 47) report, States and large metropolitan areas with relatively low concentrations of
gay and lesbian couples in the population tend to be areas where same-sex couples are more likely to have children in
the household. A recent updated brief by Gates (2011, p. F3) reinforces this: Geographically, same-sex couples are most
likely to have children in many of the most socially conservative parts of the country. Moreover, Gates notes that racial
minorities are disproportionately more likely (among same-sex households) to report having children; whites, on the other
hand, are disproportionately less likely to have children. The NFSS sample reveals the same. Gates Census-based assess-
ments further raise questions about the sampling strategies ofand the popular use of conclusions fromstudies based en-
tirely on convenience samples derived from parents living in progressive metropolitan locales.
2.4. The structure and experience of respondents families of origin
The NFSS sought to provide as clear a vision as possible of the respondents household composition during their childhood
and adolescence. The survey asked respondents about the marital status of their biological parents both in the past and pres-
ent. The NFSS also collected calendar data from each respondent about their relationship to people who lived with them in
their household (for more than 4 months) from birth to age 18, as well as who has lived with them from age 18after they
have left hometo the present. While the calendar data is utilized only sparingly in this study, such rich data enables
researchers to document who else has lived with the respondent for virtually their entire life up to the present.
For this particular study, I compare outcomes across eight different types of family-of-origin structure and/or experience.
They were constructed from the answers to several questions both in the screener survey and the full survey. It should be
noted, however, that their construction reects an unusual combination of intereststhe same-sex romantic behavior of par-
ents, and the experience of household stability or disruption. The eight groups or household settings (with an acronym or
short descriptive title) evaluated here, followed by their maximum unweighted analytic sample size, are:
1. IBF: Lived in intact biological family (with mother and father) from 0 to 18, and parents are still married at present
(N = 919).
2. LM: R reported Rs mother had a same-sex romantic (lesbian) relationship with a woman, regardless of any other
household transitions (N = 163).
3. GF: R reported Rs father had a same-sex romantic (gay) relationship with a man, regardless of any other household
transitions (N = 73).
000988
4. Adopted: R was adopted by one or two strangers at birth or before age 2 (N = 101).
5. Divorced later or had joint custody: R reported living with biological mother and father from birth to age 18, but par-
ents are not married at present (N = 116).
6. Stepfamily: Biological parents were either never married or else divorced, and Rs primary custodial parent was mar-
ried to someone else before R turned 18 (N = 394).
7. Single parent: Biological parents were either never married or else divorced, and Rs primary custodial parent did not
marry (or remarry) before R turned 18 (N = 816).
8. All others: Includes all other family structure/event combinations, such as respondents with a deceased parent
(N = 406).
Together these eight groups account for the entire NFSS sample. These eight groups are largely, but not entirely, mutually
exclusive in reality. That is, a small minority of respondents might t more than one group. I have, however, forced their
mutual exclusivity here for analytic purposes. For example, a respondent whose mother had a same-sex relationship might
also qualify in Group 5 or Group 7, but in this case my analytical interest is in maximizing the sample size of Groups 2 and 3
so the respondent would be placed in Group 2 (LMs). Since Group 3 (GFs) is the smallest and most difcult to locate ran-
domly in the population, its composition trumped that of others, even LMs. (There were 12 cases of respondents who re-
ported both a mother and a father having a same-sex relationship; all are analyzed here as GFs, after ancillary analyses
revealed comparable exposure to both their mother and father).
Obviously, different grouping decisions may affect the results. The NFSS, which sought to learn a great deal of information
about respondents families of origin, is well-poised to accommodate alternative grouping strategies, including distinguish-
ing those respondents who lived with their lesbian mothers partner for several years (vs. sparingly or not at all), or early in
their childhood (compared to later). Small sample sizes (and thus reduced statistical power) may nevertheless hinder some
strategies.
In the results section, for maximal ease, I often make use of the acronyms IBF (child of a still-intact biological family), LM
(child of a lesbian mother), and GF (child of a gay father). It is, however, very possible that the same-sex romantic relation-
ships about which the respondents report were not framed by those respondents as indicating their own (or their parents
own) understanding of their parent as gay or lesbian or bisexual in sexual orientation. Indeed, this is more a study of the chil-
dren of parents who have had (and in some cases, are still in) same-sex relationships than it is one of children whose parents
have self-identied or are out as gay or lesbian or bisexual. The particular parental relationships the respondents were
queried about are, however, gay or lesbian in content. For the sake of brevity and to avoid entanglement in interminable
debates about xed or uid orientations, I will regularly refer to these groups as respondents with a gay father or lesbian
mother.
2.5. Outcomes of interest
This study presents an overview of 40 outcome measures available in the NFSS. Table 1 presents summary statistics for all
variables. Why these outcomes? While the survey questionnaire (available online) contains several dozen outcome questions
of interest, I elected to report here an overview of those outcomes, seeking to include common and oft-studied variables of
interest from a variety of different domains. I include all of the particular indexes we sought to evaluate, and a broad list of
outcomes from the emotional, relational, and social domains. Subsequent analyses of the NFSS will no doubt examine other
outcomes, as well as examine the same outcomes in different ways.
The dichotomous outcome variables summarized in Table 1 are the following: relationship status, employment status,
whether they voted in the last presidential election, and use of public assistance (both currently and while growing up),
the latter of which was asked as Before you were 18 years old, did anyone in your immediate family (that is, in your house-
hold) ever receive public assistance (such as welfare payments, food stamps, Medicaid, WIC, or free lunch)? Respondents
were also asked about whether they had ever seriously thought about committing suicide in the past 12 months, and about
their utilization of counseling or psychotherapy for treatment of any problem connected with anxiety, depression, relation-
ships, etc.
The Kinsey scale of sexual behavior was employed, but modied to allow respondents to select the best description of
their sexual orientation (rather than behavior). Respondents were asked to choose the description that best ts how they
think about themselves: 100% heterosexual, mostly heterosexual but somewhat attracted to people of your own sex, bisex-
ual (that is, attracted to men and women equally), mostly homosexual but somewhat attracted to people of the opposite sex,
100% homosexual, or not sexually attracted to either males or females. For simplicity of presentation, I create a dichotomous
measure indicating 100% heterosexual (vs. anything else). Additionally, unmarried respondents who are currently in a rela-
tionship were asked if their romantic partner is a man or a woman, allowing construction of a measure of currently in a
same-sex romantic relationship.
All respondents were asked if a parent or other adult caregiver ever touched you in a sexual way, forced you to touch him
or her in a sexual way, or forced you to have sexual relations? Possible answers were: no, never; yes, once; yes, more than
once; or not sure. A broader measure about forced sex was asked before it, and read as follows: Have you ever been phys-
ically forced to have any type of sexual activity against your will? It employs identical possible answers; both have been
dichotomized for the analyses (respondents who were not sure were not included). Respondents were also asked if they
000989
Table 1
Weighted summary statistics of measures, NFSS.
NFSS variables Range Mean SD N
Currently married 0, 1 0.41 0.49 2988
Currently cohabiting 0, 1 0.15 0.36 2988
Family received welfare growing up 0, 1 0.34 0.47 2669
Currently on public assistance 0, 1 0.21 0.41 2952
Currently employed full-time 0,1 0.45 0.50 2988
Currently unemployed 0, 1 0.12 0.32 2988
Voted in last presidential election 0, 1 0.55 0.50 2960
Bullied while growing up 0, 1 0.36 0.48 2961
Ever suicidal during past year 0, 1 0.07 0.25 2953
Recently or currently in therapy 0, 1 0.11 0.32 2934
Identies as entirely heterosexual 0, 1 0.85 0.36 2946
Is in a same-sex romantic relationship 0, 1 0.06 0.23 1056
Had affair while married/cohabiting 0, 1 0.19 0.39 1869
Has ever had an STI 0, 1 0.11 0.32 2911
Ever touched sexually by parent/adult 0, 1 0.07 0.26 2877
Ever forced to have sex against will 0, 1 0.13 0.33 2874
Educational attainment 15 2.86 1.11 2988
Family-of-origin safety/security 15 3.81 0.97 2917
Family-of-origin negative impact 15 2.58 0.98 2919
Closeness to biological mother 15 4.05 0.87 2249
Closeness to biological father 15 3.74 0.98 1346
Self-reported physical health 15 3.57 0.94 2964
Self-reported overall happiness 15 4.00 1.05 2957
CES-D depression index 14 1.89 0.62 2815
Attachment scale (depend) 15 2.97 0.84 2848
Attachment scale (anxiety) 15 2.51 0.77 2830
Impulsivity scale 14 1.88 0.59 2861
Level of household income 113 7.42 3.17 2635
Current relationship quality index 15 3.98 0.98 2218
Current relationship is in trouble 14 2.19 0.96 2274
Frequency of marijuana use 16 1.50 1.23 2918
Frequency of alcohol use 16 2.61 1.36 2922
Frequency of drinking to get drunk 16 1.70 1.09 2922
Frequency of smoking 16 2.03 1.85 2922
Frequency of watching TV 16 3.15 1.60 2919
Frequency of having been arrested 14 1.29 0.63 2951
Frequency pled guilty to non-minor offense 14 1.16 0.46 2947
N of female sex partners (among women) 011 0.40 1.10 1975
N of female sex partners (among men) 011 3.16 2.68 937
N of male sex partners (among women) 011 3.50 2.52 1951
N of male sex partners (among men) 011 0.40 1.60 944
Age 1839 28.21 6.37 2988
Female 0, 1 0.51 0.50 2988
White 0, 1 0.57 0.49 2988
Gay-friendliness of state of residence 15 2.58 1.78 2988
Family-of-origin structure groups
Intact biological family (IBF) 0, 1 0.40 0.49 2988
Mother had same-sex relationship (LM) 0, 1 0.01 0.10 2988
Father had same-sex relationship (GF) 0, 1 0.01 0.75 2988
Adopted age 02 0, 1 0.01 0.75 2988
Divorced later/joint custody 0, 1 0.06 0.23 2988
Stepfamily 0, 1 0.17 0.38 2988
Single parent 0, 1 0.19 0.40 2988
All others 0, 1 0.15 0.36 2988
Mothers education
Less than high school 0, 1 0.15 0.35 2988
Received high school diploma 0, 1 0.28 0.45 2988
Some college/associates degree 0, 1 0.26 0.44 2988
Bachelors degrees 0, 1 0.15 0.36 2988
More than bachelors 0, 1 0.08 0.28 2988
Do not know/missing 0, 1 0.08 0.28 2988
Family-of-origin income
$020,000 0, 1 0.13 0.34 2988
$20,00140,000 0, 1 0.19 0.39 2988
$40,00175,000 0, 1 0.25 0.43 2988
$75,001100,000 0, 1 0.14 0.34 2988
$100,001150,000 0, 1 0.05 0.22 2988
(continued on next page)
000990
had ever had a sexually-transmitted infection, and if they had ever had a sexual relationship with someone else while they
(the respondent) were married or cohabiting.
Among continuous variables, I included a ve-category educational achievement measure, a standard ve-point self-
reported measure of general physical health, a ve-point measure of overall happiness, a 13-category measure of total
household income before taxes and deductions last year, and a four-point (frequency) measure of how often the respondent
thought their current relationship might be in trouble (never once, once or twice, several times, or numerous times).
Several continuous variables were constructed from multiple measures, including an eight-measure modied version of
the CES-D depression scale, an index of the respondents reported current (romantic) relationship quality, closeness to
the respondents biological mother and father, and a pair of attachment scalesone assessing dependability and the other
anxiety. Finally, a pair of indexes captures (1) the overall safety and security in their family while growing up, and (2)
respondents impressions of negative family-of-origin experiences that continue to affect them. These are part of a multidi-
mensional relationship assessment instrument (dubbed RELATE) designed with the perspective that aspects of family life,
such as the quality of the parents relationship with their children, create a family tone that can be mapped on a continuum
from safe/predictable/rewarding to unsafe/chaotic/punishing (Busby et al., 2001). Each of the scales and their component
measures are detailed in Appendix B.
Finally, I evaluate nine count outcomes, seven of which are frequency measures, and the other two counts of gender-spe-
cic sexual partners. Respondents were asked, During the past year, howoften did you. . . watch more than 3 h of television
in a row, use marijuana, smoke, drink alcohol, and drink with the intent to get drunk. Responses (05) ranged from never
to every day or almost every day. Respondents were also asked if they have ever been arrested, and if they had ever been
convicted of or pled guilty to any charges other than a minor trafc violation. Answers to these two ranged from 0 (no, never)
to 3 (yes, numerous times). Two questions about respondents number of sex partners were asked (of both men and women)
in this way: How many different women have you ever had a sexual relationship with? This includes any female you had
sex with, even if it was only once or if you did not know her well. The same question was asked about sexual relationships
with men. Twelve responses were possible: 0, 1, 2, 3, 46, 79, 1015, 1620, 2130, 3150, 5199, and 100+.
2.6. Analytic approach
My analytic strategy is to highlight distinctions between the eight family structure/experience groups on the 40 outcome
variables, both in a bivariate manner (using a simple T-test) and in a multivariate manner using appropriate variable-specic
regression techniqueslogistic, OLS, Poisson, or negative binomialand employing controls for respondents age, race/eth-
nicity, gender, mothers education, and perceived family-of-origin income, an approach comparable to Rosenfelds (2010)
analysis of differences in children making normal progress through school and the overview article highlighting the ndings
of the rst wave of the Add Health study (Resnick et al., 1997). Additionally, I controlled for having been bullied, the measure
for which was asked as follows: While growing up, children and teenagers typically experience negative interactions with
others. We say that someone is bullied when someone else, or a group, says or does nasty and unpleasant things to him or
her. We do not consider it bullying when two people quarrel or ght, however. Do you recall ever being bullied by someone
else, or by a group, such that you still have vivid, negative memories of it?
Finally, survey respondents current state of residence was coded on a scale (15) according to how expansive or restric-
tive its laws are concerning gay marriage and the legal rights of same-sex couples (as of November 2011). Emerging research
suggests state-level political realities about gay rights may discernibly shape the lives of GLB residents (Hatzenbuehler et al.,
2009; Rostosky et al., 2009). This coding scheme was borrowed from a Los Angeles Times effort to map the timeline of state-
level rights secured for gay unions. I modied it from a 10-point to a 5-point scale (Times Research Reporting, 2012). I clas-
sify the respondents current state in one of the following ve ways:
1 = Constitutional amendment banning gay marriage and/or other legal rights.
2 = Legal ban on gay marriage and/or other legal rights.
3 = No specic laws/bans and/or domestic partnerships are legal.
4 = Domestic partnerships with comprehensive protections are legal and/or gay marriages performed elsewhere are
recognized.
5 = Civil unions are legal and/or gay marriage is legal.
Each case in the NFSS sample was assigned a weight based on the sampling design and their probability of being selected,
ensuring a sample that is nationally representative of American adults aged 1839. These sample weights were used in every
Table 1 (continued)
NFSS variables Range Mean SD N
$150,001200,000 0, 1 0.01 0.11 2988
Above $200,000 0, 1 0.01 0.10 2988
Do not know/missing 0, 1 0.22 0.42 2988
000991
statistical procedure displayed herein unless otherwise noted. The regression models exhibited few (N < 15) missing values
on the covariates.
This broad overview approach, appropriate for introducing a new dataset, provides a foundation for future, more focused
analyses of the outcomes I explore here. There are, after all, far more ways to delineate family structure and experiences
and changes thereinthan I have undertaken here. Others will evaluate such groupings differently, and will construct alter-
native approaches of testing for group differences in what is admittedly a wide diversity of outcome measures.
I would be remiss to claim causation here, since to document that having particular family-of-origin experiencesor the
sexual relationships of ones parentscauses outcomes for adult children, I would need to not only document that there is a
correlation between such family-of-origin experiences, but that no other plausible factors could be the common cause of any
suboptimal outcomes. Rather, my analytic intention is far more modest than that: to evaluate the presence of simple group
differences, andwith the addition of several control variablesto assess just how robust such group differences are.
3. Results
3.1. Comparisons with still-intact, biological families (IBFs)
Table 2 displays mean scores on 15 dichotomous outcome variables which can be read as simple percentages, sorted by
the eight different family structure/experience groups described earlier. As in Tables 3 and 4, numbers that appear in bold
indicate that the groups estimate is statistically different from the young-adult children of IBFs, as discerned by a basic
T-test (p < 0.05). Numbers that appear with an asterisk (
) beside it indicate that the groups dichotomous variable estimate

from a logistic regression model (not shown) is statistically-signicantly different from IBFs, after controlling for respon-
dents age, gender, race/ethnicity, level of mothers education, perceived family-of-origins income, experience with having
been bullied as a youth, and the gay friendliness of the respondents current state of residence.
At a glance, the number of statistically-signicant differences between respondents from IBFs and respondents from the
other seven types of family structures/experiences is considerable, and in the vast majority of cases the optimal outcome
where one can be readily discernedfavors IBFs. Table 2 reveals 10 (out of 15 possible) statistically-signicant differences in
simple t-tests between IBFs and LMs (the pool of respondents who reported that their mother has had a lesbian relationship),
one higher than the number of simple differences (9) between IBFs and respondents from both single-parent and stepfam-
ilies. All but one of those associations is signicant in logistic regression analyses contrasting LMs and IBFs (the omitted
category).
Beginning at the top of Table 2, the marriage rates of LMs and GFs (those who reported that their father had a gay rela-
tionship) are statistically comparable to IBFs, while LMs cohabitation rate is notable higher than IBFs (24% vs. 9%, respec-
tively). Sixty-nine (69) percent of LMs and 57% of GFs reported that their family received public assistance at some point
while growing up, compared with 17% of IBFs; 38% of LMs said they are currently receiving some form of public assistance,
compared with 10% of IBFs. Just under half of all IBFs reported being employed full-time at present, compared with 26% of
Table 2
Mean scores on select dichotomous outcome variables, NFSS (can read as percentage: as in, 0.42 = 42%).
IBF (intact
bio family)
LM
(lesbian mother)
GF
(gay father)
Adopted by
strangers
Divorced
late (>18)
Stepfamily Single-
parent
All
other
Currently married 0.43 0.36 0.35 0.41 0.36
0.41 0.37 0.39

Currently cohabiting 0.09 0.24
0.21 0.07
^
0.31
0.19
0.19
0.13
Family received welfare growing up 0.17 0.69
0.57
0.12
^
0.47
^
0.53
^
0.48
^
0.35
^
Currently on public assistance 0.10 0.38
0.23 0.27
0.31
0.30
0.30
0.23
Currently employed full-time 0.49 0.26
0.34 0.41 0.42 0.47

^
0.43
^
0.39
Currently unemployed 0.08 0.28
0.20 0.22
0.15 0.14 0.13

^
0.15
Voted in last presidential election 0.57 0.41 0.73
^
0.58 0.63
^
0.57
^
0.51 0.48
Thought recently about suicide 0.05 0.12 0.24
0.07 0.08 0.10 0.05 0.09

Recently or currently in therapy 0.08 0.19
0.19 0.22
0.12 0.17
0.13
0.09
Identies as entirely heterosexual 0.90 0.61
0.71
0.82
^
0.83
^
0.81
^
0.83
^
0.82
^
Is in a same-sex romantic relationship 0.04 0.07 0.12 0.23 0.05 0.13
0.03 0.02
Had affair while married/cohabiting 0.13 0.40
0.25 0.20 0.12

^
0.32
0.19
^
0.16
^
Has ever had an STI 0.08 0.20
0.25
0.16 0.12 0.16
0.14
0.08
Ever touched sexually by parent/adult 0.02 0.23
0.06
^
0.03
^
0.10
0.12
0.10
0.08
^
Ever forced to have sex against will 0.08 0.31
0.25
0.23
0.24
0.16
0.16
^
0.11
^
Bold indicates the mean scores displayed are statistically-signicantly different from IBFs (currently intact, bio mother/father household, column 1),
without additional controls.
An asterisk (
) next to the estimate indicates a statistically-signicant difference (p < 0.05) between the groups coefcient and that of IBFs, controlling for
respondents age, gender, race/ethnicity, level of mothers education, perceived household income while growing up, experience being bullied as a youth,
and states legislative gay-friendliness, derived from logistic regression models (not shown).
A caret (^) next to the estimate indicates a statistically-signicant difference (p < 0.05) between the groups mean and the mean of LM (column 2), without
additional controls.
000992
LMs. While only 8% of IBF respondents said they were currently unemployed, 28% of LM respondents said the same. LMs
were statistically less likely than IBFs to have voted in the 2008 presidential election (41% vs. 57%), and more than twice
as likely19% vs. 8%to report being currently (or within the past year) in counseling or therapy for a problem connected
with anxiety, depression, relationships, etc., an outcome that was signicantly different after including control variables.
In concurrence with several studies of late, the NFSS reveals that the children of lesbian mothers seem more open to
same-sex relationships (Biblarz and Stacey, 2010; Gartrell et al., 2011a,b; Golombok et al., 1997). Although they are not sta-
tistically different from most other groups in having a same-sex relationship at present, they are much less apt to identify
entirely as heterosexual (61% vs. 90% of respondents from IBFs). The same was true of GF respondentsthose young adults
who said their father had a relationship with another man: 71% of them identied entirely as heterosexual. Other sexual dif-
ferences are notable among LMs, too: a greater share of daughters of lesbian mothers report being not sexually attracted to
either males or females than among any other family-structure groups evaluated here (4.1% of female LMs, compared to
0.5% of female IBFs, not shown in Table 2). Exactly why the young-adult children of lesbian mothers are more apt to expe-
rience same-sex attraction and behaviors, as well as self-report asexuality, is not clear, but the fact that they do seems con-
sistent across studies. Given that lower rates of heterosexuality characterize other family structure/experience types in the
Table 3
Mean scores on select continuous outcome variables, NFSS.
IBF (intact
bio family)
LM (lesbian
mother)
GF (gay
father)
Adopted by
strangers
Divorced
late (>18)
Stepfamily Single- parent All
other
Educational attainment 3.19 2.39
2.64
3.21
^
2.88
^
2.64
2.66
2.54
Family-of-origin safety/security 4.13 3.12
3.25
3.77
^
3.52
3.52
^
3.58
^
3.77
^
Family-of-origin negative impact 2.30 3.13
2.90
2.83
2.96
2.76
2.78
2.64
^
Closeness to biological mother 4.17 4.05 3.71
3.58 3.95 4.03 3.85
3.97
Closeness to biological father 3.87 3.16 3.43 3.29
3.65 3.24
3.61
Self-reported physical health 3.75 3.38 3.58 3.53 3.46 3.49 3.43
3.41
Self-reported overall happiness 4.16 3.89 3.72 3.92 4.02 3.87
3.93 3.83
CES-D depression index 1.83 2.20
2.18
1.95 2.01 1.91

^
1.89
^
1.94
^
Attachment scale (depend) 2.82 3.43
3.14 3.12
3.08
^
3.10
^
3.05
^
3.02
^
Attachment scale (anxiety) 2.46 2.67 2.66 2.66 2.71 2.53 2.51 2.56
Impulsivity scale 1.90 2.03 2.02 1.85 1.94 1.86
^
1.82
^
1.89
Level of household income 8.27 6.08 7.15 7.93
^
7.42
^
7.04 6.96 6.19
Current relationship quality index 4.11 3.83 3.63
3.79 3.95 3.80
3.95 3.94
Current relationship is in trouble 2.04 2.35 2.55
2.35 2.43 2.35
2.26
2.15
An asterisk (
and states legislative gay-friendliness, derived from OLS regression models (not shown).
Table 4
Mean scores on select event-count outcome variables, NFSS.
IBF (intact
bio family)
LM (lesbian
mother)
GF
(gay father)
Adopted by
strangers
Divorced
late (>18)
Stepfamily Single-
parent
All
other
Frequency of marijuana use 1.32 1.84
1.61 1.33
^
2.00
1.47 1.73
1.49
Frequency of alcohol use 2.70 2.37 2.70 2.74 2.55 2.50 2.66 2.44
Frequency of drinking to get drunk 1.68 1.77 2.14 1.73 1.90 1.68 1.74 1.64
Frequency of smoking 1.79 2.76
2.61
2.34
2.44
2.31
2.18
1.91
^
Frequency of watching TV 3.01 3.70
3.49 3.31 3.33 3.43
3.25 2.95
^
Frequency of having been arrested 1.18 1.68
1.75
1.31
^
1.38 1.38
^
1.35
^
1.34
^
Frequency pled guilty to non-minor offense 1.10 1.36
1.41
1.19 1.30 1.21
1.17
^
1.17
^
N of female sex partners (among women) 0.22 1.04
1.47
0.47
^
0.96
0.47
^
0.52
^
0.33
^
N of female sex partners (among men) 2.70 3.46 4.17 3.24 3.66 3.85
3.23 3.37
N of male sex partners (among women) 2.79 4.02
5.92
3.49 3.97
4.57
4.04
2.91
^
N of male sex partners (among men) 0.20 1.48
1.47
0.27 0.98
0.55 0.42 0.44

An asterisk (
and states legislative gay-friendliness, derived from Poisson or negative binomial regression models (not shown).
000993
NFSS, as Table 2 clearly documents, the answer is likely located not simply in parental sexual orientation but in successful
cross-sex relationship role modeling, or its absence or scarcity.
Sexual conduct within their romantic relationships is also distinctive: while 13% of IBFs reported having had a sexual rela-
tionship with someone else while they were either married or cohabiting, 40% of LMs said the same. In contrast to Gartrell
et al.s (2011a,b) recent, widely-disseminated conclusions about the absence of sexual victimization in the NLLFS data, 23% of
LMs said yes when asked whether a parent or other adult caregiver ever touched you in a sexual way, forced you to touch
him or her in a sexual way, or forced you to have sexual relations, while only 2% of IBFs responded afrmatively. Since such
reports are more common among women than men, I split the analyses by gender (not shown). Among female respondents,
3% of IBFs reported parental (or adult caregiver) sexual contact/victimization, dramatically below the 31% of LMs who re-
ported the same. Just under 10% of female GFs responded afrmatively to the question, an estimate not signicantly different
from the IBFs.
It is entirely plausible, however, that sexual victimization could have been at the hands of the LM respondents biological
father, prompting the mother to leave the union andat some point in the futurecommence a same-sex relationship. Ancil-
lary (unweighted) analyses of the NFSS, which asked respondents how old they were when the rst incident occurred (and
can be compared to the household structure calendar, which documents who lived in their household each year up until age
18) reveal this possibility, up to a point: 33% of those LM respondents who said they had been sexually victimized by a parent
or adult caregiver reported that they were also living with their biological father in the year that the rst incident occurred.
Another 29% of victimized LMs reported never having lived with their biological father at all. Just under 34% of LM respon-
dents who said they had at some point lived with their mothers same-sex partner reported a rst-time incident at an age
that was equal to or higher than when they rst lived with their mothers partner. Approximately 13% of victimized LMs
reported living with a foster parent the year when the rst incident occurred. In other words, there is no obvious trend
to the timing of rst victimization and when the respondent may have lived with their biological father or their mothers
same-sex partner, nor are we suggesting by whom the respondent was most likely victimized. Future exploration of the
NFSSs detailed household structure calendar offers some possibility for clarication.
The elevated LM estimate of sexual victimization is not the only estimate of increased victimization. Another more gen-
eral question about forced sex, Have you ever been physically forced to have any type of sexual activity against your will
also displays signicant differences between IBFs and LMs (and GFs). The question about forced sex was asked before the
question about sexual contact with a parent or other adult and may include incidents of it but, by the numbers, clearly in-
cludes additional circumstances. Thirty-one percent of LMs indicated they had, at some point in their life, been forced to
have sex against their will, compared with 8% of IBFs and 25% of GFs. Among female respondents, 14% of IBFs reported forced
sex, compared with 46% of LMs and 52% of GFs (both of the latter estimates are statistically-signicantly different from that
reported by IBFs).
While I have so far noted several distinctions between IBFs and GFsrespondents who said their father had a gay rela-
tionshipthere are simply fewer statistically-signicant distinctions to note between IBFs and GFs than between IBFs and
LMs, which may or may not be due in part to the smaller sample of respondents with gay fathers in the NFSS, and the much
smaller likelihood of having lived with their gay father while he was in a same-sex relationship. Only six of 15 measures in
Table 2 reveal statistically-signicant differences in the regression models (but only one in a bivariate environment). After
including controls, the children of a gay father were statistically more apt (than IBFs) to receive public assistance while grow-
ing up, to have voted in the last election, to have thought recently about committing suicide, to ever report a sexually-trans-
mitted infection, have experienced forced sex, and were less likely to self-identify as entirely heterosexual. While other
outcomes reported by GFs often differed from IBFs, statistically-signicant differences were not as regularly detected.
Although my attention has been primarily directed at the inter-group differences between IBFs, LMs, and GFs, it is worth
noting that LMs are hardly alone in displaying numerous differences with IBFs. Respondents who lived in stepfamilies or sin-
gle-parent families displayed nine simple differences in Table 2. Besides GFs, adopted respondents displayed the fewest sim-
ple differences (three).
Table 3 displays mean scores on 14 continuous outcomes. As in Table 2, bold indicates simple statistically-signicant out-
come differences with young-adult respondents from still-intact, biological families (IBFs) and an asterisk indicates a regres-
sion coefcient (models not shown) that is signicantly different from IBFs after a series of controls. Consistent with Table 2,
eight of the estimates for LMs are statistically different from IBFs. Five of the eight differences are signicant as regression
estimates. The young-adult children of women who have had a lesbian relationship fare worse on educational attainment,
family-of-origin safety/security, negative impact of family-of-origin, the CES-D (depression) index, one of two attachment
scales, report worse physical health, smaller household incomes than do respondents from still-intact biological families,
and think that their current romantic relationship is in trouble more frequently.
The young-adult GF respondents were likewise statistically distinct from IBF respondents on seven of 14 continuous out-
comes, all of which were signicantly different when evaluated in regression models. When contrasted with IBFs, GFs re-
ported more modest educational attainment, worse scores on the family-of-origin safety/security and negative impact
indexes, less closeness to their biological mother, greater depression, a lower score on the current (romantic) relationship
quality index, and think their current romantic relationship is in trouble more frequently.
As in Table 2, respondents who reported living in stepfamilies or in single-parent households also exhibit numerous sim-
ple statistical differences from IBFson nine and 10 out of 14 outcomes, respectivelymost of which remain signicant in
000994
the regression models. On only four of 14 outcomes do adopted respondents appear distinctive (three of which remain sig-
nicant after introducing controls).
Table 4 displays mean scores on nine event counts, sorted by the eight family structure/experience groups. The NFSS
asked all respondents about experience with male and female sexual partners, but I report them here separately by gender.
LM respondents report statistically greater marijuana use, more frequent smoking, watch television more often, have been
arrested more, pled guilty to non-minor offenses more, andamong womenreport greater numbers of both female and
male sex partners than do IBF respondents. Female LMs reported an average of just over one female sex partner in their life-
times, as well as four male sex partners, in contrast to female IBFs (0.22 and 2.79, respectively). Male LMs report an average
of 3.46 female sex partners and 1.48 male partners, compared with 2.70 and 0.20, respectively, among male IBFs. Only the
number of male partners among men, however, displays signicant differences (after controls are included).
Among GFs, only three bivariate distinctions appear. However, six distinctions emerge after regression controls: they are
more apt than IBFs to smoke, have been arrested, pled guilty to non-minor offenses, and report more numerous sex partners
(except for the number of female sex partners among male GFs). Adopted respondents display no simple differences from
IBFs, while the children of stepfamilies and single parents each display six signicant differences with young adults from
still-intact, biological mother/father families.
Although I have paid much less attention to most of the other groups whose estimates also appear in Tables 24, it is
worth noting how seldom the estimates of young-adult children who were adopted by strangers (before age 2) differ statis-
tically from the children of still-intact biological families. They display the fewest simple signicant differencesseven
across the 40 outcomes evaluated here. Given that such adoptions are typically the result of considerable self-selection, it
should not surprise that they display fewer differences with IBFs.
To summarize, then, in 25 of 40 outcomes, there are simple statistically-signicant differences between IBFs and LMs,
those whose mothers had a same-sex relationship. After controls, there are 24 such differences. There are 24 simple differ-
ences between IBFs and stepfamilies, and 24 statistically-signicant differences after controls. Among single (heterosexual)
parents, there are 25 simple differences before controls and 21 after controls. Between GFs and IBFs, there are 11 and 19 such
differences, respectively.
3.2. Summary of differences between LMs and other family structures/experiences
Researchers sometimes elect to evaluate the outcomes of children of gay and lesbian parents by comparing them not di-
rectly to stable heterosexual marriages but to other types of households, since it is often the caseand it is certainly true of
the NFSSthat a gay or lesbian parent rst formed a heterosexual union prior to coming out of the closet, and witnessing
the dissolution of that union (Tasker, 2005). So comparing the children of such parents with those who experienced no union
dissolution is arguably unfair. The NFSS, however, enables researchers to compare outcomes across a variety of other types of
family-structural history. While I will not explore in-depth here all the statistically-signicant differences between LMs, GFs,
and other groups besides IBFs, a few overall observations are merited.
Of the 239 possible between-group differences herenot counting those differences with Group 1 (IBFs) already de-
scribed earlierthe young-adult children of lesbian mothers display 57 (or 24% of total possible) that are signicant at
the p < 0.05 level (indicated in Tables 24 with a caret), and 44 (or 18% of total) that are signicant after controls (not
shown). The majority of these differences are in suboptimal directions, meaning that LMs display worse outcomes. The
young-adult children of gay men, on the other hand, display only 11 (or 5% of total possible) between-group differences
that are statistically signicant at the p < 0.05 level, and yet 24 (or 10% of total) that are signicant after controls (not
shown).
In the NFSS, then, the young-adult children of a mother who has had a lesbian relationship display more signicant
distinctions with other respondents than do the children of a gay father. This may be the result of genuinely different
experiences of their family transitions, the smaller sample size of children of gay men, or the comparatively-rarer expe-
rience of living with a gay father (only 42% of such respondents reported ever living with their father while he was in a
same-sex relationship, compared with 91% who reported living with their mother while she was in a same-sex
relationship).
4. Discussion
Just how different are the adult children of men and women who pursue same-sex romantic (i.e., gay and lesbian)
relationships, when evaluated using population-based estimates from a random sample? The answer, as might be expected,
depends on to whom you compare them. When compared with children who grew up in biologically (still) intact, mother
father families, the children of women who reported a same-sex relationship look markedly different on numerous out-
comes, including many that are obviously suboptimal (such as education, depression, employment status, or marijuana
use). On 25 of 40 outcomes (or 63%) evaluated here, there are bivariate statistically-signicant (p < 0.05) differences between
children from still-intact, mother/father families and those whose mother reported a lesbian relationship. On 11 of 40 out-
comes (or 28%) evaluated here, there are bivariate statistically-signicant (p < 0.05) differences between children from
still-intact, mother/father families and those whose father reported a gay relationship. Hence, there are differences in both
000995
comparisons, but there are many more differences by any method of analysis in comparisons between young-adult children
of IBFs and LMs than between IBFs and GFs.
While the NFSS may best capture what might be called an earlier generation of children of same-sex parents, and in-
cludes among them many who witnessed a failed heterosexual union, the basic statistical comparisons between this group
and those of others, especially biologically-intact, mother/father families, suggests that notable differences on many out-
comes do in fact exist. This is inconsistent with claims of no differences generated by studies that have commonly em-
ployed far more narrow samples than this one.
Goldberg (2010) aptly asserts that many existing studies were conducted primarily comparing children of heterosexual
divorced and lesbian divorced mothers, potentially leading observers to erroneously attribute to parental sexual orientation
the corrosive effects of enduring parental divorce. Her warning is well-taken, and it is one that the NFSS cannot entirely
mitigate. Yet when compared with other young adults who experienced household transitions and who witnessed parents
forming new romantic relationshipsfor example, stepfamiliesthe children of lesbian mothers looked (statistically) signif-
icantly different just under 25% of the time (and typically in suboptimal directions). Nevertheless, the children of mothers
who have had same-sex relationships are far less apt to differ from stepfamilies and single parents than they are from
still-intact biological families.
Why the divergence between the ndings in this study and those from so many previous ones? The answer lies in part
with the small or nonprobability samples so often relied upon in nearly all previous studiesthey have very likely underes-
timated the number and magnitude of real differences between the children of lesbian mothers (and to a lesser extent, gay
fathers) and those raised in other types of households. While the architects of such studies have commonly and appropri-
ately acknowledged their limitations, practicallysince they are often the only studies being conductedtheir results are
treated as providing information about gay and lesbian household experiences in general. But this study, based on a rare large
probability sample, reveals far greater diversity in the experience of lesbian motherhood (and to a lesser extent, gay father-
hood) than has been acknowledged or understood.
Given that the characteristics of the NFSSs sample of children of LMs and GFs are close to estimates of the same offered by
demographers using the American Community Study, one conclusion from the analyses herein is merited: the sample-selec-
tion bias problem in very many studies of gay and lesbian parenting is not incidental, but likely profound, rendering the abil-
ity of much past research to offer valid interpretations of average household experiences of children with a lesbian or gay
parent suspect at best. Most snowball-sample-based research has, instead, shed light on above-average household
experiences.
While studies of family structure often locate at least modest benets that accrue to the children of married biological
parents, some scholars attribute much of the benet to socioeconomic-status differences between married parents and those
parents in other types of relationships (Biblarz and Raftery, 1999). While this is likely true of the NFSS as well, the results
presented herein controlled not only for socioeconomic status differences between families of origin, but also political-geo-
graphic distinctions, age, gender, race/ethnicity, and the experience of having been bullied (which was reported by 53% of
LMs but only 35% of IBFs).
To be sure, those NFSS respondents who reported that a parent of theirs had had a romantic relationship with a member
of the same sex are a very diverse group: some experienced numerous household transitions, and some did not. Some of their
parents may have remained in a same-sex relationship, while others did not. Some may self-identify as lesbian or gay, while
others may not. I did not explore in detail the diversity of household experiences here, given the overview nature of this
study. But the richness of the NFSSwhich has annual calendar data for household transitions from birth to age 18 and from
age 18 to the presentallows for closer examination of many of these questions.
Nevertheless, to claim that there are few meaningful statistical differences between the different groups evaluated here
would be to state something that is empirically inaccurate. Minimally, the population-based estimates presented here sug-
gest that a good deal more attention must be paid to the real diversity among gay and lesbian parent experiences in America,
just as it long has been among heterosexual households. Child outcomes in stable, planned GLB families and those that are
the product of previous heterosexual unions are quite likely distinctive, as previous studies conclusions would suggest. Yet
as demographers of gay and lesbian America continue to noteand as the NFSS reinforcesplanned GLB households only
comprise a portion (and an unknown one at that) of all GLB households with children.
Even if the children in planned GLB families exhibit better outcomes than those from failed heterosexual unions, the for-
mer still exhibits a diminished context of kin altruism (like adoption, step-parenting, or nonmarital childbirth), which have
typically proven to be a risk setting, on average, for raising children when compared with married, biological parenting (Mill-
er et al., 2000). In short, if same-sex parents are able to raise children with no differences, despite the kin distinctions, it
would mean that same-sex couples are able to do something that heterosexual couples in step-parenting, adoptive, and
cohabiting contexts have themselves not been able to doreplicate the optimal childrearing environment of married, bio-
logical-parent homes (Moore et al., 2002). And studies focusing on parental roles or household divisions of labor in planned
GLB families will fail to revealbecause they have not measured ithow their children fare as adults.
The between-group comparisons described above also suggest that those respondents with a lesbian mother and those
with a gay father do not always exhibit comparable outcomes in young adulthood. While the sample size of gay fathers
in the NFSS was modest, any monolithic ideas about same-sex parenting experiences in general are not supported by these
analyses.
000996
Although the NFSS offers strong support for the notion that there are signicant differences among young adults that cor-
respond closely to the parental behavior, family structures, and household experiences during their youth, I have not and will
not speculate here on causality, in part because the data are not optimally designed to do so, and because the causal
reckoning for so many different types of outcomes is well beyond what an overview manuscript like this one could ever pur-
port to accomplish. Focused (and more complex) analyses of unique outcomes, drawing upon idiosyncratic, domain-specic
conceptual models, is recommended for scholars who wish to more closely assess the functions that the number, gender, and
sexual decision-making of parents may play in young adults lives. I am thus not suggesting that growing up with a lesbian
mother or gay father causes suboptimal outcomes because of the sexual orientation or sexual behavior of the parent; rather,
my point is more modest: the groups display numerous, notable distinctions, especially when compared with young adults
whose biological mother and father remain married.
There is more that this article does not accomplish, including closer examinations of subpopulations, consideration
of more outcomes and comparisons between other groups, and stronger tests of statistical signicancesuch as multiple
regression with more numerous independent variables, or propensity score matching. That is what the NFSS is designed
to foster. This article serves as a call for such study, as well as an introduction to the data and to its sampling and measure-
ment strengths and abilities. Future studies would optimally include a more signicant share of children from planned gay
families, although their relative scarcity in the NFSS suggests that their appearance in even much larger probability samples
will remain infrequent for the foreseeable future. The NFSS, despite signicant efforts to randomly over-sample such popu-
lations, nevertheless was more apt to survey children whose parents exhibited gay and lesbian relationship behavior after
being in a heterosexual union. This pattern may remain more common today than many scholars suppose.
5. Conclusion
As scholars of same-sex parenting aptly note, same-sex couples have and will continue to raise children. American courts
are nding arguments against gay marriage decreasingly persuasive (Rosenfeld, 2007). This study is intended to neither
undermine nor afrm any legal rights concerning such. The tenor of the last 10 years of academic discourse about gay
and lesbian parents suggests that there is little to nothing about them that might be negatively associated with child devel-
opment, and a variety of things that might be uniquely positive. The results of analyzing a rare large probability sample re-
ported herein, however, document numerous, consistent differences among young adults who reported maternal lesbian
behavior (and to a lesser extent, paternal gay behavior) prior to age 18. While previous studies suggest that children in
planned GLB families seem to fare comparatively well, their actual representativeness among all GLB families in the US
may be more modest than research based on convenience samples has presumed.
Although the ndings reported herein may be explicable in part by a variety of forces uniquely problematic for child
development in lesbian and gay familiesincluding a lack of social support for parents, stress exposure resulting from per-
sistent stigma, and modest or absent legal security for their parental and romantic relationship statusesthe empirical claim
that no notable differences exist must go. While it is certainly accurate to afrm that sexual orientation or parental sexual
behavior need have nothing to do with the ability to be a good, effective parent, the data evaluated herein using population-
based estimates drawn froma large, nationally-representative sample of young Americans suggest that it may affect the real-
ity of family experiences among a signicant number.
Do children need a married mother and father to turn out well as adults? No, if we observe the many anecdotal accounts
with which all Americans are familiar. Moreover, there are many cases in the NFSS where respondents have proven resilient
and prevailed as adults in spite of numerous transitions, be they death, divorce, additional or diverse romantic partners, or
remarriage. But the NFSS also clearly reveals that children appear most apt to succeed well as adultson multiple counts and
across a variety of domainswhen they spend their entire childhood with their married mother and father, and especially
when the parents remain married to the present day. Insofar as the share of intact, biological mother/father families contin-
ues to shrink in the United States, as it has, this portends growing challenges within families, but also heightened depen-
dence on public health organizations, federal and state public assistance, psychotherapeutic resources, substance use
programs, and the criminal justice system.
Appendix A. Comparison of weighted NFSS results with parallel national survey results on selected demographic and
lifestyle variables, US adults (in percentages)
NFSS 2011,
N = 941
(1823)
NSYR
20072008,
N = 2520
(1823)
NFSS 2011,
N = 1123
(2432)
Add Health
20072008,
N = 15,701
(2432)
NFSS 2011,
N = 2988
(1839)
NSFG
20062010,
N = 16,851
(1839)
CPS ASEC
2011,
N = 58,788
(1839)
Gender
Male 52.6 48.3 47.3 50.6 49.4 49.8 50.4
Female 47.4 51.7 52.8 49.4 50.6 50.2 49.6
000997
Appendix A (continued)
NFSS 2011,
N = 941
(1823)
NSYR
20072008,
N = 2520
(1823)
NFSS 2011,
N = 1123
(2432)
Add Health
20072008,
N = 15,701
(2432)
NFSS 2011,
N = 2988
(1839)
NSFG
20062010,
N = 16,851
(1839)
CPS ASEC
2011,
N = 58,788
(1839)
Age
1823 28.9 28.6 28.2
2432 41.2 40.6 42.1
3339 29.9 30.9 29.8
Race/ethnicity
White, NH 54.2 68.3 60.2 69.2 57.7 61.6 59.6
Black, NH 11.0 15.0 13.0 15.9 12.6 13.3 13.2
Hispanic 24.9 11.2 20.7 10.8 20.8 18.6 19.5
Other (or multiple),
NH
10.0 5.5 6.2 4.2 8.9 6.5 7.8
Region
Northeast 18.9 11.8 16.5 17.6 17.5
Midwest 18.7 25.6 23.3 21.1 21.2
South 34.3 39.1 39.6 36.7 37.0
West 28.2 23.5 20.6 24.6 24.4
Mothers education
(BA or above)
28.4 33.3 24.6 21.9 25.3 22.2
Respondents education
(BA or above)
5.3 3.8 33.7 30.0 26.5 24.2
Household income
(current)
Under $10,000 21.0 9.7 5.6 11.9 9.5 5.7
$10,00019,999 13.3 9.1 6.9 9.2 13.1 7.4
$20,00029,999 11.6 10.3 10.1 10.5 13.5 9.5
$30,00039,999 8.0 11.0 11.1 9.6 13.4 9.4
$40,00049,999 6.5 12.8 11.8 9.9 8.5 9.1
$50,00074,999 14.9 22.3 24.3 19.2 19.5 20.3
$75,000 or more 24.7 24.9 30.2 29.8 22.7 38.6
Ever had sex 66.5 75.6 90.6 93.9 85.6 91.2
Never been married 89.3 92.8 45.7 50.0 51.7 52.3 54.4
Currently married 8.0 6.9 44.9 44.6 40.6 39.2 37.9
Church attendance
Once a week or more 18.4 20.2 22.1 16.0 22.3 26.2
Never 32.3 35.6 31.2 32.1 31.7 25.8
Not religious 21.1 24.7 22.5 20.2 22.0 21.7
Self-reported health
Poor 1.8 1.5 1.0 1.2 1.5 0.7
Fair 8.4 9.2 11.0 7.9 10.7 5.3
Good 28.7 26.7 37.6 33.5 33.9 24.9
Very Good 39.6 37.5 35.7 38.2 37.3 40.9
Excellent 21.5 25.2 14.8 19.1 16.7 28.3
Never drinks alcohol 30.5 21.9 22.4 26.1 25.4 18.7
000998
Appendix B. Construction of outcome indexes
B.1. CES-D (depression) index (8 items, a = 0.87)
Respondents were asked to think about the past 7 days, and assess howoften each of the following things were true about
them. Answer categories ranged from never or rarely (0) to most of the time or all of the time (3). Some items were re-
verse-coded for the index variable (e.g., You felt happy.):
1. You were bothered by things that usually do not bother you.
2. You could not shake off the blues, even with help from your family and your friends.
3. You felt you were just as good as other people.
4. You had trouble keeping your mind on what you were doing.
5. You felt depressed.
6. You felt happy.
7. You enjoyed life.
8. You felt sad.
B.2. Current romantic relationship quality (6 items, a = 0.96)
Respondents were asked to assess their current romantic relationship. Answer categories ranged from strongly disagree
(1) to strongly agree (5):
1. We have a good relationship.
2. My relationship with my partner is very healthy.
3. Our relationship is strong.
4. My relationship with my partner makes me happy.
5. I really feel like part of a team with my partner.
6. Our relationship is pretty much perfect.
B.3. Family-of-origin relationship safety/security (4 items, a = 0.90)
Respondents were asked to evaluate the overall atmosphere in their family while growing up by responding to four state-
ments whose answer categories ranged from strongly disagree (1) to strongly agree (5):
1. My family relationships were safe, secure, and a source of comfort.
2. We had a loving atmosphere in our family.
3. All things considered, my childhood years were happy.
4. My family relationships were confusing, inconsistent, and unpredictable.
B.4. Family-of-origin negative impact (3 items, a = 0.74)
Respondents were asked to evaluate the present-day impact of their family-of-origin experiences by responding to three
statements whose answer categories ranged from strongly disagree (1) to strongly agree (5):
1. There are matters from my family experience that I am still having trouble dealing with or coming to terms with.
2. There are matters from my family experience that negatively affect my ability to form close relationships.
3. I feel at peace about anything negative that happened to me in the family in which I grew up.
B.5. Impulsivity (4 items, a = 0.76)
Respondents were asked to respond to four statements about their decision-making, especially as it concerns risk-taking
and new experiences. Answer categories ranged from 1 (never or rarely) to 4 (most or all of the time):
1. When making a decision, I go with my gut feeling and do not think much about the consequences of each
alternative.
2. I like new and exciting experiences, even if I have to break the rules.
3. I am an impulsive person.
4. I like to take risks.
000999
B.6. Closeness to biological mother and father (6 items, a = 0.89 and 0.92)
Respondents were asked to evaluate their current relationship with up to four parent gureswho they reported living
with for at least 3 years when they were 018 years oldby reporting the frequency of six parentchild interactions. For each
parent gure, these six items were coded and summed into a parental closeness index. From these, I derived indices of close-
ness to the respondents biological mother and biological father. Response categories ranged from never (1) to always (5):
1. How often do you talk openly with your parent about things that are important to you?
2. How often does your parent really listen to you when you want to talk?
3. How often does your parent explicitly express affection or love for you?
4. Would your parent help you if you had a problem?
5. If you needed money, would you ask your parent for it?
6. How often is your parent interested in the things you do?
B.7. Attachment (depend, 6 items, a = 0.80; anxiety, 6 items, a = 0.82)
For a pair of attachment measures, respondents were asked to rate their general feelings about romantic relationships,
both past and present, in response to 12 items. Response categories ranged from not at all characteristic of me (1) to very
characteristic of me (5). Items 16 were coded and summed into a depend scale, with higher scores denoting greater com-
fort with depending upon others. Items 712 were coded and summed into an anxiety scale, with higher scores denoting
greater anxiety in close relationships, in keeping with the original Adult Attachment Scale developed by Collins and Read
(1990). The measures employed were:
1. I nd it difcult to allow myself to depend on others.
2. I am comfortable depending on others.
3. I nd that people are never there when you need them.
4. I know that people will be there when I need them.
5. I nd it difcult to trust others completely.
6. I am not sure that I can always depend on others to be there when I need them.
7. I do not worry about being abandoned.
8. In relationships, I often worry that my partner does not really love me.
9. I nd that others are reluctant to get as close as I would like.
10. In relationships, I often worry that my partner will not want to stay with me.
11. I want to merge completely with another person.
12. My desire to merge sometimes scares people away.
References
Anderssen, Norman, Amlie, Christine, Erling, Ytteroy A., 2002. Outcomes for children with lesbian or gay parents. A review of studies from 1978 to 2000.
Scandinavian Journal of Psychology 43 (4), 335351.
Balsam, Kimberly F., Beauchaine, Theodore P., Rothblum, Esther D., Solomon, Sondra E., 2008. Three-year follow-up of same-sex couples who had civil
unions in Vermont, same-sex couples not in civil unions, and heterosexual married couples. Developmental Psychology 44, 102116.
Baumle, Amanda K., Compton, DLane R., Poston Jr., Dudley L., 2009. Same-Sex Partners: The Demography of Sexual Orientation. SUNY Press, Albany, NY.
Berg, Sven, 1988. Snowball sampling. In: Kotz, Samuel, Johnson, Norman L. (Eds.), Encyclopedia of Statistical Sciences, vol. 8. Wiley-Interscience, New York.
Biblarz, Timothy J., Raftery, Adrian E., 1999. Family structure, educational attainment, and socioeconomic success: rethinking the pathology of matriarchy.
American Journal of Sociology 105, 321365.
Biblarz, Timothy J., Stacey, Judith, 2010. How does the gender of parents matter? Journal of Marriage and Family 72 (1), 322.
Bos, Henny M.W., Sandfort, Theo G.M., 2010. Childrens gender identity in lesbian and heterosexual two-parent families. Sex Roles 62, 114126.
Bos, Henny M.W., van Balen, Frank, van den Boom, Dymphna C., 2007. Child adjustment and parenting in planned lesbian parent families. American Journal
of Orthopsychiatry 77, 3848.
Brewaeys, Anne, Ponjaert, Ingrid, Van Hall, Eylard V., Golombok, Susan, 1997. Donor insemination: child development and family functioning in lesbian
mother families. Human Reproduction 12, 13491359.
Brown, Susan L., 2004. Family structure and child well-being: the signicance of parental cohabitation. Journal of Marriage and Family 66 (2), 351367.
Busby, Dean M., Holman, Thomas B., Taniguchi, Narumi, 2001. RELATE: relationship evaluation of the individual, family, cultural, and couple contexts.
Family Relations 50, 308316.
Collins, Nancy L., Read, Stephen J., 1990. Adult attachment, working models, and relationship quality in dating couples. Journal of Personality and Social
Psychology 58, 644663.
Crowl, Alicia L., Ahn, Soyeon, Baker, Jean, 2008. A meta-analysis of developmental outcomes for children of same-sex and heterosexual parents. Journal of
GLBT Family Sciences 4 (3), 385407.
Finer, Lawrence B., Henshaw, Stanley K., 2006. Disparities in rates of unintended pregnancy in the United States, 1994 and 2001. Perspectives on Sexual and
Reproductive Health 38, 9096.
Fulcher, Megan, Sutn, Erin L., Patterson, Charlotte J., 2008. Individual differences in gender development: associations with parental sexual orientation,
attitudes, and division of labor. Sex Roles 57, 330341.
Gartrell, Nanette K., Bos, Henny M.W., 2010. US national longitudinal lesbian family study: psychological adjustment of 17-year-old adolescents. Pediatrics
126 (1), 111.
001000
Gartrell, Nanette K., Bos, Henny M.W., Goldberg, Naomi G., 2011a. Adolescents of the U.S. national longitudinal lesbian family study: sexual orientation,
sexual behavior, and sexual risk exposure. Archives of Sexual Behavior 40, 11991209.
Gartrell, Nanette K., Bos, Henny M.W., Goldberg, Naomi G., 2011b. New trends in same-sex sexual contact for American adolescents? Archives of Sexual
Behavior. http://dx.doi.org/10.1007/s10508-011-9883-5.
Gates, Gary J., 2011. Family formation and raising children among same-sex couples. NCFR Report 56 (4), F1F3.
Gates, Gary J., Ost, Jason, 2004. The Gay and Lesbian Atlas. The Urban Institute Press, Washington, DC.
Goldberg, Abbie E., 2010. Lesbian and Gay parents and Their Children: Research on the Family Life Cycle. APA Books, Washington, DC.
Golombok, Susan, Perry, Beth, Burston, Amanda, Murray, Clare, Mooney-Somers, Julie, Stevens, Madeleine, Golding, Jean, 2003. Children with lesbian
parents: a community study. Developmental Psychology 39, 2033.
Golombok, Susan., Tasker, Fiona., Murray, Clare., 1997. Children raised in fatherless families from infancy: family relationships and the socioemotional
development of children of lesbian and single heterosexual mothers. Journal of Child Psychology and Psychiatry 38, 783792.
Hatzenbuehler, Mark L., Keyes, Katherine M., Hasin, Deborah S., 2009. State-level policies and psychiatric morbidity in lesbian, gay, and bisexual
populations. American Journal of Public Health 99 (12), 22752281.
Herbenick, Debby, Reece, Michael, Schick, Vanessa, Sanders, Stephanie A., Dodge, Brian, Fortenberry, J.Dennis, 2010. Sexual behavior in the United States:
results from a national probability sample of men and women ages 1494. Journal of Sexual Medicine 7 (Suppl. 5), 255265.
Hufngton Post: Healthy Living, 2011. Child Abuse Rate at Zero Percent in Lesbian Households, New Report Finds. The Hufngton Post. <http://
www.hufngtonpost.com/2010/11/10/lesbians-child-abuse-0-percent_n_781624.html> (accessed 01.13.12).
Times Research Reporting, 2012. Interactive: Gay Marriage Chronology. Los Angeles Times. <http://www.latimes.com/news/local/la-gmtimeline-
,0,5345296.htmlstory> (accessed 01.03.12).
MacCallum, Fiona, Golombok, Susan, 2004. Children raised in fatherless families from infancy: a follow-up of children of lesbian and single heterosexual
mothers at early adolescence. Journal of Psychology and Psychiatry 45, 14071419.
Manning, Wendy D., Smock, Pamela J., Majumdar, Debarun, 2004. The relative stability of cohabiting and marital unions for children. Population Research
and Policy Review 23, 135159.
McLanahan, Sara, Sandefur, Gary, 1994. Growing Up with a Single Parent: What Hurts, What Helps. Harvard University Press, Cambridge.
Miller, Brent C., Fan, Xitao, Christensen, Matthew, Grotevant, Harold, van Dulmen, Manfred, 2000. Comparisons of adopted and nonadopted adolescents in a
large, nationally representative sample. Child Development 71 (5), 14581473.
Moore, Kristin Anderson, Jekielek, Susan M., Emig, Carol, 2002. Marriage from a Childs Perspective: How Does Family Structure Affect Children, and What
Can We Do About It? Child Trends Research Brief, Child Trends, Washington, DC.
Movement Advancement Project, Family Equality Council and Center for American Progress, 2011. All Children Matter: How Legal and Social Inequalities
Hurt LGBT Families, Full Report.
National Center for Family and Marriage Research, 2010. Same-Sex Couple Households in the US, 2009. Family Proles, FP-10-08.
Nock, Steven L., 2001. Afdavit of Steven Nock. Halpern et al. v. Canada and MCCT v. Canada. ON S.C.D.C. <http://marriagelaw.cua.edu/Law/cases/Canada/
ontario/halpern/aff_nock.pdf> (accessed 12.20.11).
Patterson, Charlotte J., 1997. Children of lesbian and gay parents. In: Ollendick, Thomas H., Prinz, Ronald J. (Eds.), Advances in Clinical Child Psychology, vol.
19. Plenum, New York.
Patterson, Charlotte J., 2000. Family relationships of lesbians and gay men. Journal of Marriage and the Family 62, 10521069.
Patterson, Charlotte J., 2006. Children of lesbian and gay parents. Current Directions in Psychological Science 15 (5), 241244.
Perrin, Ellen C., Committee on Psychosocial Aspects of Child and Family Health, 2002. Technical report: coparent or second-parent adoption by same-sex
partners. Pediatrics 109, 341344.
Redding, Richard R., 2008. Its really about sex: same-sex marriage, lesbigay parenting, and the psychology of disgust. Duke Journal of Gender Law and Policy
16, 127193.
Resnick, Michael D., Bearman, Peter S., Blum, Robert W., Bauman, Karl E., Harris, Kathleen M., Jones, Jo, Tabor, Joyce, Beuhring, Trish, Sieving, Renee E., Shew,
Marcia, Ireland, Marjorie, Bearinger, Linda H., Udry, J.R., 1997. Protecting adolescents from harm: ndings from the national longitudinal study on
adolescent health. Journal of the American Medical Association 278 (10), 823832.
Rosenfeld, Michael, 2007. The Age of Independence: Interracial Unions, Same-Sex Unions and the Changing American Family. Harvard University Press,
Cambridge, MA.
Rosenfeld, Michael J., 2010. Nontraditional families and childhood progress through school. Demography 47, 755775.
Rostosky, Sharon Scales, Riggle, Ellen D.B., Horne, Sharon G., Miller, Angela D., 2009. Marriage amendments and psychological distress in lesbian, gay, and
bisexual (LGB) adults. Journal of Counseling Psychology 56 (1), 5666.
Sirota, Theodora, 2009. Adult attachment style dimensions in women who have gay or bisexual fathers. Archives of Psychiatric Nursing 23 (4), 289297.
Snijders, Tom A.B., 1992. Estimation on the basis of snowball samples: how to weight? Bulletin de Mthodologie Sociologique 36, 5970.
Stacey, Judith, Biblarz, Timothy J., 2001a. (How) does the sexual orientation of parents matter? American Sociological Review 66 (2), 159183.
Stacey, Judith, Biblarz, Timothy, 2001b. Afdavit of Judith Stacey and Timothy Biblarz. Halpern et al. v. Canada and MCCT v. Canada. ON S.C.D.C. <http://
www.samesexmarriage.ca/docs/stacey_biblarz.pdf> (accessed 12.20.11).
Tasker, Fiona, 2005. Lesbian mothers, gay fathers, and their children: a review. Developmental and Behavioral Pediatrics 26 (3), 224240.
Tasker, Fiona, 2010. Same-sex parenting and child development: reviewing the contribution of parental gender. Journal of Marriage and Family 72, 3540.
Tasker, Fiona L., Golombok, Susan, 1997. Growing Up in a Lesbian Family. Guilford, New York.
Vanfraussen, Katrien, Ponjaert-Kristoffersen, Ingrid, Brewaeys, Anne, 2003. Family functioning in lesbian families created by donor insemination. American
Journal of Orthopsychiatry 73 (1), 7890.
Veldorale-Brogan, Amanda, Cooley, Morgan, 2011. Child outcomes for children with LGBT parents. NCFR Report 56 (4), F15F16.
Wainright, Jennifer L., Patterson, Charlotte J., 2006. Delinquency, victimization, and substance use among adolescents with female same-sex parents. Journal
of Family Psychology 20 (3), 526530.
Wainright, Jennifer L., Russell, Stephen T., Patterson, Charlotte J., 2004. Psychosocial adjustment, school outcomes, and romantic relationships of adolescents
with same-sex parents. Child Development 75 (6), 18861898.
001001

TAB 58

While sample size issuesas well as concerns about representativenesshave long hampered the general line of inquiry
into same-sex parents and child outcomes, prior to the NFSS most suppositions about possible problems with studies based
on nonrandom samples were intellectual rather than data-based. That is, it was easy for scholars to admit the limitations of
their study samples. What was more difcult, however, was to grasp just how nonrandom they were and how that might
affect their results (Marks, 2012). Even while family scholars have long acknowledged the likelihood of demographic diver-
sity among same-sex households, most have been unable to document the extent of this diversity in a statistically-meaning-
ful way. National probability surveys have typically been constrained by the relatively small number of same-sex households
in the general population, resulting in small sample sizes and limited statistical power to detect between-group differences.
Most research has instead relied on snowball and convenience samples, which often minimize genuine racial, socioeco-
nomic, and geographic heterogeneity (Tasker, 2005). Others have turned to the Census and the American Community Survey
for more representative demographic characteristics of same-sex couples with children (Rosenfeld, 2010; Gates and Ost,
2004). However, these population-based resources are not able to tell us about gay or lesbian single parents or non-residen-
tial parents. In addition, Census data provide very little detail about the diversity of family structures experienced by chil-
dren of same-sex parents over time.
Thus the original NFSS study, while subject to its own documented limitations, suggested the possibility that previous
nonrandom studies were painting a rosier picture of child outcomes than would be the case were a more random sample
to be employed or if the outcomes were based on the reports of young adults themselves rather than relying on parental
self-reports. In other words, the original study muddied what had largely been, up to that time, a relatively consistent, po-
sitive portrait of child outcomes in gay and lesbian households (however dened).
In this article, I address six areas of concern with the original study, including an extended discussion of the challenges of
dealing with household and relational instability in analyses, before briey reporting the results of alternative approaches to
presenting overview data. Throughout the article I make greater use of the NFSSs detailed family history calendar data to
look at the variety of family structure experiences in the households in which young adults reported maternal same-sex rela-
tionship behavior.
2. Responses to criticisms
2.1. What constitutes an LM or GF respondent?
Concern about the use of the acronyms LM (lesbian mother) and GF (gay father) in the original study is arguably the most
reasonable criticism. In hindsight, I wish I would have labeled LMs and GFs as MLRs and FGRs, that is, respondents who re-
port a maternal (or mothers) lesbian relationship, and respondents who report a paternal (or fathers) gay relationship.
While in the original studys description of the LM and GF categories I carefully and accurately detailed what respondents
t the LM and GF categories, I recognize that the acronyms LM and GF are prone to conate sexual orientation, which the
NFSS did not measure, with same-sex relationship behavior, which it did measure. The original study, indeed the entire data
collection effort, was always focused on the respondents awareness of parental same-sex relationship behavior rather than
their own assessment of parental sexual orientation, which may have differed from how their parent would describe it.
Therefore, I will use the (albeit awkward) dual acronyms of LM/MLR and GF/FGR to provide orienting reference to the ori-
ginal studys acronym while capitalizing on the more appropriate acronym, which I begin using exclusively in the section on
new analyses.
Some critics have correctly noted that the LM/MLR measure includes respondents who appear to have lived both with
their mother and her romantic partner for many years, as well as respondents who never lived with their mothers romantic
partner. The relationship(s) may or may not have been briefthe NFSS survey did not directly inquire about their number or
duration. While it is possible that a one-night stand might have sufced as a denition here, it stretches the imagination to
hold that many respondents would have (a) been aware of such solitary experiences, (b) classify it/them as a romantic rela-
tionship, and (c) list it when queried. In my own studies of heterosexual behavior, romantic relationships are typically per-
ceived as enduring for far longer than an evening. In Wave III of the National Longitudinal Study of Adolescent Health, less
than three percent of all young adults sexual relationships that were identied by respondents as romantic in content
(rather than nonromantic) lasted for only a day (Regnerus and Uecker, 2011). However, it is a fair request to assess those
LM/MLR respondents who lived with their mother and her romantic partner separately from those that did not. I do so
below.
2.2. Comparing apples to oranges?
The most consistent criticism is that the original studys analyses compare apples to oranges. That is, the primary com-
parison is between LM/MLRs, GF/FGRs, and intact biological families (IBFs), and that given prevalent instability in the NFSS
sample of the former pairs households, that to compare them to IBFs is to cause the former pair to look poorly. However, if
stability is a key asset for households with children, then it is sensible to use intact biological families in any comparative
assessment. But this has rarely been the approach employed in past research: Rosenfeld (2010: 757) notes that of the 45
001003
studies listed in Taskers (2005) review article, only two included a more traditional family control group built into the
study.
Moreover, it is inaccurate to imply that the original study did not evaluate distinctions between LM/MLRs and other cat-
egories that displayed some degree of instability. Tables 24 in the original study (not shown) displayed indicators of sta-
tistically-signicant differences between LM/MLRs and all other groups, and I briey describe on page 13 (Section 3.2) of
the original study text the number of (and percent of possible) statistically-signicant differences both before and after con-
trols between both LM/MLR and GF/FGR categories and all non-IBF groups.
The primary concern here, I presume, is that the LM/MLR and GF/FGR categories are comprised of households that have
experienced varying degrees of instability, and that similar experiences of instability in the one ought to be compared with
similar experiences in the other. In an ideal data world, that makes sense. But this is not as simple as it might seem, since
there is likewise varying degrees of instability in the groups denoted as stepfamily and single parent in the original
study. The household rosters, assessed over the course of 18 years, reveal quite diverse degrees of instability in stepfamilies
and single-parent households. For example, some respondents in the single parent category certainly witnessed their
never-married mother enter and exit multiple relationships, and yet I combined them with respondents whose mother never
entered another relationship after divorcing the respondents father. Some respondents entered a stepfamily as young chil-
dren, while others later in adolescence. Thus the apples versus oranges criticism is, upon closer inspection, not a very real-
istic one in social reality. Americans households, traced over the course of respondents rst 18 years of life, reveal
considerable family diversity that requires challengingand subjectivemeasurement decisions from researchers, as I noted
in the original text.
Many critics have focused on the small number of stably-coupled lesbian families in the NFSS data, and some have taken
this as a sign of a suspect dataset. It could be an undercount, but it may not be. A closer look at the respondents who stated
that their mother had a same-sex romantic relationship and that they lived with both her and her partner at some point fur-
ther reveals the short-term nature of many of the relationships. Of the 85 respondents who claimed such, 31 reported living
with their mothers partner for up to 1 year only.
2
An additional 20 reported this relationship for up to 2 years, ve for 3 years,
and eight for 4 years.
2.2.1. Relationship Instability: Control variable or pathway in analyses of child outcomes?
What should social scientists do about household (and by inference, parental relationship) instability that is nearly coter-
minous with a key independent variable, in this case the LM/MLR and GF/FGR categories? It is not a simple decision. Control
for instability?
3
But what does it mean to control for instability in this scenario? It is quite possible that household instabil-
ityvia parental romantic-relationship fragilitywas a key pathway or mechanism linking the LM/MLRs with the comparatively
higher emotional and social challenges they report. This tendency to overlook pathways in favor of control variables more
broadly reects a typical misguided tendency in social science research to always search for independent effects of variables,
often missing the pathways explaining how social phenomena actually operate. In this case, parental same-sex relationships,
family instability, and more problematic young-adult life outcomes are quite possibly linked. In assessing young-adult out-
comes, controlling for the effect of a parents same-sex relationship with a family instability variable and concludingpre-
sumablythat there is no association could well be the wrong thing to do. This is controlling for the pathways, a model
that is unhelpful for understanding social reality. If, for example, most men smoked, but very few women ever did so, it is en-
tirely unhelpful to declare thatcontrolling for smokingthere is no effect of gender on lung cancer. In that case, mens predi-
lection for smoking would merit close scrutiny and concern. Indeed, a key purpose of social science is to identify and
understand the various underlying causal mechanisms that produce identiable outcomes and events of interest (Smith,
2010: 293).
2.2.2. Gay and lesbian relationship instability: An artifact of the past?
Since the NFSS did not select by design a group of unstable gay or lesbian parents, a key issue is whether or not the LM/
MLR and GF/FGR households are more unstable than those of heterosexual couples. If stability was comparatively rarer in the
lives of MLRs and FGRs growing up some decades ago when stigma was more pronounced and social support for lesbian and
gay parents far more modest than today, is it a safe assumption that the NFSS study is a dated one by denition and that if
the study could be replicated in the future that the associations here would very likely disappear? Perhaps, but hardly cer-
tain: assumptions about comparative relationship stability among gay and lesbian couplesincluding parentscan and have
been empirically tested using other data on current relationships.
2
As I note below in greater detail, I have included in the LM/MLR group the 12 cases in which the respondent indicated that both parents had had a same-sex
relationship. In the previous study, I analyzed them only as GF/FGRs, given sample-size concerns.
3
One option is to utilize the NFSS calendars and create a measure of the number of household transitions rather than the experience of one or more
transitions (Potter, 2012). But the household calendars could well miss the exact number of transitions, since the NFSS only asked respondents to denote when
someone else lived with them for at least 4 months. This also overlooks parental romantic relationships which were either brief or else not residential (yet
potentially still inuential). And in cases of excessive household instability, respondents may experience survey fatigue and may underreport transitions when
lling out what amounts to be for them a rather complicated household calendar. Moreover, to suggest that all romantic partner dissolution creates problems
for respondents is short-sighted. Indeed, some dissolutions solve problems (Amato, 2000). Such is the messy business of documenting and assessing household
histories.
001004
A study of Norwegian and Swedish same-sex marriages notes that divorce risk is higher in same-sex marriages and that
the risk of divorce for female partnerships actually is more than twice that for male unions (Andersson et al., 2006: 89).
Moreover, early same-sex marriagesthose occurring shortly after a shift in marriage lawexhibited a similar risk of divorce
as did more recent marriages, suggesting no notable variation in instability over time as a function of new law or pent-up
demand among more stable, longstanding relationships. The study authors estimate that in Sweden, 30% of female marriages
are likely to end in divorce within 6 years of formation, compared with 20% for male marriages and 13% for heterosexual
ones. Moreover, they found lesbian couples to be more sociodemographically homogamous than other couples, and spec-
ulate that this situation may be conducive to a high level of dynamism in the relationship, but perhaps not to the kind of
inertia that is related to marital stability (Andersson et al., 2006: 96). Biblarz and Stacey (2010: 17) similarly note this phe-
nomenon in their review of research on lesbian parents, asserting that they face a somewhat greater risk of splitting up,
due in part, they suggest, to their their high standards of equality. A follow-up assessment of more recent Norwegian sta-
tistics, presented at the 2012 annual meeting of the Population Association of America (PAA), found no evidence that the
gender gap in same-sex divorce has closed (Noack et al., 2012).
Michael Rosenfeld detects the same pattern in a study of nationally-representative data on American relationships pre-
sented at the 2012 annual meeting of the American Sociological Association. He nds that lesbian couples report higher rela-
tionship satisfaction alongside higher break-up rates. The highest stability rates appear among heterosexual married couples,
while notably better stability is located among married gay and lesbian couples than among those in civil unions (as would
be expected). Yet his analysis too detects greater instability among lesbian couples in general, a nding that persists even
after a lengthy series of control variables are included. While lesbian couples in the study are more apt to be raising children,
the presence of children does not appear to be a factor in the diminished relationship stability evident among them.
That few LM/MLR respondents reported stability in their mother and her partners relationship (in the domicile in which
the respondent lived) ought not be simply chalked up to greater stigma or insufcient social support as factors that account
for the entirety of the association. In light of evidence of the same pattern among current lesbian couples in the US and Scan-
dinavia, it remains an open question.
While the cited study authors tend to nd the difference in divorce behavior between lesbians and gay men intriguing,
this lesbian effect is anticipated in a sexual economics approach to romantic relationships (e.g., Baumeister, 2010). This
perspective places no blame for instability on sexual orientation per se, but rather on stable gender differences and prefer-
ences in relationships (e.g., for women, a signicantly higher bar for the relationships quality and emotional satisfaction).
Gay mens relationships thus appear predictably more stable than lesbian relationships, but are less likely to be sexually
monogamous when compared with lesbian or heterosexual relationships (Hoff and Beougher, 2010). Here again, this is be-
lieved to be due not to sexual orientation but stable gender differences in relationship preferences and sex drive (Baumeister
and Vohs, 2004). While the effect of relationship stability on child health and development is well-documented and apparent
in the original NFSS studys ndingsas well as this follow-up explorationthe effect on children of parental nonmonogamy
is not well understood.
2.3. Is the NFSS a representative sample?
As an extension of the second concern, many critics have focused on the small number of stably-coupled lesbian families
in the NFSS data. Indeed, only two cases of LM/MLRs reported living with their mother and her partner uninterrupted from
age 1 to 18. Of the 85 cases (out of 175 total LM/MLRs) wherein the respondent indicated living in residence for a time with
both their mother and her female partner, only 19 spent at least ve consecutive years together, and six cases spent 10 or
more consecutive years together. Some have taken this as a sign of a suspect and non-representative dataset. It could be an
undercount, but it may well not be. Rather, readers would do well to keep in mind anachronistic expectations concerning an
era in which enduring same-sex relationships with children were simply less common, and those that existed certainly sub-
ject to greater social scrutiny and stigma. And, as noted above, there may be stability distinctions that foster unreasonable
expectations, especially following upon decades of research conclusions based on nonrandom samples.
Moreover, such expectations also tend to reveal a class bias that may hamper studies in this domain, given that families
wherein same-sex couples pursue the complicatedand potentially quite expensiveprocess of deciding just how and when
they will have a child tend to be more educated, wealthy, and white than the families of many NFSS LM/MLRs. Rosenfeld
(2010: 757) notes:
. . .the literature on same-sex couple parenting has tended to feature studies of the kind of women who can afford ART:
white, upper-middle-class women. Nationally representative data tend to paint a different picture: in the US census,
same-sex couple parents tend to be more working class and are much more likely to be nonwhite compared with heter-
osexual married couples.
The children of such a selective groupthose who conceive by ART, or assisted reproductive technologywould be ex-
pected to witness greater stability and to fare better, enjoying advantages that tend to benet children regardless of their
parents race, age, or sexual orientation. While this selective group is hardly the only face of same-sex parents in America,
they are the ones who receive the majority of popular and scholarly attention.
In his assessment of group differences in academic progress, moreover, Rosenfeld (2010) restricted his Census-based sam-
ple to the children of same-sex couples who had been living with both parents for at least 5 years, thus raising the like-
001005
Table 1
Mean scores on select dichotomous outcome variables, NFSS (can read as percentage: as in, 0.43 = 43%).
1-IBF 2-MLR no partner 3-MLR + partner 4-FGR 5 6 7 8 9 10 11 12 13 14 15
Currently married 0.43 0.31 0.38 0.38 0.36
0.49 0.37 0.41 0.27 0.21
0.17
0.47 0.63 0.41 0.45

Currently cohabiting 0.09 0.18 0.27
0.23 0.31
0.11 0.20
0.10 0.25
0.18
0.31
0.22
0.28
0.07
^
0.32
Family received welfare growing up 0.17 0.72
0.70
0.51
0.47
0.41
,^
0.49
0.37
^
0.70
0.75
0.56 0.58
0.13
^
0.12
^
0.47
Currently on public assistance 0.10 0.32 0.49
0.14
^
0.31
0.21
^
0.22
,^
0.27
0.52
0.44
0.49
0.28 0.11
^
0.27
0.25
Currently employed full-time 0.49 0.36 0.17
0.36 0.42
,^
0.48
^
0.44
^
0.55
^
0.42
^
0.31 0.09
0.52
^
0.75
^
0.41
^
0.42
^
Currently unemployed 0.08 0.10
^
0.40
0.23 0.15
^
0.13
^
0.15
^
0.06
^
0.18 0.19 0.34
0.03
^
0.00
^
0.22 0.12
^
Voted in last presidential election 0.57 0.46 0.43 0.71
,^
0.63 0.53 0.58 0.52 0.58 0.43 0.37 0.70
,^
0.44 0.58 0.59
Thought recently about suicide 0.05 0.23
0.09 0.17 0.08 0.10 0.06 0.02 0.11 0.09 0.03 0.04 0.01 0.07 0.11
Recently or currently in therapy 0.08 0.30
0.17 0.10 0.12 0.17
0.20
0.11 0.24
0.13 0.09 0.13 0.01

,^
0.22
0.09
Identies as entirely heterosexual 0.90 0.45
0.68
0.80
0.83 0.82
0.82
0.89
^
0.80 0.77
0.83 0.91
^
0.96
^
0.82 0.72
Is in a same-sex romantic relationship 0.04 0.12 0.02 0.14 0.05 0.15
0.01 0.00
0.05 0.01 0.04 0.13
- 0.23 0.21
Had an affair while married/cohabiting 0.13 0.42
0.38
0.26 0.12
^
0.28
0.17 0.09
^
0.48
0.23 0.18 0.23 0.18 0.20 0.30

Has ever had an STI 0.08 0.21 0.26
0.18
0.12 0.12 0.17
0.06
^
0.25
0.19
0.26
0.17 0.08 0.16 0.12

Ever touched sexually by an adult 0.02 0.16
0.26
0.07 0.10
0.09
0.10
0.10
0.20
0.15
0.11 0.05
^
0.02
^
0.03
^
0.09
Ever forced to have sex against will 0.08 0.42
0.27
0.17
0.24
0.18
0.20
0.13 0.17 0.17 0.11 0.12 0.10 0.23
0.17
1 = Lived with both bio mother and father from 0 to 18 or until left home (N = 919).
2 = MLR, but never lived with mothers same-sex romantic partner (N = 90).
3 = MLR, spent time in residence with mothers same-sex romantic partner (N = 85).
4 = FGR (N = 61).
5 = Lived with both bio mom and dad until 18, but subsequently theyve gotten a divorce (N = 116).
6 = Parents were married, but got a divorce, R lived with mother, and R reported subsequent relationship(s) and remarriage (N = 223).
7 = Parents were married, but got a divorce, R lived with mother, and R reported subsequent relationship(s) but no remarriage (N = 278).
8 = Parents were married, but got a divorce, R lived with mother, and R reported NO subsequent relationship before 18 (N = 108).
9 = Parents never married, R lived with mother, and R reported subsequent relationship(s) and marriage (N = 104).
10 = Parents never married, R lived with mother, and R reported subsequent relationship(s) but no marriage (N = 221).
11 = Parents never married, R lived with mother, and R reported NO subsequent relationship (N = 48).
12 = Parents were married, but one parent died, and R reported subsequent relationship(s), possibly including remarriage (N = 117).
13 = Parents were married, but one parent died, and R reported NO subsequent relationship (N = 28).
14 = Adopted by strangers at birth or 1 year (at some point, either one or two adopted parents) (N = 101).
15 = Parents were married, but got a divorce, R lived with father (84% of the time, R said father had another relationship) (N = 95).
Bold indicates the mean scores displayed are statistically-signicantly different from IBFs (currently intact, bio mother/father household, column 1), without additional controls.
An asterisk (
) next to the estimate indicates a statistically-signicant difference (p < 0.05) between the groups coefcient and that of IBFs, controlling for respondents age, gender, race/ethnicity, level of mothers
education, perceived household income while growing up, experience being bullied as a youth, and states legislative gay-friendliness, derived from logistic regression models (not shown).
A caret (
^
) next to the estimate indicates a statistically-signicant difference (p < 0.05) between the groups mean and the mean of Group 3 (MLR + partner), without additional controls.
1
3
7
2
M
.
R
e
g
n
e
r
u
s
/
S
o
c
i
a
l
S
c
i
e
n
c
e
R
e
s
e
a
r
c
h
4
1
(
2
0
1
2
)
1
3
6
7
1
3
7
7
001007
Table 2
Mean scores on select continuous outcome variables, NFSS.
Educational attainment 3.19 2.34
2.41
2.70 2.88
,^
2.72
2.82
,^
3.06
^
2.41
2.18
2.01
2.78 2.92 3.21

^
2.79
Family-of-origin safety/security 4.13 3.23
2.97
3.35
3.52
,^
3.70
,^
3.45
,^
3.71
,^
3.35
3.44
,^
3.59
,^
3.63
,^
4.02
^
3.77
,^
3.12
Family-of-origin negative impact 2.30 3.30
2.97
2.89
2.96
2.67
2.97
2.55 3.04
2.74
3.02
2.72
2.62 2.83
2.67
Closeness to biological mother 4.17 4.07 4.03 3.71
3.95 4.26
3.88 3.90 3.63 3.50
4.20 3.87 4.17 3.58 3.79

Closeness to biological father 3.87 3.16 3.18 3.44 3.29
3.53 3.29 2.77
- 1.57
,^
3.01 3.28 3.27 - 3.89
Self-reported physical health 3.75 3.50 3.24 3.67 3.46 3.51 3.58
^
3.42 3.40 3.28
3.09
3.54 3.66 3.53 3.54

Self-reported overall happiness 4.16 3.63 4.04 3.79 4.02 3.94 3.93 3.83 3.88 3.70 3.64 4.03 4.58
,^
3.92 3.80
CES-D depression index 1.83 2.37
2.12 2.07 2.01 1.88 1.92 1.84 2.02 2.08 1.99 1.76
^
1.48
,^
1.95 1.90
Attachment scale (depend) 2.82 3.63
3.27 3.10 3.08 3.00 3.12
2.84 3.26 3.22
3.40
3.16 2.52
^
3.12
3.10
Attachment scale (anxiety) 2.46 2.77 2.63 2.60 2.71 2.47 2.54 2.41 2.66 2.65 2.77 2.51 2.03 2.66
2.49
Impulsivity scale 1.90 2.03 2.06 1.95 1.94 1.79
^
1.93 1.84
^
1.98 1.79
^
1.81 1.86 1.66
,^
1.85 1.76
^
Level of household income 8.27 6.45
5.96 7.08 7.42 7.46

^
7.67
^
7.34 5.72
5.38
3.67
,^
7.68
^
9.03
,^
7.93
^
7.73
^
Current relationship quality index 4.11 3.80 3.76 3.73
3.95 3.88 3.94 3.92 3.65 3.66
3.92 3.77 4.36

^
3.79 4.02
Current relationship is in trouble 2.04 2.60
2.21 2.47 2.43 2.15 2.32
2.19 2.77
,^
2.45
2.60 2.31 1.85 2.35 2.31

An asterisk (
) next to the estimate indicates a statistically-signicant difference (p < 0.05) between the groups coefcient and that of IBFs, controlling for respondents age, gender, race/ethnicity, level of
mothers education, perceived household income while growing up, experience being bullied as a youth, and states legislative gay-friendliness, derived from OLS regression models (not shown).
A caret (
^
M
.
R
e
g
n
e
r
u
s
/
S
o
c
i
a
l
S
c
i
e
n
c
e
R
e
s
e
a
r
c
h
4
1
(
2
0
1
2
)
1
3
6
7
1
3
7
7
1
3
7
3
001008
Table 3
Mean scores on select event-count outcome variables, NFSS.
Frequency of marijuana use 1.32 1.78 1.85
1.62 2.00
1.32 1.71
1.61 1.86
1.99
1.70 1.50 1.62 1.33 1.50

Frequency of alcohol use 2.70 2.58 2.41 2.42 2.55 2.35 2.64 2.87 2.84
2.63 1.89 2.55 2.59 2.74 2.84

Frequency of drinking to get drunk 1.68 1.89 1.88 1.89 1.90 1.58 1.75 1.91 1.96
1.78 1.37 1.73 1.32 1.73 1.68

Frequency of smoking 1.79 2.95
2.84
2.22 2.44 2.25
2.03 2.31 2.38 2.27 2.14 1.90 2.59 2.34
2.44
Frequency of watching TV 3.01 4.21
3.46 3.17 3.33 3.21 3.24 3.47 3.98
3.50 3.51 3.37 2.27

^
3.31 2.77
Frequency of having been arrested 1.18 1.82
1.76
1.52 1.38 1.39

,^
1.37
,^
1.17
^
1.34
^
1.43
1.47
1.27
^
1.37 1.31
^
1.53
Freq pled guilty to non-minor offense 1.10 1.43
1.35
1.36 1.30 1.20 1.21
1.10
^
1.22 1.15 1.18 1.20 1.23 1.19 1.24
N of female sex partners (among women) 0.22 1.66
,^
0.70
0.74
0.96
0.52
0.41 0.14
^
0.51 0.64
0.94 0.52 0.36 0.47 0.47
N of female sex partners (among men) 2.70 2.37

^
3.97 4.16 3.66 3.79 3.30 2.03
3.91
4.38
2.06 4.52
3.43 3.24 3.60

N of male sex partners (among women) 2.79 5.73
,^
2.98 4.51
3.97
4.55
,^
4.05
,^
3.70 4.90
,^
4.42
,^
4.13 3.38 3.36 3.49 4.53
,^
N of male sex partners (among men) 0.20 2.13
1.18
1.47
0.98 0.37 0.10

^
0.72 0.20 0.35
0.62
0.21 0.47
0.27 1.76
An asterisk (
) next to the estimate indicates a statistically-signicant difference (p < 0.05) between the groups coefcient and that of IBFs, controlling for respondents age, gender, race/ethnicity, level of mothers
education, perceived household income while growing up, experience being bullied as a youth, and states legislative gay-friendliness, derived from Poisson or negative binomial regression models (not shown).
A caret (
^
1
3
7
4
M
.
R
e
g
n
e
r
u
s
/
S
o
c
i
a
l
S
c
i
e
n
c
e
R
e
s
e
a
r
c
h
4
1
(
2
0
1
2
)
1
3
6
7
1
3
7
7
001009
their early years were spent with both their biological mother and her same-sex partner. The household presence of a same-sex
partner begins emerging slowly but steadily through the course of childhood. In numerous cases LM/MLR respondents indicated
rst living with their mothers girlfriend/partner at a comparatively older age (for example, 54 began at or after age 10, 40 at or
after age 13, and 18 at or after age 16).
Whether these were in fact mixed-orientation marriages or relationships is of course impossible to discern with con-
dence, since the study did not ask the respondents to identify their parents sexual orientation, a decision I remain comfort-
able with given the era the data are describing. Many LM/MLR and GF/FGR respondents may well have witnessed their
parents mixed-orientation marriage. On the other hand, given the documented uidity of womens sexuality, I would hes-
itate to assert that a same-sex relationshipespecially if relatively briefis indicative of a xed sexual orientation (Diamond,
2008).
While the etiology of homosexuality is not under study here, the matter seems tacitly embedded in criticisms about clas-
sication. As such, the original study should be understood in the manner in which it is explicitly titledabout the adult
children of parents who have same-sex relationships. If for whatever reason that is an unsatisfying anchorparental sexual
behavior rather than orientationit is beyond the scope of an academic study to be something it is not. Nevertheless, it sug-
gests the importance of consistently employing the acronyms MLR and FGR.
2.5. Bisexuality in the NFSS?
As an extension of this, a few critics have raised the possibility that plenty of the NFSS LM/MLRs and GF/FGRs may in real-
ity be bisexual in orientation. In an unpublished study of the most recent two series of data from the National Survey of Fam-
ily Growthpresented at the 2012 PAA conferenceDanielle Wondra reports that self-identied bisexual men and women
are notably more likely to desire a (or another) child than self-identied gay or lesbian respondents. Sufce it to say that
more research needs to be conducted on bisexual parents outside of a simplistic mixed-orientation rubric that may not
reect the reality of many couples history of sexual experiences or preferences. Moreover, claims about mixed orientation
marriages unnecessarily problematize bisexuality by prioritizing a dualistic (either/or) essentialism about sexual orienta-
tion that may not t social reality (Diamond, 2008).
If the complex calendar histories are any clue, bisexuality is probable among some NFSS respondents parents. Such fre-
quencies of opposite-sex relationship behavior or opposite-sex attraction are not out of step with other studies of same-sex
partnerships (Andersson et al., 2006; Potter, 2012; Rosenfeld, 2012). Nevertheless, only four LM/MLRs reported an opposite-
sex parent gurea stepfatherliving in the household after having reported a same-sex parent gure (i.e., a mothers girl-
friend/partner). In sum, the B in LGBT parenting deserves more attention than it has been given, and may constitute a more
signicant share of such households-with-children than has often been recognized.
2.6. Foster care experiences
A few critics have raised the suggestion that in the era represented by the NFSS respondents, gay and lesbian parents were
more apt to either adopt foster children, orat the other extremefaced the forcible placement of their own children in fos-
ter care. Either scenario raises concern about the original studys claim that LM/MLR respondents were the most apt to report
experience with the foster care system. This concern prompted a detailed exploration of the calendar data for the 21 LM/MLR
respondents who reported such an experience, in order to discern the timing of their foster system experience. As with the
original studys discussion about the timing of sexual victimization, here too the story is muddied. Three of the 21 LM/MLRs
who spent time in foster care did so immediately prior to reporting living in a household with their mother and her female
partnerone of the two scenarios anticipated by critics. Four of the 21 spent some time in foster care following their report
of living in a household with their mother and her partnerthe other scenario that concerned critics. Whether any of these
seven cases actually match those scenarios in reality is impossible to know from the data. The remaining 14 cases display
calendar data less apt to suggest either of these two scenarios as a likely t. Just under half of the 21 respondents reported
their foster care experience beginning before age 10.
3. Alternative analyses
Tables 13 display results in a manner similar to Tables 24 in the original study (not shown), with several changes made
in response to criticisms:
1. I split the LM/MLRs (hereafter, MLRs) between those who never lived with their mothers same-sex romantic partner and
those that have.
Why this particular division? Of the 85 cases wherein the respondent indicated living in residence with both their mother
and her female partner, only 19 spent ve consecutive years together, and six cases spent 10 consecutive years together.
While this is not quite the comparison some critics seek, the statistical power is simply not present for a direct comparison
001010
of the most stable MLRs, given uncommon relationship longevity in their households-of-origin. It is true, though, that greater
longevity of such in residence relationships tended to reveal better outcomes at face value.
2. I shifted the 12 cases wherein a respondent reported that both parents had had a same-sex relationship from FGR to MLR.
As noted in the original study, analyses of the household calendar data for these 12 cases revealed comparable exposure
to both their mother and father. As a result, there are now 90 MLR cases who never reported living with their mothers part-
ner/girlfriend, 85 MLRs who did, and 63 FGRs. As reported in the original study, the latter group very infrequently reported
living with their father and his partner/boyfriend, so this group remains unaltered in its structure.
3. I expanded the total number of groups to 15 in order to better reect the different experiences of stability and partnering
in American households. I did not include an others catch-all group in this set of analyses. As a result, the nal tables
reect just under 400 fewer cases than in the original study.
Given the outcome measures are the same as employed in the original study, I do not describe their operationalization
here. That can be located in the original studys text and its Appendix B. The analytic strategyan overview featuring both
simple between-group means tests as well as an indicator of statistical signicance after controlling for several independent
variables via outcome-appropriate forms of regression analysesremains the same as well, for comparability.
As was the case in the original analyses, Tables 13 reveal that those adult children who report a maternal same-sex rela-
tionshipregardless of whether their mother ever resided with her same-sex partnerlook far more similar to adult children
of other types of households than they do to those from stably-intact biological families. There are 20 simple statistically-
signicant differences between group 2 (MLRs who never lived with their mothers same-sex partner) and IBFs, and an iden-
tical number between group 3 (MLRs who did live with their mothers same-sex partner for a time) and IBFs. After controls
via regression analysisthere are 21 and 19 statistically-signicant differences between groups 2 and 3, respectively, and
IBFs. These numbers are a dip from those reported in the original study.
Most of the distinctions between IBFs and groups 2 and 3 are consistent with those reported in the original study. On 16
different outcomes, both groups 2 and 3 appear statistically different from IBFs prior to controls (i.e., regression models); the
same is true of 13 outcomes after controls. There are nine simple differences between FGRs and IBFs prior to controls, and 12
after them. As in the original study, distinctions between the two MLR groups and IBFs appear in the domains of sexuality,
sexual behavior, sexual victimization, household economics and work, educational attainment, smoking, arrests, and retro-
spective sentiment about family life while growing up.
7
Carets denote a simple statistically-signicant difference between group 3 (MLRs who spent time living with their
mothers partner) and all non-IBF groups. Of the 517 possible between-group differences, 89% (or 17%) appear signicant
at the bivariate level, a decline from the 24% gure when assessing all MLRs together in the original study. Several groups
compare similarly to group 3 in terms of very few simple differences:
Group 4 (FGRs): two differences.
Group 11 (never-married single mothers with no subsequent relationships): two differences.
Group 9 (single mothers who subsequently remarried): four differences.
Group 10 (never-married single mothers with relationships but no marriage): four differences.
Group 2 (MLRs who did not live with their same-sex partner): four differences.
Group 10 displays by far the most pre- and post-regression statistically-signicant differences with IBFs (31 and 23,
respectively), and tends to fare consistently poorly across most outcomes which are agreeably suboptimal. Group 3 (MLRs
who lived with their mothers partner) compare less favorably with:
Group 8 (divorced, lived with mother, no subsequent relationships): 12 differences.
Group 13 (parents married until one died, no subsequent relationships): 15 differences.
In general, groups 8 and 13 fared rather well on many outcomes, shedding light on the likely importance of avoiding fur-
ther household transitions. Where outcomes are clearly discernible as optimal or suboptimalfor example, educational
attainment or STI, respectivelygroup 8 fares better than groups 67, whose only distinction is subsequent maternal roman-
tic relationships and, in group 6s case, remarriage. Additional parental romantic partners, even remarriages, seem to make a
(negative) difference. As in the original study, there is much that these analyses cannot document, including causation as
well as any effects of sexual orientation. Selectivity is very likely at work on multiple outcomes.
Analyses comparing younger versus older NFSS respondents may prove a fertile avenue of exploration. Initial ancillary
analyses suggest that older young adult MLRs seem to have struggled more than younger ones. Whether this is a function
7
As noted in the original study text, the NFSS data is insufciently capable of discerning much information about the context surrounding respondents
sexual victimization. No simplistic conclusions about it ought to be discerned from the analyses.
001011
of time exposure, or more pronounced social stigma further in the past than among the newest young adult MLRs, is dif-
cult to say, given the interpretive limitations of this data. Alternately, some challenges may cumulate over time; it may be
that the older respondents have simply had more time to experience particular outcomes.
4. Conclusion
This follow-up study has sought to address six common criticisms that have arisen following the July 2012 publication in
this journal of the original study entitled, How different are the adult children of parents who have same-sex relation-
ships? One in particular, about comparing stable heterosexual couples to stable same-sex couples, is particularly challeng-
ing to accomplish with all but the very largest datasets (which, in turn, tend to have fewer interesting outcome measures). It
also raises important conceptual and analytic questions about how to navigate persistent instability in the NFSSs MLR and
FGR cases. This is complicated by contemporary evidence in the US and Scandinavia suggesting that lesbian relationships in
particularincluding legally married couplescontinue to exhibit instability in excess of heterosexual relationships and
even gay male relationships.
Perhaps in social reality there really are two gold standards of family stability and context for childrens ourishinga
heterosexual stably-coupled household and the same among gay/lesbian householdsbut no population-based sample anal-
yses is yet able to consistently conrm wide evidence of the latter. Moreover, a stronger burden of proof than has been em-
ployed to date ought to characterize studies which conclude no differences, especially in light of longstanding reliance
on nonrandom samples of unknown bias and the high risk of making Type II errors in small-sample studies (Marks,
2012; Nock, 2001). In other words, the science here remains young. Until much larger random samples can be drawn and
evaluated, the probability-based evidence that existsincluding additional NFSS analyses hereinsuggests that the biolog-
ically-intact two-parent household remains an optimal setting for the long-term ourishing of children.
Of course the ourishing of children involves many other factors besides parental relationship structure and decision-
making, as analyses of the NFSS and numerous other datasets conrm. Indeed, most young-adult respondents in the NFSS
report ample success and largely avoid problematic physical and emotional difculties, regardless of their parents experi-
ences, decisions, and actions.
References
Amato, Paul R., 2000. The consequences of divorce for adults and children. Journal of Marriage and Family 62, 12691287.
Andersson, Gunnar, Noack, Turid, Seierstad, Ane, Weedon-Fekjaer, Harald, 2006. The demographics of same-sex marriages in Norway and Sweden.
Demography 43, 7998.
Baumeister, Roy F., 2010. Is There Anything Good About Men? How Cultures Flourish by Exploiting Men. Oxford University Press, New York.
Baumeister, Roy F., Vohs, Kathleen D., 2004. Sexual economics: sex as female resource for social exchange in heterosexual interactions. Personality and
Social Psychology Review 8 (4), 339363.
Biblarz, Timothy J., Stacey, Judith, 2010. How does the gender of parents matter? Journal of Marriage and Family 72 (1), 322.
Black Dan, Gates Gary, Sanders Seth, Taylor Lowell. 2007. The Measurement of Same-Sex Unmarried Partner Couples in the 2000 US Census. California
Center for Population Research, Los Angeles.
Diamond, Lisa M., 2008. Sexual Fluidity: Understanding Womens Love and Desire. Harvard University Press, Cambridge, MA.
Gates, Gary J., Ost, Jason, 2004. The Gay & Lesbian Atlas. The Urban Institute Press, Washington, DC.
Hoff, Colleen C., Beougher, Sean C., 2010. Sexual agreements among gay male couples. Archives of Sexual Behavior 39, 774787.
Marks, Loren., 2012. Same-sex parenting and childrens outcomes: a closer examination of the American Psychological Associations brief on lesbian and gay
parenting. Social Science Research 41, 735751.
Nock, Steven L., 2001. Afdavit of Steven Nock. Halpern et al. v. Canada and MCCT v. Canada. ON S.C.D.C. <http://marriagelaw.cua.edu/Law/cases/Canada/
ontario/halpern/aff_nock.pdf> (accessed 12.20.11).
Potter, Daniel, 2012. Same-sex parent families and childrens academic achievement. Journal of Marriage and Family 74, 556571.
Regnerus, Mark, Uecker, Jeremy, 2011. Premarital Sex in America: How Young Americans Meet, Mate, and Think about Marrying. Oxford University Press,
New York.
Rosenfeld, Michael J., 2010. Nontraditional families and childhood progress through school. Demography 47, 755775.
Rosenfeld, Michael J., Thomas, Reuben J., 2012. Searching for a mate: the rise of the internet as a social intermediary. American Sociological Review 77 (4),
523547.
Smith, Christian, 2010. What is a Person? Rethinking Humanity, Social Life, and the Moral Good from the Person Up. The University of Chicago Press,
Chicago.
Tasker, Fiona, 2005. Lesbian mothers, gay fathers, and their children: a review. Developmental and Behavioral Pediatrics 26 (3), 224240.
001012

TAB 59

By way of comparison, social science research with small convenience samples has repeatedly reported no signicant dif-
ferences between children from gay/lesbian households and heterosexual households. These recurring ndings of no signif-
icant differences have led some researchers and professional organizations to formalize related claims. Perhaps none of these
claims has been more inuential than the following from the 2005 American Psychological Association (APA) Brief on Les-
bian and Gay Parenting.
12,13
Not a single study has found children of lesbian or gay parents to be disadvantaged in any signicant respect relative to
children of heterosexual parents.
Are we witnessing the emergence of a new family form that provides a context for children that is equivalent to the tra-
ditional marriage-based family? Many proponents of same-sex marriage contend that the answer is yes. Others are skeptical
and wondergiven that other departures from the traditional marriage-based family form have been correlated with more
negative long-term child outcomesdo children in same-sex families demonstrably avoid being disadvantaged in any sig-
nicant respect relative to children of heterosexual parents as the APA Brief asserts? This is a question with important
implications, particularly since the 2005 APA Brief on Lesbian and Gay Parenting has been repeatedly invoked in the cur-
rent same-sex marriage debate.
2. Statement of purpose
The overarching question of this paper is: Are the conclusions presented in the 2005 APA Brief on Lesbian and Gay Parenting
valid and precise, based on the cited scientic evidence?
14
In the present paper, seven questions relating to the cited scientic
evidence are posed, examined, and addressed.
15
Two portions of the APA Brief are of particular concern to us in connection with these questions: (a) the Summary of
Research Findings (pp. 522), and (b) the rst and largest section of the annotated bibliography, entitled Empirical Studies
Specically Related to Lesbian and Gay Parents and Their Children (pp. 2345). In the latter section (pp. 2345), the APA
references 67 manuscripts. Eight of these studies are unpublished dissertations.
16
The 59 published studies are listed in
Table 1 of this paper, providing clear parameters from which to formulate responses to the seven outlined questions, next.
2.1. Question 1: how representative and culturally, ethnically, and economically diverse were the gay/lesbian households in the
published literature behind the APA brief?
In response to question 1, more than three-fourths (77%) of the studies cited by the APA brief are based on small, non-
representative, convenience samples of fewer than 100 participants. Many of the non-representative samples contain far
fewer than 100 participants, including one study with ve participants (Wright, 1998; see Table 1). As Strasser (2008) notes:
Members of the LGBT community. . .vary greatly in their attitudes and practices. For this reason, it would be misleading to
cite a study of gay men in urban southern California as if they would represent gay men nationally (p. 37).
By extension, it seems that inuential claims by national organizations should be based, at least partly, on research that is
nationally representative.
Lack of representativeness often entails lack of diversity as well.
17
A closer examination of the APA-cited literature from the
Empirical Studies (pp. 2345) section of the APA Brief reveals a tendency towards not only non-representative but racially
homogeneous samples. For example:
12
The APA Briefs stated objective was primarily to inuence family law. The preface states that the focus of the publication. . .[is] to serve the needs of
psychologists, lawyers, and parties in family law cases. . .. Although comprehensive, the research summary is focused on those issues that often arise in family
law cases involving lesbian mothers or gay fathers (APA Brief, 2005, p. 3). Redding (2008) reports that leading professional organizations including the
American Psychological Association have issued statements and that advocates have used these research conclusions to bolster support for lesbigay parenting
and marriage rights, and the research is now frequently cited in public policy debates and judicial opinions (p. 136).
13
Patterson, p. 15 (from APA Brief, 2005).
14
Kuhn (1970/1996) has stated that there is an insufciency of methodological directives, by themselves, to dictate a unique substantive conclusion to many
sorts of scientic questions (p. 3). To draw substantive conclusions, a socially and historically inuenced paradigm is needed. Research is then directed to the
articulation of those phenomena and theories that the paradigm already supplies (p. 24). Indeed, paradigmatic biases, and other inuences, can make us
vulnerable to discrepancies between warranted and stated conclusions in the social sciences (Glenn, 1989, p. 119; see also Glenn, 1997).
15
Kuhn (1970/1996) has noted that when scientists disagree about whether the fundamental problems of their eld have been solved, the search for rules
gains a function that it does not ordinarily possess (p. 48).
16
These unpublished dissertations include: Hand (1991), McPherson (1993), Osterweil (1991), Paul (1986), Puryear (1983), Rees (1979), Sbordone (1993), and
Steckel (1985). An adapted portion of one of these dissertations (Steckel, 1985) was eventually published (Steckel, 1987) and is included in the present
examination; the other unpublished work is not included in Table 1 of this paper.
17
Of the 59 published Empirical Studies Specically Related to Lesbian and Gay Parents and Their Children, no studies mention African-American, Hispanic,
or Asian-American families in either their titles or subtitles. The reference list in the APA Briefs Summary of Research Findings (pp. 1522) is also void of any
studies focusing on African-American, Hispanic, or Asian-American families. None of the Empirical Studies Specically Related to Lesbian and Gay Parents and
Their Children (pp. 2345) holds, as its focus, any of these minorities. (Note: Three years after the 2005 APA Brief, Moore (2008) published a small but
pioneering study on AfricanAmerican lesbians.)
736 L. Marks / Social Science Research 41 (2012) 735751
001014
Table 1
Publications Cited in APA brief on lesbian and gay parenting (pp. 2345).
Author and year GayLes N Hetero N Stat used Cohen
N
Stat
power
Outcome studied Hetero compar
group
Bailey et al. (1995) 55par; 82chl 0 T-test/Chi 393 N/A Sexual orientation None
Barrett and Tasker
(2001)
101 0 T-test/Chi 393 N/A Child responses to a gay parent None
Bigner and Jacobsen
(1989a)
33 33 T-test 393 No Parents reports of values of
children
Fathers
Bigner and Jacobsen
(1989b)
33 33 T-test 393 No Parent reports of parent behavior Fathers
Bos et al. (2003) 100 100 MANOVA 393 No Parental motives and desires Families
Bos et al. (2004) 100 100 MANOVA 393 No Parent reports of couple
relations
Families
Bozett (1980) 18 0 Qualitative N/A N/A Father disclosure of
homosexuality
None
Brewaeys et al. (1997) 30 68 ANOVA 393 No Emotional/gender development DI/Non-DI
Couples
Chan et al. (1998a) 30 16 Various 393 No Division of labor/child
adjustment
DI Couples
Chan et al. (1998b) 55 25 Various 393 Reported Psychosocial adjustment DI Couples
Ciano-Boyce and
Shelley-Sireci (2002)
67 44 ANOVA 393 No Division of child care Adoptive Parents
Crawford et al. (1999) 0 0 MANOVA 393 N/A 388 Psychologists attitudes N/A
Flaks et al. (1995) 15 15 MANOVA 393 No Cognitive/behavioral/parenting Married Couples
Fulcher et al. (2002) 55 25 T-test/Chi 393 Reported DI/adult-child relationships Parents
Gartrell et al. (1996) 154 0 Descript. N/A N/A Prospective Parent Reports None
Gartrell et al. (1999) 156 0 Descript. N/A N/A Reports on parenting issues None
Gartrell et al. (2000) 150 0 Descript. N/A N/A Reports on parenting issues None
Gartrell et al. (2005) 74 0 Descript. N/A N/A Health, school/education None
Gershon et al. (1999) 76 0 Reg. 390 N/A Adolescent coping None
Golombok et al. (1983) 27 27 T-test/Chi 393 No Psychosexual development Single mother
families
Golombok et al. (2003) 39 134 Various 393 No Socioemotional dev./relations Couples &
singles
Golombok and Rust
(1993)
N/A N/A N/A N/A N/A Reliability testing of a pre-school
gender inventory
Golombok and Tasker
(1996)
25 21 Pearson 783 Reported Sexual orientation Children of
single mothers
Golombok et al. (1997) 30 83 MANOVA 393 No. Parentchild interactions Couples &
singles
Green (1978) 37 0 Descript. N/A N/A Sexual identity None
Green et al. (1986) 50par; 56chl 40par; 48chl Various 390 No Sexual identity/social relations Single mothers
Harris and Turner
(1986)
23 16 ANOVA/Chi 393 No Sex roles/relationship with child Single moth. &
fath.
Hoeffer (1981) 20 20 ANOVA 393 No Sex-role behavior Single mothers
Huggins (1989) 18 18 T-test 393 No Self-esteem of adolescent
children
Divorced
mothers
Johnson and OConnor
(2002)
415 0 Various N/A No Parenting beliefs/division of
labor/etc.
None
King and Black (1999) N/A N/A F 393 N/A 338 College students
perceptions
N/A
Kirkpatrick et al. (1981) 20 20 Descript. N/A No Gender development Single mothers
Koepke et al. (1992) 47 couples 0 MANOVA N/A N/A Relationship quality None
Kweskin and Cook, 1982 22 22 Chi-Sqr 785 No Sex-role behavior Single mothers
Lewis, 1980 21 0 Qualitative N/A N/A Child response to m. disclosure None
Lott-Whitehead and
Tully, 1993
45 0 Descriptive N/A N/A Adult reports of impacts on
children
None
Lyons, 1983 43 37 Descriptive N/A No Adult self-reports Divorced
mothers
McLeod et al., 1999 0 0 Mult. regr. N/A No 151 College student reports N/A
Miller, 1979 54 0 Qualitative N/A N/A Father behavior & f-child bond None
Miller et al., 1981 34 47 Chi-Sqr 785 No Mother role/home environment Mothers
Morris et al., 2002 2431 0 MANCOVA N/A N/A Adult reports on coming out None
Mucklow and Phelan,
1979
34 47 Chi-Sqr 785 No Behavior and self-concept Married mothers
OConnell, 1993 11 0 Qualitative N/A N/A Social and sexual identity None
Pagelow, 1980 20 23 Qual/Descr. N/A N/A Problems and coping Single mothers
Patterson (1994) 66 0 T-test 393 No Social/behavioral/sexual identity Available norms
Patterson (1995) 52 0 T-test/Chi/F 393 No Division of labor/child
adjustment
None
(continued on next page)
L. Marks / Social Science Research 41 (2012) 735751 737
001015
1. All of [the fathers in the sample] were Caucasian (Bozett, 1980, p. 173).
2. Sixty parents, all of whom were White comprised the sample (Flaks et al., 1995, p. 107).
3. [All 40] mothers. . .were white (Hoeffer, 1981, p. 537).
4. All the children, mothers, and fathers in the sample were Caucasian (Huggins, 1989, p. 126).
5. The 25 women were all white (Rand et al., 1982, p. 29).
6. All of the women. . .[were] Caucasian (Siegenthaler and Bigner, 2000, p. 82).
7. All of the birth mothers and co-mothers were white (Tasker and Golombok, 1998, p. 52).
8. All [48] parents were Caucasian (Vanfraussen et al., 2003, p. 81).
Many of the other studies do not explicitly acknowledge all-White samples, but also do not mention or identify a single
minority participantwhile a dozen others report almost all-white samples.
18
Same-sex family researchers Lott-Whitehead
and Tully (1993) cautiously added in the discussion of their APA Brief-cited study:
Results from this study must be interpreted cautiously due to several factors. First, the study sample was small (N = 45)
and biased toward well-educated, white women with high incomes. These factors have plagued other [same-sex parent-
ing] studies, and remain a concern of researchers in this eld (p. 275).
Similarly, in connection with this bias, Patterson (1992), who would later serve as sole author of the 2005 APA Briefs
Summary of Research Findings on Lesbian and Gay Families, reported
19
:
Despite the diversity of gay and lesbian communities, both in the United States and abroad, samples of children [and par-
ents] have been relatively homogeneous. . .. Samples for which demographic information was reported have been
described as predominantly Caucasian, well-educated, and middle to upper class.
In spite of the privileged and homogeneous nature of the non-representative samples employed in the studies at that
time, Pattersons (1992) conclusion was as follows
20
:
Despite shortcomings [in the studies], however, results of existing research comparing children of gay or lesbian parents
with those of heterosexual parents are extraordinarily clear, and they merit attention. . . There is no evidence to suggest
that psychosocial development among children of gay men or lesbians is compromised in any respect relative to that
among offspring of heterosexual parents.
Table 1 (continued)
Author and year GayLes N Hetero N Stat used Cohen
N
Stat
power
Outcome studied Hetero compar
group
Patterson (2001) 66 0 Various 393 No Maternal mental health/child
adjustment
None
Patterson et al., 1998 66 0 Various 393 No Contact w/grandparents & adults None
Rand et al. (1982) 25 0 Correlations 783 No Mothers psychological health None
Sarantakos, 1996 58 116 F-test 393 N/A Childrens educational/social
outcomes
Married/non-
married
Siegenthaler and Bigner,
2000
25 26 T-test 393 No Mothers value of children Mothers
Steckel (1987) (Review) N/A N/A N/A No Psychosocial development of
children
None
Sullivan, 1996 34 couples 0 Qualitative N/A N/A Division of labor None
Tasker and Golombok,
1995
25 21 Pearson/T 783 No Psychosocial/sexual orientation Single mothers
Tasker and Golombok
(1997)
27 27 Various 393 Reported Psychological outcomes/family
rel.
Single mothers
Tasker and Golombok
(1998)
15 84 ANCOVA/
Chi
785 N/A Work and family life DI & NC couples
Vanfraussen et al.
(2003)
24 24 ANOVA 393 No Donor insemination/family
funct.
Families
Wainwright et al. (2004) 44 44 Various 393 No Psychosocial/school/romantic Couples
Wright (1998) 5 0 Qualitative N/A N/A Family issues/processes/
meaning
None
N/A = Not applicable (e.g., In connection with statistical power, qualitative studies and studies without heterosexual comparison groups are coded as N/A).
18
Examples of explicit or implicitly all-White (or nearly all-White) samples include, but are not limited to: Bigner andJacobsen (1989a,b), Bozett (1980), Flaks
et al. (1995), Green (1978), Green etal. (1986), Hoeffer (1981), Huggins (1989), Koepke et al. (1992), Rand et al. (1982), Siegenthaler and Bigner (2000), Tasker
and Golombok (1995, 1998), Vanfraussen et al. (2003).
19
Patterson (1992, p. 1029).
20
Patterson (1992, p. 1036) (emphasis added).
001016
Pattersons conclusion in a 2000 review was essentially the same
21
:
[C]entral results of existing research on lesbian and gay couples and families with children are exceptionally clear. . .. [The]
home environments provided by lesbian and gay parents are just as likely as those provided by heterosexual parents to
enable psychosocial growth among family members.
Although eight years had passed, in this second review, Patterson (2000) reported the continuing tendency of same-sex
parenting researchers to select privileged lesbian samples. Specically, she summarized, Much of the research [still] in-
volved small samples that are predominantly White, well-educated [and] middle-class (p. 1064).
22
Given the privileged,
homogeneous, and non-representative samples of lesbian mothers employed in much of the research, it seems warranted
to propose that Patterson was empirically premature to conclude that comparisons between gay or lesbian parents and het-
erosexual parents were extraordinarily clear
23
or exceptionally clear.
24
There is an additional point that warrants attention here. In Pattersons statements above, there are recurring references
to research on children of gay men/parents. In 2000, Demo and Cox reported that children living with gay fathers was a
rarely studied household conguration.
25
In 2005, how many of the 59 published studies cited in the APAs list of Empirical Stud-
ies Specically Related to Lesbian and Gay Parents and Their Children (pp. 2345) specically addressed the outcomes of children
from gay fathers? A closer examination reveals that only eight studies did so.
26
Of these eight studies, four did not include a het-
erosexual comparison group.
27
In three of the four remaining studies (with heterosexual comparison groups), the outcomes
studied were:
(1) the value of children to. . .fathers (Bigner and Jacobsen, 1989a, p. 163);
(2) parenting behaviors of. . .fathers (Bigner and Jacobsen, 1989b, p. 173);
(3) problems and relationship with child (Harris and Turner, 1986, pp. 1078).
The two Bigner and Jacobsen (1989a,b) studies focused on fathers reports of fathers values and behaviors, not on chil-
drens outcomesillustrating a recurring tendency in the same-sex parenting literature to focus on the parent rather than
the child. Harris and Turner (1986) addressed parentchild relationships, but their studys male heterosexual comparison
group was composed of two single fathers. Although several studies have examined aspects of gay fathers lives, none of
the studies comparing gay fathers and heterosexual comparison groups referenced in the APA Brief (pp. 2345) appear to
have specically focused on childrens developmental outcomes, with the exception of Sarantakos (1996), a study to which
we will later return.
In summary response to question 1 (How representative and culturally, ethnically, and economically diverse were the
gay/lesbian households in the published literature behind the APA Brief?), we see that in addition to relying primarily
on small, non-representative, convenience samples, many studies do not include any minority individuals or families. Fur-
ther, comparison studies on children of gay fathers are almost non-existent in the 2005 Brief. By their own reports, social
researchers examining same-sex parenting have repeatedly selected small, non-representative, homogeneous samples of
privileged lesbian mothers to represent all same-sex parents. This pattern across three decades of research raises signicant
questions regarding lack of representativeness and diversity in the same-sex parenting studies.
2.2. Question 2: how many studies of gay/lesbian parents had no heterosexual comparison group?
Of the 59 publications cited by the APA in the annotated bibliography section entitled Empirical Studies Specically
Related to Lesbian and Gay Parents and Their Children (pp. 2345), 33 included a heterosexual comparison group. In direct
response to question 2, 26 of the studies (44.1%) on same-sex parenting did not include a heterosexual comparison group. In
well-conducted science, it is important to have a clearly dened comparison group before drawing conclusions regarding
differences or the lack thereof. We see that nearly half of the Empirical Studies Specically Related to Lesbian and Gay Par-
ents and Their Children referenced in the APA Brief allowed no basis for comparison between these two groups (see Table
1). To proceed with precision, this fact does not negate the APA claim. It does, however, dilute it considerably as we are left
with not 59, but 33, relevant studies with heterosexual comparison groups.
2.3. Question 3: when heterosexual comparison groups were used, what were the more specic characteristics of those groups?
We now turn to a question regarding the nature of comparison samples. Of the 33 published Empirical Studies Specif-
ically Related to Lesbian and Gay Parents and Their Children (APA Brief, pp. 2345) that did directly include a heterosexual
21
Patterson (2000, , p. 1064) (emphasis added).
22
Patterson (2000, p. 1064).
23
Patterson (1992, p. 1036).
24
Patterson (2000, p. 1064).
25
Demo and Cox (2000, p. 890).
26
Bailey et al. (1995), Barrett and Tasker (2001), Bigner and Jacobsen (1989a,b), Bozett (1980), Harris and Turner (1986), Miller (1979), Sarantakos (1996).
27
Bailey et al. (1995), Barrett and Tasker (2001), Bozett (1980), Miller (1979).
001017
comparison group, what were the more specic characteristics of the groups that were compared? The earlier examination and
response related to question 1 documented that, by Pattersons reports, Despite the diversity of gay and lesbian communi-
ties. . .in the United States,
28
the repeatedly selected representatives of same-sex parents have been small samples [of lesbi-
ans] that are predominantly White, well-educated [and] middle-class (p. 1064).
29
In spite of homogeneous sampling, there is considerable diversity among gay and lesbian parents. Considerable diversity
exists among heterosexual parents as well. Indeed, the opening paragraph of the present article noted recurring differences
in several outcomes of societal concern for children in marriage-based intact families compared with children in cohabiting,
divorced, step, and single-parent families.
30
Many of the cited ndings are based on probability samples of thousands (see
Table 2).
Because children in marriage-based intact families have historically fared better than children in cohabiting, divorced,
step, or single-parent families on the above outcomes, the question of what groups researchers selected to represent het-
erosexual parents in the same-sex parenting studies becomes critical. A closer examination of the 33 published same-sex
parenting studies (APA Brief, pp. 2345) with comparison groups, listed chronologically, reveals that:
1. Pagelow (1980) used single mothers as a comparison group (p. 198).
2. Hoeffer (1981) used heterosexual single mothers (p. 537).
3. Kirkpatrick et al. (1981) used single, heterosexual mothers (p. 545).
4. Kweskin and Cook (1982) used women from Parents without Partners (p. 969).
Table 2
Brief overview of 15 intact/divorce/step/single family studies.
(N) Number of reported participants
Probability Is the study based on a probability sample?
Comp Grp Is a probability sample used as a comparison group?
Long Does the study employ measurements across time?
Key ! = Yes; X = No
(N) Probability Comp Grp Long
Amato (1991) 9643 ! ! !
Aquilino (1994) 4516 ! ! !
Brown (2004)
a
35,938 ! ! X
Chase-Lansdale et al. (1995)
b
17,414 ! ! !
Cherlin et al. (1998)
c
11,759 ! ! !
Ellis et al. (2003) 762 ! ! !
Harper and McLanahan (2004)
d
2846 ! ! !
Hetherington and Kelly (2002)
e
1400 ! ! !
Jekielek (1998) 1640 ! ! !
Lichter et al. (2003)
f
7665 ! ! X
Manning and Lamb (2003) 13,231 ! ! X
McLanahan and Sandefur (1994) (based on four data sets)
PSID
g
2900 ! ! !
NLSY
h
5246 ! ! !
HSBS
i
10,400 ! ! !
NSFH
j
13,017
k
! ! !
Mitchell et al. (2009)
l
4663 ! ! !
Nock (1998)
m
3604 ! ! !
Page and Stevens (2005)
n
2023 ! ! !
Total 148,667
a
National Survey of Americas Families (NSAF).
b
United Kingdom study and sample.
c
United Kingdom study and sample.
d
National Longitudinal Survey of Young Men and Women (NLSY).
e
Virginia Longitudinal Study (VLS).
f
National Survey of Family Growth (NSFG).
g
Panel Study of Income Dynamics (PSID).
h
i
The High School and Beyond Study (HSBS).
j
National Survey of Families and Households (NSFH).
k
This is the total original sample. The sub-sample is unlisted but is likely smaller.
l
National Longitudinal Study of Adolescent Health (Add Health).
m
n
Panel Study of Income Dynamics (PSID).
28
Patterson (1992, p. 1029).
29
Patterson (2000, p. 1064).
30
See Footnotes 210 for documentation.
001018
5. Lyons (1983) used heterosexual single mothers (p. 232).
6. Golombok et al. (1983) used single-parent households (p. 551).
7. Green et al. (1986) used solo parent heterosexual mothers (p. 175).
8. Harris and Turner (1986) used 2 male single parents and 14 female single parents (p. 105).
9. Huggins (1989) used divorced heterosexual mothers
31
(p. 123).
10. Tasker and Golombok (1995) used heterosexual single mothers (p. 203).
11. Tasker and Golombok (1997) used single heterosexual mothers (p. 38).
We see that in selecting heterosexual comparison groups for their studies, many same-sex parenting researchers have not
used marriage-based, intact families as heterosexual representatives, but have instead used single mothers (see Table 1).
Further, Bigner and Jacobsen used 90.9 percent single-father samples in two other studies (1989a, 1989b).
32
In total, in at
least 13 of the 33 comparison studies listed in the APA Briefs list of Empirical Studies (pp. 2345) that include heterosexual
comparison groups, the researchers explicitly sampled single parents as representatives for heterosexual parents. The re-
peated (and perhaps even modal) selection of single-parent families as a comparison heterosexual-parent group is noteworthy,
given that a Child Trends (2002) review has stated that children in single-parent families are more likely to have problems than
are children who live in intact families headed by two biological parents.
33
Given that at least 13 of the 33 comparison studies listed in the APA Briefs list of Empirical Studies (pp. 2345) used
single-parent families as heterosexual comparison groups, what group(s) did the remaining 20 studies use as heterosexual
representatives? In closely examining the 20 remaining published comparison group studies, it is difcult to formulate pre-
cise reports of the comparison group characteristics, because in many of these studies, the heterosexual comparison groups
are referred to as mothers or couples without appropriate specicity (see Table 1). Were these mothers continuously
marriedor were they single, divorced, remarried, or cohabiting? When couples were used, were they continuously mar-
riedor remarried or cohabiting? These failures to explicitly and precisely report sample characteristics (e.g., married or
cohabiting) are signicant in light of Browns (2004) nding based on her analysis of a data set of 35,938 US children and
their parents, that regardless of economic and parental resources, the outcomes of adolescents (1217 years old) in cohab-
iting families. . .are worse. . .than those. . .in two-biological-parent married families.
34
Because of the disparities noted by
Brown and others, scientic precision requires that we know whether researchers used: (a) single mothers, (b) cohabiting moth-
ers and couples, (c) remarried mothers, or (d) continuously married mothers and couples as heterosexual comparison groups.
Due to the ambiguity of the characteristics of the heterosexual samples in many same-sex parenting studies, let us frame
a question that permits a more precise response, namely: How many of the studies in the APA Briefs Empirical Studies section
(pp. 2345) explicitly compare the outcomes of children from intact, marriage-based families with those from same-sex families? In
an American Psychologist article published the year after the APA Brief, Herek (2006) referred to a large, national study by
McLanahan and Sandefur (1994) comparing the children of intact heterosexual families with children being raised by a sin-
gle parent. Herek then emphasized that this [large scale] research literature does not include studies comparing children
raised by two-parent same-sex couples with children raised by two-parent heterosexual couples.
35
Isolated exceptions exist
with relatively small samples (as discussed shortly in response to question 4 and as listed in Table 1), but they are rare.
Given what we have seen regarding heterosexual comparison group selection, let us revisit three related claims. First, in
1992, Patterson posited that
36
:
[N]ot a single study has found children of gay and lesbian parents to be disadvantaged in any respect relative to children
of heterosexual parents.
Pattersons (2000) claim was similar
37
:
[C]entral results of existing research on lesbian and gay couples and families with children are exceptionally clear. . ..
[The] home environments provided by lesbian and gay parents are just as likely as those provided by heterosexual par-
ents to enable psychosocial growth among family members.
Lastly, and most signicantly, we turn to the APA Briefs Summary of Research Findings on Lesbian and Gay Parenting,
also single-authored by Patterson (see p. 5)
38
:
Not a single study has found children of lesbian or gay parents to be disadvantaged in any signicant respect relative to
children of heterosexual parents.
31
Four of the 16 [divorced] heterosexual mothers were either remarried or currently living with a heterosexual lover (p. 127).
32
Of the 66 respondents, six were married, 48 were divorced, eight were separated, and four had never been married (Bigner and Jacobsen (1989a, p. 166).
This means the sample was 90.9% single.
33
Moore et al. (2002); for an extensive review, see Wilcox et al. (2011).
34
Brown (2004, p. 364) (emphasis added).
35
Herek (2006, p. 612).
36
37
38
Patterson, p. 15 (from APA Brief, 2005), (emphasis added).
001019
In all three of these claims (including that latter from the 2005 APA Brief), Patterson uses the broad and plural term het-
erosexual parents, a term that includes marriage-based, intact families. This broad claim is not nuanced by the information
that, with rare exceptions, the research does not include studies comparing children raised by two-parent, same-sex couples
with children raised by marriage-based, heterosexual couples. Further, no mention is made that in at least 13 of the 33 ex-
tant comparison studies referenced in the Brief (pp. 2345), the groups selected to represent heterosexual parents were
composed largely, if not solely, of single parents. We now move to another related examination of the APA Brief.
2.4. Question 4: does a scientically-viable study exist to contradict the conclusion that not a single study has found children of
lesbian or gay parents to be disadvantaged?
There is at least one notable exception
39
to the APAs claim that Not a single study has found children of lesbian or gay
parents to be disadvantaged in any signicant respect relative to children of heterosexual parents.
40
In the Summary of Find-
ings section, the APA Brief references a study by Sarantakos (1996),
41
but does so in a footnote that critiques the study (p. 6,
Footnote 1). On page 40 of the APA Briefs annotated bibliography, a reference to the Sarantakos (1996) article is offered, but
there is no summary of the studys ndings, only a note reading No abstract available.
Upon closer examination, we nd that the Sarantakos (1996) study is a comparative analysis of 58 children of heterosex-
ual married parents, 58 children of heterosexual cohabiting couples, and 58 children living with homosexual couples that
were all matched according to socially signicant criteria (e.g., age, number of children, education, occupation, and so-
cio-economic status).
42
The combined sample size (174) is the seventh-largest sample size of the 59 published studies listed
in the APA Briefs Summary of Research Findings on Lesbian and Gay Parenting (see Table 1). However, the six studies with
larger sample sizes were all adult self-report studies,
43
making the Sarantakos combined sample the largest study (APA Brief, pp.
2345) that examined childrens developmental outcomes.
Key ndings of the Sarantakos study are summarized below. To contextualize these data, the numbers are based on a tea-
cher rating-scale of performance ranging from 1 (very low performance), through 5 (moderate performance) to 9 (very high
performance).
44
Based on teacher (not parent) reports, Sarantakos found several signicant differences between married fam-
ilies and homosexual families.
45
Language Achievement Married 7.7, Cohabiting 6.8, Homosexual 5.5
Mathematics Achievement Married 7.9, Cohabiting 7.0, Homosexual 5.5
Social Studies Achievement Married 7.3, Cohabiting 7.0, Homosexual 7.6
Sport Interest/Involvement Married 8.9, Cohabiting 8.3, Homosexual 5.9
Sociability/Popularity Married 7.5, Cohabiting 6.5, Homosexual 5.0
School/Learning Attitude Married 7.5, Cohabiting 6.8, Homosexual 6.5
Parent-School Relationships Married 7.5, Cohabiting 6.0, Homosexual 5.0
Support with Homework Married 7.0, Cohabiting 6.5, Homosexual 5.5
Parental Aspirations Married 8.1, Cohabiting 7.4, Homosexual 6.5
a
a
Sarantakos, 1996, pp. 2427.
Sarantakos concluded, Overall, the study has shown that children of married couples are more likely to do well at school
in academic and social terms, than children of cohabiting and homosexual couples.
46
The APAs decision to de-emphasize the Sarantakos (1996) study was based, in part, on the criticism that nearly all indi-
cators of the childrens functioning were based on subjective reports by teachers.
47
The Sarantakos study was based, in part,
on teacher reports. However, teacher reports included tests and normal school assessment (p. 24). Subsequently, it may be
39
Other arguably contradictory studies are reviewed by Schumm (2011).
40
41
Among the diverse types of gay/lesbian parents there are at least two major categories that warrant scholarly precision: (a) two lesbian or gay parents
raising an adopted or DI (donor insemination) child from infancy with these and only these two parents; and (b) two lesbian or gay parents raising a child who
is the biological offspring of one of the parents, following a separation or divorce from a heterosexual partner. The Sarantakos sample is of the latter (b) type. In
terms of scholarly precision, it is important to differentiate and not draw strong implications from a to b or b to a. Indeed, the author would posit that
adopted versus DI children may also warrant separate consideration. The core issue is that precision is essential and overextension of ndings should be
avoided. This same issue is of serious concern in connection with the tendency to overextend ndings regarding lesbian mothers to apply to gay fathers (see
Regnerus, this volume).
42
Sarantakos (1996, p. 23).
43
In order, these six studies include: (1) Morris et al., 2002 (N = 2431), who addressed adults reports of coming out; (2) Johnson and OConnor (2002)
(N = 415), who addressed adults reports of parenting beliefs, division of labor, etc.; (3) Crawford et al. (1999) (N = 388), who addressed psychologists self-
reports of gay adoption; (4) King and Black (1999) (N = 338), who addressed college students perceptions of gay parents; (5) Bos et al. (2003) (N = 200), who
addressed parental motives and desires; and (6) Bos et al. (2004) (N = 200), who addressed parental reports of couple relations. These foci are not childrens
outcomes.
44
45
Social Studies Achievement is signicant at the p = .008 level; the eight other differences are signicant at the p = .000 level.
46
47
APA Brief (2005), Footnote 1, p. 6 (emphasis added).
001020
argued that Sarantakos decision not to rely solely or extensively on parent reports, as is done in most same-sex parenting stud-
ies, is a strength, given parents tendencies towards bias when reporting on their own children.
48
Sarantakos
49
also drew data
from school aptitude tests and observations, thereby modeling a research ideal of triangulation of sources.
50
In fact, the study
integrated not only three data sources to triangulate; it featured at least four (i.e., teachers, tests, observations, and child re-
ports). Further, the study controlled for education, occupation, and socio-economic status and then, based on teacher reports,
compared marriage-based families with gay/lesbian families and found nine signicant differenceswith children from mar-
riage-based families rating higher in eight areas. By objective standards, compared with the studies cited by the APA Brief,
the 1996 Sarantakos study was:
(a) The largest comparison study to examine childrens outcomes,
51
(b) One of the most comparative (only about ve other studies used three comparison groups),
52
and
(c) The most comprehensively triangulated study (four data sources) conducted on same-sex parenting.
53
Accordingly, this study deserves the attention of scientists interested in the question of homosexual and heterosexual
parenting, rather than the footnote it received.
As we conclude the examination of question 4, let us review a portion of APAs published negation of Sarantakos (1996)
study
54
:
[Children Australia, the journal where the article was published] cannot be considered a source upon which one should
rely for understanding the state of scientic knowledge in this eld, particularly when the results contradict those that
have been repeatedly replicated in studies published in better known scientic journals.
For other scientists, however, the salient point behind the Sarantakos ndings is that the novel comparison group of mar-
riage-based families introduced signicant differences in childrens outcomes (as opposed to the recurring no difference
nding with single-mother and couple samples). We now turn to the fth question.
2.5. Question 5: what types of outcomes have been investigated?
With respect to the APA Briefs claim that not a single study has found children of lesbian or gay parents to [have] dis-
advantaged [outcomes], what types of outcomes have been examined and investigated? Specically, how many of the same-
sex parenting studies in Table 1 address the societal concerns of intergenerational poverty, collegiate education and/or labor
force contribution, serious criminality, incarceration, early childbearing, drug/alcohol abuse, or suicide that are frequently
the foci of national studies on children, adolescents, and young adults, as discussed at the outset of this paper?
Anderssen and colleagues cataloged the foci of same-sex parenting studies in a 2002 review and reported
55
:
Emotional functioning was the most often studied outcome (12 studies), followed by sexual preference (nine studies),
gender role behavior (eight studies), behavioral adjustment (seven studies), gender identity (six studies), and cognitive
functioning (three studies).
Examination of the articles cited in the 2005 APA Brief on Lesbian and Gay Parenting yields a list of studied outcomes that
are consistent with Anderssens summary, including: sexual orientation
56
; behavioral adjustment, self-concepts, and
sex-role identity
57
; sexual identity
58
; sex-role behavior
59
; self-esteem
60
; psychosexual and psychiatric appraisal
61
;
socioemotional development
62
; and maternal mental health and child adjustment.
63
48
It is well replicated that individuals tend to rate the group with which they most identify more positively than they do other groups. This positive bias
includes within-family ratings Roese and Olson (2007).
49
Sarantakos is the author of several research methods textbooks (2005, 2007b) and the author/editor of a four-volume, 1672-page work in Sage Publications
Benchmarks in Social Research Series (2007a).
50
Triangulation is a means of checking the integrity of the inferences one draws. It can involve the use of multiple data sources, . . .multiple theoretical
perspectives, multiple methods, or all of these (Schwandt, 2001, p. 257). In effect, the standard of triangulation is advocacy for checks and balances.
51
Six of the 59 studies listed in the 2005 APA Brief (pp. 2345) had larger samples, but, as discussed earlier, they all focused on adult reports of adult
perceptions and outcomes.
52
For example, Brewaeys et al. (1997), Golombok et al. (2003, 1997), MacCallum and Golombok (2004), and Tasker and Golombok (1998).
53
In spite of the strong design with respect to triangulation, the Sarantakos study does not appear to be based on a true probability sample, nor is it or a large
sample (although it is a subsample of a 900-plus study). The study is rigorous by comparison to other same-sex parenting studies, but is limited compared with
most of the nationally representative studies on intact families listed in Table 2.
54
Patterson (2005) in APA Brief, p. 7, Footnote 1.
55
Anderssen et al. (2002, p. 343).
56
Bailey et al. (1995) and Golombok and Tasker (1996).
57
Patterson (1994).
58
Green (1978).
59
Hoeffer (1981) and Kweskin and Cook (1982).
60
Huggins (1989).
61
Golombok et al. (1983).
62
Golombok et al. (1997).
63
Patterson (2001).
001021
With these focal outcomes identied, it is noteworthy that all of the aforementioned outcomes of societal-level concern
are absent from the list of most often studied outcome(s) as identied by Anderssen et al.
64
In response to the present arti-
cles question 5 (what types of outcomes have been investigated for children of gay/lesbian families?), it may be concluded: In
the same-sex parenting research that undergirded the 2005 APA Brief, it appears that gender-related outcomes were the dom-
inant research concern. To be more precise, Table 1 lists several categories of information regarding the 59 published empirical
studies; one of these categories is the outcome studied. More than 20 studies have examined gender-related outcomes, but
there was a dearth of peer-reviewed journal articles from which to form science-based conclusions in myriad areas of societal
concern.
65
One book-length empirical study
66
entitled Same-Sex Couples (Sarantakos, 2000, Harvard Press) did examine several issues
of societal concern. In connection with the questions raised in the present article, this study:
(1) includes a diverse sample of lesbian and gay parents instead of focusing on privileged lesbian mothers (question 1);
(2) uses not only one but two heterosexual comparison samples; one married parent sample and one cohabitating parent
sample (questions 2 and 3);
(3) examines several outcomes of societal concern (question 5); and
(4) is unique in presenting long-term (post-18 years old) outcomes of children with lesbian and gay parents (question 6,
addressed later).
This studys conclusion regarding outcomes of gay and lesbian parents reads, in part:
If we perceive deviance in a general sense, to include excessive drinking, drug use, truancy, sexual deviance, and criminal
offenses, and if we rely on the statements made by adult children (over 18 years of age). . .[then] children of homosexual
parents report deviance in higher proportions than children of (married or cohabiting) heterosexual couples (Sarantakos,
2000, p. 131).
The 2005 APA Brief does not cite this study, again leaving us to more closely examine the claim that Not a single study
has found children of lesbian or gay parents to be disadvantaged in any signicant respect relative to children of heterosex-
ual parents (p. 15).
The Sarantakos (2000) study also includes the report that the number of children who were labeled by their parents as
gay, or identied themselves as gay, is much higher than the generally expected proportion (p. 133). However, the study
also notes areas of no signicant heterosexualhomosexual differences (i.e., Physical and emotional well-being, p. 130),
consistent with the 2005 APA Briefs claims. All of these ndings warranted attention in the 2005 APA Brief but were over-
looked. Of most interest to us here, however, is the novel attention of Sarantakos (2000) on multiple concerns of societal
importance, including drug and alcohol abuse, education (truancy), sexual activity, and criminality.
In any less-developed area of empirical inquiry it takes time, often several decades, before many of the central and most
relevant questions can be adequately addressed. This seems to be the case with same-sex parenting outcomes, as several
issues of societal concern were almost entirely unaddressed in the 2005 APA Brief.
2.6. Question 6: what do we know about the long-term outcomes of children of lesbian and gay parents?
In the preceding response to question 5, the outcomes of intergenerational poverty, criminality, college education and/or
labor force contribution, drug/alcohol abuse, suicide, early sexual activity, early childbearing, and eventual divorce as adults
were mentioned. Close consideration reveals that the majority of these outcomes are not child outcomes. Indeed, most of
these outcomes are not optimally observable until (at the earliest) mid-late adolescence or early adulthood (and in the case
of divorce, not until middle adulthood). As discussed in question 5, virtually none of the peer-reviewed, same-sex parenting
comparison studies addressed these outcomes.
67
Additionally, of the 59 published studies cited by the APA 2005 Brief (pp. 2345), it is difcult to nd comparison studies
of any kind that examine late adolescent outcomes of any kind. The few that utilize comparison groups have comparison
groups of 44 or fewer.
68
Let us further explore the importance of a lack of data centered on adolescents and young adults.
Table 2 identies 15 of the hundreds of available studies on outcomes of children from intact families (as contrasted with
comparison groups such as cohabiting couples and single parents). One of these studies included a data set of 35,938 chil-
drenone of the largest. . .nationally representative survey[s] of US children and their parents.
69
Based on analysis of this
64
65
Including: intergenerational poverty, criminality, college education and/or labor force contribution, drug/alcohol abuse, suicide, sexual activity and early
childbearing, and eventual divorce.
66
This study is a later, larger, and more detailed report on the earlier mentioned Sarantakos (1996) study. The sample of that study was larger than the other
comparison samples in Table 1.
67
Gartrell and colleagues (1999, 2000, 2005) have commenced to do so, but in 2005 they were reporting on children who were only 10 years old (with a
sample size of 74 and no heterosexual comparison group).
68
I.e. Wainwright et al. (2004).
69
Brown (2004), p. 355.
001022
nationally representative sample, Susan Brown emphasized, The ndings of this study. . .demonstrate the importance of sep-
arately examining children and adolescents. She then explained
70
:
Although the outcomes of children (611 years old) in cohabiting families. . .are worse. . .than those of children in two-
biological-parent married families, much of this difference. . .is economic. . .. In contrast, regardless of economic and
parental resources, the outcomes of adolescents (1217 years old) in cohabiting families. . .are worse. . .than those. . .in
two-biological-parent married families.
In short, in the case of cohabiting families and two-biological-parent married families the differences in childrens out-
comes increase in signicance as the children grow older. The likelihood of signicant differences arising between children
from same-sex and married families may also increase across timenot just into adolescence but into early and middle
adulthood. For example, research indicates that [d]aughters raised outside of intact marriages are. . .more likely to end
up young, unwed mothers than are children whose parents married and stayed married, and that [p]arental divorce in-
creases the odds that adult children will also divorce.
71
Longitudinal studies that follow children across time and into adulthood to examine such outcomes are comparatively
rare and valuable. We briey turn to a key nding from one such study that followed children of divorce into middle adult-
hood. Based on a 25-year longitudinal study, Wallerstein and colleagues (2001) state:
Contrary to what we have long thought, the major impact of divorce does not occur during childhood or adolescence.
Rather, it rises in adulthood as serious romantic relationships move center stage. When it comes time to choose a life
mate and build a new family, the effects of divorce crescendo (p. xxix).
Wallersteins research, like nearly all of the studies in the same-sex parenting literature, is based on a small, non-repre-
sentative sample that should not be generalized or overextended. Her longitudinal work does, however, indicate that effects
[can] crescendo in adulthood. Did any published same-sex parenting study cited by the 2005 APA Brief (pp. 2345) track the
societally signicant long-term outcomes into adulthood? No. Is it possible that the major impact of same-sex parenting
might not occur during childhood or adolescence. . .[but that it will rise] in adulthood as serious romantic relationships
move center stage? Is it also possible that when it comes time to choose a life mate and build a new family that the effects
of same-sex parenting will similarly crescendo as they did in Wallersteins study of divorce effects? In response to this or
any question regarding the long-term, adult outcomes of lesbian and gay parenting we have almost no empirical basis for
responding. An exception is provided by the ndings from self-reports of adult children (18 + years of age) in Sarantakos
(2000) book-length study, but those results not encouraging. This is a single study howevera study that, like those cited by
the APA Brief, lacks the statistical power and rigor of the large, random, representative samples used in marriage-based fam-
ily studies (see Table 2). We now move to a nal related empirical question regarding the same-sex parenting literature.
2.7. Question 7: have the studies in this area committed the type II error and prematurely concluded that heterosexual couples and
gay and lesbian couples produce parental outcomes with no differences?
The Summary of Research Findings in the APA brief reads, As is true in any area of research, questions have been raised
with regard to sampling issues, statistical power, and other technical matters (p. 5). However, neither statistical power nor
the related concern of Type II error is further explained or addressed. This will be done next.
In social science research, questions are typically framed as follows: Are we 95% sure the two groups being compared are
different? (p < .05). If our statistics seem to conrm a difference with 95% or greater condence, then we say the two groups
are signicantly different. But what if, after statistical analysis, we are only 85% sure that the two groups are different? By
the rules of standard social science, we would be obligated to say we were unable to satisfactorily conclude that the two
groups are different. However, a reported nding of no statistically signicant difference (at the p < .05 level; 95%-plus cer-
tainty) is a grossly inadequate basis upon which to offer the science-based claim that the groups were conclusively the
same. In research, incorrectly concluding that there is no difference between groups when there is in fact a difference is
referred to as a Type II error. A Type II error is more likely when undue amounts of random variation are present in a study.
Specically, small sample size, unreliable measures, imprecise research methodology, or unaccounted for variables can all
increase the likelihood of a Type II error. All one would have to do to be able to come to a conclusion of no difference
is to conduct a study with a small sample and/or sufcient levels of random variation. These weaknesses compromise a
studys statistical power (Cohen, 1988).
It must be re-emphasized that a conclusion of no signicant difference means that it is unknown whether or not a dif-
ference exists on the variable(s) in question (Cohen, 1988). This conclusion does not necessarily mean that the two groups
are, in fact, the same on the variable being studied, much less on all other characteristics. This point is important with same-
sex parenting research because Patterson (1992, 2000) and the 2005 APA Brief seem to drawinferences of sameness based on
the observation that gay and lesbian parents and heterosexual parents appear not to be statistically different from one an-
other based on small, non-representative samplesthereby becoming vulnerable to a classic Type II error.
70
Brown (2004), p. 364.
71
Wilcox et al. (2011), p. 11.
001023
To make the APA Briefs proposition of sameness more precarious, in a review published one year after the APA Brief in
the agship APA journal, American Psychologist, Herek (2006) acknowledged that many same-sex parenting studies have
utilized small, select convenience samples and often employed unstandardized measures.
72
Anderssen et al. (2002) simi-
larly indicated in their review of same-sex parenting studies, The samples were most often small, increasing the chance to con-
clude that no differences exist between groups when in fact the differences do exist. This casts doubt on the external validity of
the studies.
73
With these limitations noted, the 2005 APA Brief explicitly claimed that ndings of non-signicant differences
between same-sex and heterosexual parents had been repeatedly replicated (p. 7, Footnote 1).
Reasons for skepticism regarding the APA Briefs claim that ndings have been repeatedly replicated rest in Neumans
(1997) point that the logic of replication implies that different researchers are unlikely to make the same errors.
74
How-
ever, if errors (e.g., similarly biased sampling approaches employing small, select convenience samples
75
and comparison
groups) are repeated by different researchers, the logic behind replication is undermined. As has been previously detailed in
the response to question 1 in this article, same-sex parenting researchers have repeatedly selected White, well-educated, mid-
dle- and upper-class lesbians to represent same-sex parents. This tendency recurred even after this bias was explicitly identied
by Patterson (1992, 2000).
76
Further, repeated sampling tendencies in connection with heterosexual comparison groups (e.g.,
single mothers), were documented in response to Question 3 in this paper. These repeated (convenience) sampling tendencies
across studies that employed different measures do not seem to constitute valid scientic replication.
An additional scientic question raised by the above information regarding small, select convenience
77
samples is
framed by Stacey and Biblarz (2001) who reveal that many of these [comparative same-sex parenting] studies use conventional
levels of signicance. . .on miniscule samples, substantially increasing their likelihood of failing to reject the null hypothesis.
78
Was the APAs claim that Not a single study has found children of lesbian or gay parents to be disadvantaged. . .
79
based on
clear scientic evidence or (perhaps) Type II errors? In response, we now turn to the APA-acknowledged but unexplained cri-
tique of low statistical power in these studies (p. 5).
The last three editions of the APA Publication manual (1994, 2001, 2010) have urged scholars to report effect sizes and to
take statistical power into consideration when reporting their results. The APA 5th Publication manual (2001) in use at the
time of APAs 2005 Brief on Lesbian and Gay Parenting stated:
Take seriously the statistical power considerations associated with your tests of hypotheses. Such considerations relate to
the likelihood of correctly rejecting the tested hypotheses, given a particular alpha level, effect size, and sample size. In
that regard, you should routinely provide evidence that your study has power to detect effects of substantive interest
(e.g., see Cohen, 1988). You should be similarly aware of the role played by sample size in cases in which not rejecting
the null hypothesis is desirable (i.e., when you wish to argue that there are no differences [between two groups]). . .
(p. 24).
This awareness of statistical power in cases when you wish to argue that there are no differences bears directly on
same-sex comparative research. The APA 5th Publication manual (2001) continues:
Neither of the two types of probability [alpha level or p value] directly reects the magnitude of an effect or the strength
of a relationship. For the reader to fully understand the importance of your ndings, it is almost always necessary to
include some index of effect size or strength of relationship in your Results section (p. 25).
Let us review three statements from the APA 5th Publication Manual for emphasis:
(1) The APA urges researchers to: Take seriously the statistical power considerations and routinely provide evidence
(p. 24).
(2) The APA identies a specic concern with sample size and statistical power in connection with cases where authors
wish to argue that there are no differences between compared groups (p. 24).
(3) The APA concludes: It is almost always necessary to include some index of effect size or strength of relationship in
your Results section (p. 25).
The APAs rst highlighted exhortation is that an author should routinely provide evidence that your study has sufcient
power. . .(e.g., see Cohen, 1988) (p. 24). The reference cited here by the APA is the volume Statistical Power Analysis for the
Behavioral Sciences (2nd ed.) by the late psychometrician Jacob Cohen, who has been credited with foundational work in sta-
tistical meta-analysis (Borenstein, 1999). In his APA-cited volume, Cohen states:
72
Herek (2006), p. 612.
73
Anderssen et al. (2002), p. 348.
74
Neuman (1997), p. 150.
75
Herek (2006), p. 612.
76
Further, single mothers have been repeatedly selected to represent heterosexual parents as documented in this papers response to question 3.
77
Herek (2006), p. 612.
78
Stacey and Biblarz (2001, p. 168), Footnote 9.
79
001024
Most psychologists of whatever stripe believe that samples, even small samples, mirror the characteristics of their parent
populations. In effect, they operate on the unstated premise that the law of large numbers holds for small numbers as
well. . .. [Citing Tversky and Kahneman] The believer in the law of small numbers has incorrect intuitions about signif-
icance level, power, and condence intervals. Signicance levels are usually computed and reported, but power and con-
dence levels are not. Perhaps they should be.
But as we have seen, too many of our colleagues have not responded to [this] admonition. . .. They do so at their
peril (p. xv).
Let us contextualize the law of small numbers with respect to the same-sex parenting studies cited in the APA Brief. The
combined non-representative sample total of all 59 same-sex parenting studies in the 2005 APA Brief (pp. 2345) is 7800
80
(see Table 1). By comparison, Table 2 lists 15 prominent studies that contrast childrens outcomes in intact, single-parent, di-
vorced, and/or step-family forms using large probability samples and comparison groups.
81
The average sample size in these
studies is 9911
82
a gure larger than all 59 same-sex parenting studies combined (7800).
We now turn to another question relating to Cohens statements: How many of the published same-sex parenting studies
with a heterosexual comparison group cited in APAs Brief (pp. 2345) provide[d] evidence of statistical power, consistent
with APAs Publication Manual and the admonition of Jacob Cohen who is cited in the APA manual? An examination of the
studies indicates that only four of the 59 did so.
83
In addition to Cohens (1988) statement that statistical power is ignored at our own peril, he offered several tables in his
volume for researchers to reference. Employing these tables, statistical experts Lerner and Nagai (2001) computed the sam-
ple sizes required for a power level of .80, or a Type II error rate of .20, or one in ve ndings (p. 102). At this power level,
the minimum number of cases required to detect a small effect size
84
is 393 for a T-test or ANOVA, or 780-plus for Chi-Square
or Pearson Correlation Coefcient tests.
85
In Table 1 of this report, the 59 published same-sex parenting studies cited in the APA
Brief (pp. 2345) are compared against these standards. A close examination indicates that not a single study, including the few
that reported power, meets the standards needed to detect a small effect size. Indeed, it appears that only two of the comparison
studies (Bos et al., 2003, 2004) have combined sample sizes of even half of the minimum number of cases.
86
In their book-length examination of same-sex parenting studies, Lerner and Nagai (2001) further indicate that 17 of the
22 same-sex parenting comparison studies they reviewed had been designed in such a way that the odds of failing to nd a
signicant difference [between homo- and hetero-sexual groups] was 85% or higher.
87
Indeed, only one of the 22 studies they
analyzed revealed a probability of Type II error below 77 percent, and that study did nd differences.
88
These methodological
concerns (and others) were raised and explained in Lerner and Nagais monograph (see pp. 95108), and in an 81-page report by
Nock (2001) preceding the APA Brief.
89
Nock concluded:
All of the [same-sex parenting] articles I reviewed contained at least one fatal aw of design or execution. Not a single one
was conducted according to generally accepted standards of scientic research. . .. [I]n my opinion, the only acceptable
conclusion at this point is that the literature on this topic does not constitute a solid body of scientic evidence (Nock,
2001, pp. 39, 47).
80
This gure (7800) includes same-sex parents and their children, as well as heterosexual comparison samples (1404), psychologists (388), and college
students perception reports (489).
81
Table 2 lists 15 studies that contrast childrens outcomes in intact families compared with other family forms using large, probability samples and
comparison groups. The focal topics of these studies are not sexual preference, gender role behavior. . .[and] gender identity (Anderssen et al., 2002, p. 343),
but outcomes such as educational attainment, labor force attachment, and early childbearing (McLanahan and Sandefur, 1994, pp. 2021 ), as
recommended in the earlier examination of question 5. Further, all but two of the 15 studies employ longitudinal designs, as recommended in the earlier
examination of question 6.
82
This gure is the result of 148,667 divided by 15 studies.
83
These include Chan et al. (1998b), Fulcher et al. (2002), Golombok and Tasker (1996), and Tasker and Golombok (1997).
84
By way of context, in a 67 study meta-analysis of the average differences in outcomes between children with divorced and continuously marriedparents,
Amato (2001) reported an average weighted effect size of between 0.12 and 0.22 (a 0.17 average) with an advantage in all ve domains considered to
children of continuously married parents (p. 360). These effect sizes of about .20, although statistically robust, would be classied by Cohen (1992) as small
effect sizes. Even so, based on the data, most family scholars would agree that children whose parents remain continuously married tend to fare slightly to
moderately better than when parents divorce. However, large numbers were needed to determine this small but important effect. Indeed, most effect sizes in
social science research tend to be small. Rigorous and sound social science tends to include and account for many inuential factors that each has a small but
meaningful effect size. In social science, detecting a novel large effect from a single variable (whether it is divorce, remarriage, or same-sex parenting), is a
comparatively rare occurrence. If we are to examine possible effects of same-sex parenting with scientic precision and rigor, related examinations would, like
Amatos work, be designed and rened to detect small effect sizes.
85
Cohen (1988) proposes a relatively high power of .90 for cases where one is trying to demonstrate the r [difference] is trivially small (p. 104). If the .90
power were applied, the required sample sizes would further increase. However, because none of the studies in Table 1 of the present report approach the .80
power levels, .90 calculations are unnecessary here.
86
The minimum number of cases is 393. The two Bos et al. studies both have combined samples of 200. Four other larger samples are not comparison
studies Crawford et al. (1999), Johnson and OConnor (2002), King and Black (1999), and Morris et al. (2002).
87
Lerner and Nagai (2001, p. 103).
88
The single exception was Cameron and Cameron (1996) with a comparatively low probability error rate of 25%. This study, like the Sarantakos (1996) study
mentioned earlier, did report some signicant differences between children of heterosexual and homosexual parents but, like Sarantakos (1996), was not
addressed in the body of the 2005 APA brief but was instead moved to a footnote on p. 7. See Redding (2008) for additional discussion (p. 137).
89
For similar critiques preceding the 2005 APA brief, seeNock (2001), Schumm (2004), Wardle (1997), and Williams (2000). For similar critiques post-dating
the 2005 APA brief, see Byrd (2008), Schumm (2010a,b, 2011), and Redding (2008, p. 138).
001025
More specically, Nock identied: (a) several aws related to sampling (including biased sampling, non-probability sam-
pling, convenience sampling, etc.); (b) poorly operationalized denitions; (c) researcher bias; (d) lack of longitudinal studies;
(e) failure to report reliability; (f) low response rates; and (g) lack of statistical power (pp. 3940).
90
Although some of these
aws are briey mentioned in the 2005 APA Summary of Research Findings on Lesbian and Gay Parenting, many of the signif-
icant concerns raised by Nock or Lerner and Nagai are not substantively addressed.
91
Indeed, the Lerner and Nagai volume and
the Nock report are neither mentioned nor referenced.
To restate, in connection with the APAs published urging that researchers: Take seriously the statistical power consid-
erations and routinely provide evidence, the academic reader is left at a disadvantage.
92
Only a few comparison studies
specically reported statistical power at all and no comparison study approached the minimum sample size of 393 needed
to nd a small effect.
The authors response to question 7 has examined how comparisons have been made from a research methods stand-
point. In summary, some same-sex parenting researchers have acknowledged that miniscule samples
93
signicantly in-
crease the chance to conclude that no differences exist between groups when in fact the differences do existthereby
casting doubt on the external validity of the studies.
94
An additional concern is that the APA Briefs claim of repeatedly rep-
licated ndings of no signicant difference rested almost entirely on studies that were published without reports of the APA-
urged effect sizes and statistical power analyses.
95
This inconsistency seems to justify scientic skepticism, as well as the effort
of more closely assessing the balance, precision, and rigor behind the conclusions posed in the 2005 APA Brief.
3. Conclusion
The 2005 APA Brief, near its outset, claims that even taking into account all the questions and/or limitations that may
characterize research in this area, none of the published research suggests conclusions different from that which will be
summarized (p. 5). The concluding summary later claims, Indeed, the evidence to date suggests that home environments
provided by lesbian and gay parents are as likely as those provided by heterosexual parents to support and enable childrens
psychosocial growth (p. 15).
96
We now return to the overarching question of this paper: Are we witnessing the emergence of a new family form that
provides a context for children that is equivalent to the traditional marriage-based family? Even after an extensive reading
of the same-sex parenting literature, the author cannot offer a high condence, data-based yes or no response to this
question. To restate, not one of the 59 studies referenced in the 2005 APA Brief (pp. 2345; see Table 1) compares a large,
random, representative sample of lesbian or gay parents and their children with a large, random, representative sample of
married parents and their children. The available data, which are drawn primarily from small convenience samples, are
insufcient to support a strong generalizable claim either way. Such a statement would not be grounded in science. To make
a generalizable claim, representative, large-sample studies are neededmany of them (e.g., Table 2).
Some opponents of same-sex parenting have made egregious overstatements
97
disparaging gay and lesbian parents.
Conversely, some same-sex parenting researchers seem to have contended for an exceptionally clear
98
verdict of no differ-
ence between same-sex and heterosexual parents since 1992. However, a closer examination leads to the conclusion that
strong, generalized assertions, including those made by the APA Brief, were not empirically warranted.
99
As noted by Shiller
(2007) in American Psychologist, the line between science and advocacy appears blurred (p. 712).
The scientic conclusions in this domain will increase in validity as researchers: (a) move from small convenience sam-
ples to large representative samples; (b) increasingly examine critical societal and economic concerns that emerge during
adolescence and adulthood; (c) include more diverse same-sex families (e.g., gay fathers, racial minorities, and those without
middle-high socioeconomic status); (d) include intact, marriage-based heterosexual families as comparison groups; and (e)
90
Four of these seven issues are addressed in the present paper. The exceptions include researcher bias, failure to report reliability, and low response rates.
91
The 2005 APA Briefs Summary on Research Findings acknowledges criticisms of same-sex parenting research including: (a) non-representative sampling,
(b) poorly matched or no control groups, (c) well-educated, middle class [lesbian] families, and (d) relatively small samples (p. 5). The respective
responses to these criticisms in the APA brief are: (a) contemporary research on children of lesbian and gay parents involves a wider array of sampling
techniques than did earlier studies; (b) contemporary research on children of lesbian and gay parents involves a wider array of research designs (and hence,
control groups) than did earlier studies; (c) contemporary research on children of lesbian and gay parents involves a greater diversity of families than did
earlier studies; and (d) contemporary research has beneted from such criticisms (p. 5). The APA Brief does not challenge the validity of these research
criticisms but notes that improvements are being made.
92
See Schumm (2010b) for more comprehensive, article-length treatment of these statistical issues.
93
Stacey and Biblarz (2001, p. 168).
94
95
Schumm (2010b).
96
The APA Brief also states that the existing data are still limited, and any conclusions must be seen as tentative. Also, that it should be acknowledged that
research on lesbian and gay parents and their children, though no longer new, is still limited in extent (p. 15). For some scientists, these salient points seem to
be overridden by the APA Briefs conclusions.
97
This reality has been disapprovingly documented by Shiller (2007).
98
Patterson (1992).
99
In 2006, the year following APAs release of the brief on Lesbian and Gay Parenting, former APA president Nicholas Cummings argued that there has been
signicant erosion of the APAs established principle (Shiller (2007), p. 712). . .that when we speak as psychologists we speak from research evidence and
clinical experience and expertise (Cummings (2006), p. 2).
001026
constructively respond to criticisms from methodological experts.
100
Specically, it is vital that critiques regarding sample
size, sampling strategy, statistical power, and effect sizes not be disregarded. Taking these steps will help produce more meth-
odologically rigorous and scientically informed responses to signicant questions affecting families and children.
References
Akerlof, G., 1998. Men without children. Economic Journal 108, 287309.
Amato, P., 1991. Parental absence during childhood and depression in later life. Sociological Quarterly 32, 543556.
Amato, P., 2001. Children of divorce in the 1990s: an update of the Amoato and Keith (1991) meta-analysis. Journal of Family Psychology 15, 355370.
Amato, P., 2005. The impact of family formation change on the cognitive, social, and emotional well-being of the next generation. The Future of Children 15,
7596.
Amato, P., Booth, A., 2000. A Generation at Risk: Growing Up in an Era of Family Upheaval. Harvard University Press, Cambridge, MA.
Anderssen, N., Amlie, C., Ytteroy, E.A., 2002. Outcomes for children with lesbian or gay parents: a review of studies from 1978 to 2000. Scandinavian Journal
of Psychology 43, 335351.
Publication manual of the American Psychological Association (forth ed.), 1994. APA, Washington, DC.
Publication manual of the American Psychological Association (fth ed.), 2001. APA, Washington, DC.
Publication manual of the American Psychological Association (sixth ed.), 2010. APA, Washington, DC.
Aquilino, W.S., 1994. Impact of childhood family disruption on young adults relationships with parents. Journal of Marriage and the Family 56, 295313.
Bachman, J.G. et al, 1997. Smoking, Drinking and Drug Abuse in Young Adulthood. Erlbaum, Mahwah, NJ.
Bailey, J.M., Bobrow, D., Wolfe, M., Mikach, S., 1995. Sexual orientation of adult sons of gay fathers. Developmental Psychology 31, 124129.
Barrett, H., Tasker, F., 2001. Growing up with a gay parent: views of 101 gay fathers on their sons and daughters experiences. Educational and Child
Battle, J., 1998. What beats having two parents?: educational outcomes for AfricanAmerican students in single- versus dual-parent families. Journal of
Black Studies 28, 783801.
Bigner, J.J., Jacobsen, R.B., 1989a. The value of children to gay and heterosexual fathers. Journal of Homosexuality 19, 163172.
Bigner, J.J., Jacobsen, R.B., 1989b. Parenting behaviors of homosexual and heterosexual fathers. Journal of Homosexuality 19, 173186.
Blackmon, L., Clayton, O., Glenn, N., Malone-Colon, L., Roberts, A., 2005. The Consequences of Marriage for African Americans: A Comprehensive Literature
Review. Institute for American Values, New York.
Borenstein, M., 1999. Jacob Cohen, PhD, 19231998. Archives of General Psychiatry 56, 581.
Bos, H.M.W., van Balen, F., van den Boom, D.C., 2003. Planned lesbian families: their desire and motivation to have children. Human Reproduction 10, 2216
2224.
Bos, H.M.W., van Balen, F., van den Boom, D.C., 2004. Experience of parenthood, couple relationship, social support, and child-rearing goals in planned
lesbian mother families. Journal of Child Psychology and Psychiatry 45, 755764.
Bozett, F.W., 1980. Gay fathers: how and why they disclose their homosexuality to their children. Family Relations 29, 173179.
Brewaeys, A., Ponjaert, I., Van Hall, E.V., Golombok, S., 1997. Donor insemination: child development and family functioning in lesbian mother families.
Human Reproduction 12, 13491359.
Brown, S.L., 2004. Family structure and child well-being: the signicance of parental cohabitation. Journal of Marriage and Family 66, 351367.
Byrd, A.D., 2008. Conjugal marriage fosters healthy human and societal development. In: Wardle, L. (Ed.), Whats the Harm? Does Legalizing Same-Sex
Marriage Really Harm Individuals, Families or Society? University Press, Lanham, MD, pp. 326.
Cameron, P., Cameron, K., 1996. Homosexual parents. Adolescence 31, 757776.
Chan, R.W., Brooks, R.C., Raboy, B., Patterson, C.J., 1998a. Division of labor among lesbian and heterosexual parents: associations with childrens adjustment.
Journal of Family Psychology 12, 402419.
Chan, R.W., Raboy, B., Patterson, C.J., 1998b. Psychosocial adjustment among children conceived via donor insemination by lesbian and heterosexual
mothers. Child Development 69, 443457.
Chase-Lansdale, P.L., Cherlin, A.J., Kiernan, K.K., 1995. The long-term effects of parental divorce on the mental health of young adults: a developmental
perspective. Child Development 66, 16141634.
Cherlin, A.J. et al, 1995. Parental divorce in childhood and demographic outcomes in young adulthood. Demography 32, 299318.
Cherlin, A.J., Chase-Lansdale, P.L., McRae, C., 1998. Effects of parental divorce on mental health throughout the life course. American Sociological Review 63,
239249.
Ciano-Boyce, C., Shelley-Sireci, L., 2002. Who is mommy tonight? Lesbian parenting issues. Journal of Homosexuality 43, 113.
Cohen, J., 1988. Statistical Power Analysis for the Behavioral Sciences, second ed. Erlbaum, Hillsdale, NJ.
Cohen, J., 1992. A power primer. Psychological Bulletin 112, 155159.
Crawford, I., McLeod, A., Zamboni, B.D., Jordan, M.B., 1999. Psychologists attitudes toward gay and lesbian parenting. Professional Psychology: Research and
Practice 30, 394401.
Cummings, N.A., 2006. The APA and psychology need reform. Paper presented at the annual convention of the American Psychological Association (August
12). New Orleans, LA.
Cutler, D.M. et al., 2000. Explaining the rise in youth suicide. Working Paper 7713. National Bureau of Economic Research, Cambridge.
Demo, D.H., Cox, M.J., 2000. Families with young children: a review of research in the 1990s. Journal of Marriage and the Family 62, 876895.
Ellis, B.J. et al, 2003. Does father absence place daughters at special risk for early sexual activity and teenage pregnancy? Child Development 74, 801821.
Flaks, D., Ficher, I., Masterpasqua, F., Joseph, G., 1995. Lesbians choosing motherhood: a comparative study of lesbian and heterosexual parents and their
children. Developmental Psychology 31, 104114.
Flewelling, R.L., Bauman, K.E., 1990. Family structure as a predictor of initial substance use and sexual intercourse in early adolescence. Journal of Marriage
and the Family 52, 171181.
Fulcher, M., Chan, R.W., Raboy, B., Patterson, C.J., 2002. Contact with grandparents among children conceived via donor insemination by lesbian and
heterosexual mothers. Parenting: Science and Practice 2, 6176.
Gartrell, N., Hamilton, J., Banks, A., Mosbacher, D., Reed, N., Sparks, C.H., Bishop, H., 1996. The national lesbian family study: 1. Interviews with prospective
mothers. American Journal of Orthopsychiatry 66, 272281.
Gartrell, N., Banks, A., Hamilton, J., Reed, N., Bishop, H., Rodas, C., 1999. The national lesbian family study: II. Interviews with mothers of toddlers. American
Journal of Orthopsychiatry 69, 362369.
Gartrell, N., Banks, A., Reed, N., Hamilton, J., Rodas, C., Deck, A., 2000. The national lesbian family study: III. Interviews with mothers of ve-year olds.
American Journal of Orthopsychiatry 70, 542548.
Gartrell, N., Deck, A., Rodas, C., Peyser, H., Banks, A., 2005. The national lesbian family study: 4. Interviews with the 10-year old children. American Journal of
Orthopsychiatry 75, 518524.
Gaudino, J.A. et al, 1999. No fathers names: a risk factor for infant mortality in the state of Georgia. Social Science and Medicine 48, 253265.
100
At least one such study (Rosenfeld, 2010) has emerged in the years since the 2005 APA brief was issued. This study features a very large sample but has also
received criticism (Schumm, 2011).
001027
Gershon, T.D., Tschann, J.M., Jemerin, J.M., 1999. Stigmatization, self-esteem, and coping among the adolescent children of lesbian mothers. Journal of
Adolescent Health 24, 437445.
Glenn, N.D., 1989. What we know, what we say we know: discrepancies between warranted and stated conclusions in the social sciences. In: Eulau, H. (Ed.),
Crossroads of Social Science. The ICPSR 25th Anniversary Volume. Agathon Press, New York, pp. 119140.
Glenn, N.D., 1997. A reconsideration of the effect of no-fault divorce on divorce rates. Journal of Marriage and the Family 59, 10231025.
Golombok, S., Rust, J., 1993. The pre-school activities inventory: a standardized assessment of gender role in children. Psychological Assessment 5, 131136.
Golombok, S., Tasker, F., 1996. Do parents inuence the sexual orientation of their children? Findings from a longitudinal study of lesbian families.
Developmental Psychology 32, 311.
Golombok, S., Spencer, A., Rutter, M., 1983. Children in lesbian and single-parent households: psychosexual and psychiatric appraisal. Journal of Child
Psychology and Psychiatry 24, 551572.
Golombok, S., Tasker, F., Murray, C., 1997. Children raised in fatherless families from infancy: family relationships and the socioemotional development of
children of lesbian and single heterosexual mothers. Journal of Child Psychology and Psychiatry 38, 783791.
Golombok, S., Perry, B., Burston, A., Murray, C., Mooney-Somers, J., Stevens, M., Golding, J., 2003. Children with lesbian parents: a community study.
Developmental Psychology 39, 2033.
Green, R., 1978. Sexual identity of 37 children raised by homosexual or transsexual parents. American Journal of Psychiatry 135, 692697.
Green, R., Mandel, J.B., Hotvedt, M.E., Gray, J., Smith, L., 1986. Lesbian mothers and their children: a comparison with solo parent heterosexual mothers and
their children. Archives of Sexual Behavior 7, 175181.
Harper, C., McLanahan, S., 2004. Father absence and youth incarceration. Journal of Research on Adolescence 14, 369397.
Harris, M.B., Turner, P.H., 1986. Gay and lesbian parents. Journal of Homosexuality 12, 101113.
Heiss, J., 1996. Effects of African American family structure on school attitude and performance. Social Problems 43, 246267.
Herek, G.M., 2006. Legal recognition of same-sex relationships in the United States: a social science perspective. American Psychologist 61, 607621.
Hetherington, M., Kelly, J., 2002. For Better or for Worse: Divorce Reconsidered. Norton, New York.
Hoeffer, B., 1981. Childrens acquisition of sex-role behavior in lesbian mother families. American Journal of Orthopsychiatry 5, 536544.
Horwitz, A.V., White, H.R., Howell-White, S., 1996. Becoming married and mental health: a longitudinal study of a cohort of young adults. Journal of
Marriage and the Family 58, 895907.
Huggins, S.L., 1989. A comparative study of self-esteem of adolescent children of divorced lesbian mothers and divorced heterosexual mothers. Journal of
Homosexuality 18, 123135.
Jekielek, S., 1998. Parental conict, marital disruption, and childrens emotional well-being. Social Forces 76, 905936.
Johnson, S.M., OConnor, E., 2002. The Gay Baby Boom: The Psychology of Gay Parenthood. NYU Press, New York.
Johnson, R.A. et al, 1996. The Relationship Between Family Structure and Adolescent Substance Abuse. US Dept. of Health and Human Services, Rockville,
MD.
Kamark, E.C., Galston, W.A., 1990. Putting Children First: A Progressive Family Policy for the 1990s. Progressive Policy Institute, Washington, DC.
King, B.R., Black, K.N., 1999. College students perceptual stigmatization of the children of lesbian mothers. American Journal of Orthopsychiatry 69, 220
227.
Kirkpatrick, M., Smith, C., Roy, R., 1981. Lesbian mothers and their children: a comparative survey. American Journal of Orthopsychiatry 51, 545551.
Koepke, L., Hare, J., Moran, P., 1992. Relationship quality in a sample of lesbian couples with children and child-free lesbian couples. Family Relations 41,
224229.
Kuhn, T.S., 1970/1996. The Structure of Scientic Revolutions, third ed. University of Chicago Press, Chicago.
Kweskin, S.L., Cook, A.S., 1982. Heterosexual and homosexual mothers self-described sex-role behavior and ideal sex-role behavior in children. Sex Roles 8,
967975.
Lansford, J.E., 2009. Parental divorce and childrens adjustment. Perspectives on Psychological Science 4, 140152.
Lerner, R., Nagai, A., 2001. No Basis: What the Studies Dont Tell us About Same-Sex Parenting. Marriage Law Project, Washington, DC.
Lewis, K.G., 1980. Children of lesbians: Their point of view. Social Work, 198203.
Lichter, D.T., Graefe, D.R., Brown, J.B., 2003. Is marriage a panacea? Union formation among economically disadvantaged unwed mothers. Social Problems
50, 6086.
Lott-Whitehead, L., Tully, C.T., 1993. The family lives of lesbian mothers. Smith College Studies in Social Work 63, 265280.
Lyons, T.A., 1983. Lesbian mothers custody fears. Women & Therapy 2, 231240.
MacCallum, F., Golombok, S., 2004. Children raised in fatherless families from infancy: a follow-up of children of lesbian and single heterosexual mothers at
early adolescence. Journal of Child Psychology and Psychiatry 45, 14071419.
Manning, W.D., Lamb, K.A., 2003. Adolescent well-being in cohabiting, married, and single-parent families. Journal of Marriage and Family 65, 876893.
Margolin, L., 1992. Child abuse by mothers boyfriends: why the overrepresentation? Child Abuse and Neglect 16, 541551.
McLanahan, S., Sandefur, G., 1994. Growing Up with a Single Parent: What Hurts, What Helps. Harvard University Press, Cambridge, MA.
McLeod, A.C., Crawford, I., Zecheister, J., 1999. Heterosexual Undergraduates Attitudes Toward Gay Fathers and their Children. Journal of Psychology and
Human Sexuality 11, 4362.
Miller, B., 1979. Gay fathers and their children. Family Coordinator 28, 544552.
Miller, J.A., Jacobsen, R.B., Bigner, J.J., 1981. The childs home environment for lesbian versus heterosexual mothers: a neglected area of research. Journal of
Mitchell, K.S., Booth, A., King, V., 2009. Adolescents with nonresident fathers: are daughters more disadvantaged than sons? Journal of Marriage and Family
71, 650662.
Moore, M.R., 2008. Gendered power relations among women: a study of household decision making in Black, lesbian step families. American Sociological
Review 73, 335356.
Moore, K.A., Jekielek, S.M., Emig, C.E., 2002. Marriage From a Childs Perspective: How Does Family Structure Affect Children, and What Can We Do About It?
Child Trends Research Brief, Washington, DC.
Morris, J.F., Balsam, K.F., Rothblum, E.D., 2002. Lesbian and bisexual mothers and non-mothers: demographics and the coming out process. Journal of Family
Mucklow, B.M., Phelan, G.K., 1979. Lesbian and traditional mothers responses to adult responses to child behavior and self concept. Psychological Reports
44, 880882.
Neuman, W.L., 1997. Social Research Methods, third ed. Allyn & Bacon, Boston.
Nock, S.L., 1998. Marriage in Mens Lives. Oxford University Press, New York.
Nock, S., 2001. The Afdavit of Steven Nock, Halpern v. Attorney General of Canada, No. 684/00 (Ont. Sup. Ct. of Justice).
OConnell, A., 1993. Voices from the heart: the developmental impact of a mothers lesbianism on her adolescent children. Smith College Studies in Social
Work 63, 281299.
Oliver, M.L., Shapiro, T.M., 1997. Black Wealth/White Wealth. Routledge, New York.
Page, M.E., Stevens, A.H., 2005. Understanding racial differences in the economic costs of growing up in a single-parent family. Demography 42, 7590.
Pagelow, M.D., 1980. Heterosexual and lesbian single mothers: a comparison of problems, coping, and solutions. Journal of Homosexuality 5, 198204.
Patterson, C.J., 1992. Children of lesbian and gay parents. Child Development 63, 10251042.
Patterson, C.J., 1994. Children of the lesbian baby boom: Behavioral adjustment, self-concepts, and sex-role identity. In: Greene, B., Herek, G. (Eds.),
Contemporary Perspectives on Lesbian and Gay Psychology: Theory, Research, and Application. Sage, Beverly Hills, CA, pp. 156175.
Patterson, C.J., 1995. Families of the lesbian baby boom: parents division of labor and childrens adjustment. Developmental Psychology 31, 115123.
001028
Patterson, C.J., 2000. Family relationships of lesbians and gay men. Journal of Marriage and the Family 62, 10521069.
Patterson, C.J., 2001. Families of the lesbian baby boom: maternal mental health and child adjustment. Journal of Gay and Lesbian Psychotherapy 4 (3/4),
91107.
Patterson, C.J., 2005. Lesbian and gay parents and their children: summary of research ndings. Lesbian and Gay Parenting: American Psychological
Association, pp. 522.
Patterson, C.J., Hurst, S., Mason, C., 1998. Families of the lesbian baby boom: childrens contacts with grandparents and other adults. American Journal of
Phillips, C.P., Asbury, C.A., 1993. Parental divorce/separation and the motivational characteristics and educational aspirations of African American university
students. The Journal of Negro Education 62, 204210.
Rand, C., Graham, D.L.R., Rawlings, E.I., 1982. Psychological health and factors the court seeks to control in lesbian mother custody trials. Journal of
Rank, M.R., Hirschl, T.A., 1999. The economic risk of childhood in America: estimating the probability of poverty across the formative years. Journal of
Marriage and the Family 61, 10581067.
Redding, R., 2008. Its really about sex: same-sex marriage, lesbigay parenting, and the psychology of disgust. Duke Journal of Gender Law & Policy 15, 128
193.
Roese, N.J., Olson, J.M., 2007. Better, stronger, faster self-serving judgment, affect regulation, and the optimal vigilance hypothesis. Perspectives on
Psychological Science 2, 124141.
Rosenfeld, M.J., 2010. Nontraditional families and childhood progress through school. Demography 47, 755775.
Sarantakos, S., 1996. Children in three contexts: family, education, and social development. Children Australia 21, 2331.
Sarantakos, S., 2000. Same-Sex Couples. Harvard Press, Sydney.
Sarantakos, S., 2005. Social Research, third ed. Palgrave Macmillan.
Sarantakos, S., 2007a. Data analysis, vol. 4. Sage, Thousand Oaks, CA.
Sarantakos, S., 2007b. Tool Kit for Quantitative Data Analysis. Palgrave Macmillan.
Schumm, W.R., 2004. What was really learned from Tasker and Golomboks (1995) study of lesbian and single parent mothers? Psychological Reports 94,
422424.
Schumm, W.R., 2010a. Comparative relationship stability of lesbian mother and heterosexual mother families: a review of evidence. Marriage & Family
Review 46, 499509.
Schumm, W.R., 2010b. Statistical requirements for properly investigating a null hypothesis. Psychological Reports 107, 953971.
Schumm, W.R., 2011. Child outcomes associated with lesbian parenting: Comments on Biblarz and Stacys 2010 report. Journal of Human Sexuality 3, 35
80.
Schwandt, T., 2001. Dictionary of Qualitative Inquiry. Sage, Thousand Oaks, CA.
Shiller, V.M., 2007. Science and advocacy issues in research on children of gay and lesbian parents. American Psychologist 62, 712713.
Siegel, C. et al, 1996. Mortality from intentional and unintentional injury among infants of young mothers in Colorado, 19821992. Archives of Pediatric and
Adolescent Medicine 150, 10771083.
Siegenthaler, A.L., Bigner, J.J., 2000. The value of children to lesbian and non-lesbian mothers. Journal of Homosexuality 39, 7391.
Simon, R.W., 2002. Revisiting the relationships among gender, marital status, and mental health. American Journal of Sociology 107, 10651096.
Stacey, J., Biblarz, T.J., 2001. (How) does the sexual orientation of parents matter? American Sociological Review 66, 159183.
Steckel, A., 1985. Separation-Individuation in Children of Lesbian and Heterosexual Couples. Unpublished doctoral dissertation, The Wright Institute
Graduate School, Berkeley, CA.
Steckel, A., 1987. Psychosocial development of children in lesbian mothers. In: Bozett, F.W. (Ed.), Gay and Lesbian Parents. Praeger, New York, pp. 7585.
Strasser, M., 2008. The alleged harms of recognizing same-sex marriage. In: Wardle, L. (Ed.), Whats the Harm? Does Legalizing Same-Sex Marriage Really
Harm Individuals, Families or Society? University Press, Lanham, MD, pp. 2746.
Sullivan, M., 1996. Rozzie and Harriett? Gender and family parents of lesbian coparents. Gender and Society 10, 747767.
Tasker, F.L., Golombok, S., 1995. Adults raised as children in lesbian families. American Journal of Orthopsychiatry 65, 203215.
Tasker, F.L., Golombok, S., 1997. Growing Up in a Lesbian Family: Effects on Child Development. Guilford, New York.
Tasker, F.L., Golombok, S., 1998. The role of co-mothers in planned lesbian-led families. Journal of Lesbian Studies 2, 4968.
Teachman, J.R. et al, 1998. Sibling resemblance in behavioral and cognitive outcomes: the role of father presence. Journal of Marriage and the Family 60,
835848.
Vanfraussen, K., Ponjaert-Kristoffersen, I., Brewaeys, A., 2003. Family functioning in lesbian families created by donor insemination. American Journal of
Wainright, J.L., Russell, S.T., Patterson, C.J., 2004. Psychosocial adjustment, school outcomes, and romantic relationships of adolescents with same-sex
parents. Child Development 75, 18861898.
Waite, L., 1995. Does marriage matter? Demography 32, 483507.
Waite, L., Gallagher, M., 2000. The Case for Marriage: Why Married People are Happier, Healthier, and Better off Financially. Doubleday, New York.
Wallerstein, J., Lewis, J.M., Blakeslee, S., 2001. The Unexpected Legacy of Divorce. Hyperion, New York.
Wardle, L.D., 1997. The potential impact of homosexual parenting on children. University of Illinois Law Review 1997, 833919.
Weitoft, G.R. et al, 2003. Mortality, severe morbidity, and injury in children living with single parents in Sweden: a population-based study. The Lancet 361,
289295.
Wilcox, W.B. et al, 2005. Why Marriage Matters, second ed. Institute for American Values, New York.
Wilcox, W.B. et al, 2011. Why Marriage Matters, third ed. Institute for American Values, New York.
Williams, R.N., 2000. A critique of the research on same-sex parenting. In: Dollahite, D.C. (Ed.), Strengthening Our Families. Bookcraft, Salt Lake City, UT, pp.
325355.
Wolnger, N.H., 2005. Understanding the Divorce Cycle: The Children of Divorce in their Own Marriages. Cambridge University Press, New York.
Wright, J.M., 1998. Lesbian Stepfamilies: An Ethnography of Love. Harrington Park, New York.
001029

TAB 60

iMAPP Research Brief
Institute for Marriage and Public Policy Vol. 5, No.6, October 2012
MISPLACED RELIANCE ON SOCIAL SCIENCE EVIDENCE IN THE PROPOSITION 8 CASE
William C. Duncan, Legal Analyst
In its opinion invalidating the California marriage amendment, Proposition 8, the trial court
relied heavily on social scientist witnesses and statements of professional organizations for a
number of propositions the court believed required the result.
1
This has been a common tactic in
the litigation aimed at securing a judicial redefinition of marriage.
There's reason to believe that this reliance is misplaced.
For example, the court cites a "Position Statement" of the American Psychiatric Association for
the proposition that children of same-sex couples are harmed because the couple cannot marry
and an American Medical Association "Policy" that same-sex couples experience health
disparities as a result of the current definition of marriage. The American Psychiatric Association
statement is, as the court note, a "Position" rather than a scholarly article. The sources cited in the
statement are all other policy documents rather than research findings. The American Medical
Association policy document is likewise devoid of citations.
These kinds of problems seem to be endemic in similar organizational statements on same-sex
parenting issues. The statements of the American Anthropological Association and the American
Academy of Child and Adolescent Psychiatry
2
are likewise free of cites. The groups issuing many
of these statements also endorse a range of causes to which social science considerations are
tangential at best.
3
They have also been subjected to criticism within their ranks.4
The politically motivated nature of the American Academy of Child and Adolescent Psychiatry
was highlighted by a colloquy from the Proposition 8 trial. After reading the Academy's
statement that "Lesbian, gay, bisexual or transgender people have faced more rigorous scrutiny
than heterosexual people regarding their rights to be or become parents," counsel said to the
plaintiffs' child development expert, Dr. Michael Lamb (who is prominently used as an expert in
these types of cases): "Dr. Lamb, there is not a rich empirical literature relating to child outcomes
of transgender individuals, is that right?" Dr. Lamb responded, "I'm not familiar with it, no."
Counsel then asked, "And there is not a rich literature on the child outcomes of the children of
bisexuals, correct?" Dr. Lamb responded, "That's correct." Counsel then stated the obvious
conclusion: "So this statement is not based in empirics, but, rather, in politics, correct?" Dr. Lamb
concurred: "Well, I can't speak to the basis. That would be my understanding, yes."s
There are many problems endemic in a reliance on these kinds of professional organization
statements. Daniel N. Robinson, professor of philosophy at Oxford and Georgetown Universities
and past president of two divisions of the American Psychological Association, points out: "Even
when there is relevant expertise within the discipline of psychology there is no a priori basis
upon which to credit a given Board or Council with possession of it. Scholars and scientists do
not vote on matters of fact, nor do they install theories into positions of supremacy by fiat."6 This
is true because "[m]oral authority does not inhere in committees, boards, councils, or
memberships, but in the weight of reason attached to arguments for and against actions of a
certain kind."
7
The organization's "concern" "is not an argument."
8
Ultimately, Dr. Robinson
WWW.IMAPP.ORG
001030
explains what "is unsound in the APA's policy of advocacy is the apparent conviction that the
moral categories of right and wrong are simply brought into being by a show of hands."
9
Indeed, these kinds of statements by professional organizations are susceptible to direct political
influence as illustrated by the recent controversy over a statement of the American College of
Obstetricians and Gynecologists that was edited by an associate White House Counsel (who
inserted a crucial sentence into the statement) and then relied on by courts in cases involving
abortion laws.1o This led journalist William Saletan to argue, "Judges have put too much faith in
statements from scientific organizations. . . . Judges should stop treating the statements of
scientific organizations as apolitical. Such statements, like the statements of any other group, can
be loaded with spin. This one is a telling example."
11
Dr. Robinson points out that "the composition of the APA could change and could come to reflect
any number of reactionary impulses."
12
Would it then be reasonable for another court to rule that
marriage must be defined as the union of a man and a woman if the APA's leadership were to
change and vote to rescind its' current policy endorsement?
To take another example, the findings of the Proposition 8 trial court regarding child outcomes
relied heavily, almost exclusively, on one expert witness (a psychologist discussed below) and on
a document produced by the American Psychological Association. The statement does not
contain citations and admits to serious limitations. The four findings that proceed the language
quoted by the court below say that (1) sexual identities of "children of lesbian mothers" develop
similarly to those of heterosexual parents though "[f]ew studies are available regarding children
of gay fathers," (2) children of lesbian mothers have few differences in personality development
than children of heterosexual parents (with the same note about the lack of studies of gay
fathers), (3) children of gay parents have normal social relationships, and (4) children of gay
parents are not more likely to be sexually abused.
13
Then, based on this thin record, it comes to
the conclusion quoted by the court which conflates "lesbian and gay parents" even though the
statement notes a dearth of research about the latter. This hardly promotes confidence in the
"scientific" nature of the assertions.
Dr. Michael Lamb, the developmental psychologist whose testimony was so influential with the
court, seems to exemplify the risk that non-scientific considerations may play an inordinate role
in expert evaluations of research. For example, he acknowledged at trial that a number of his
earlier articles had noted the importance of fathers in children's development. For instance, that
"it is disturbing that there appears to have been a devaluation of the father's role in western
society such that many children may suffer affective paternal deprivation."
14
Some other
examples from his early articles include:
"[b]oys growing up without fathers seem to have problems in the area of sex role and gender
identity development, school performance, psychosocial adjustment, and perhaps in the
control of aggression" IS
"[b]oys growing up without fathers seem especially prone to exhibit problems in the areas of
sex role and gender identity development"l6
He had also written about differences between men and women in terms of parenting: "[t]he data
suggests that the differences between maternal and paternal behavior are more strongly related
-2-
001031
to either the parents' biological gender or sex roles, than to either their degree of involvement in
infant care or their attitudes regarding the desirability of paternal involvement in infant care."
17
Yet, in the decision of the Proposition 8 court, Dr. Lamb's testimony was relied on by the court
for the conclusions that "[t]he gender of a child's parent is not a factor in the child's adjustment"
18
and "[c]hildren don't need to be raised by a male parent and a female parent to be well-adjusted,
and having both a male and female parent does not increase the likelihood that a child will be
well-adjusted."
How then to explain this discrepancy? One might assume that research developments since Dr.
Lamb's earlier statements had contradicted the earlier conclusions. Yet this is not the case. In fact,
Dr. Lamb admitted at trial that there are a number of areas in which a father's absence is
meaningful for child outcomes. For instance, admitting:
that "[t]he increase in father's absence is particularly troubling because it is consistently
associated with poor school achievement, diminished involvement in the labor force, early
child bearing, and heightened levels of risk-taking behavior."
1
9
that boys without fathers are "prone to poor school performance."
2
o
that boys without fathers fare worse in terms of psychosocial adjustment.
21
that there are differences in terms of self-control and delinquent behavior in adolescence
between boys with and without fathers.
22
"that nurturant fathers may contribute greatly to the psychological adjustment of their
daughters."23
that nurturant fathers "may facilitate their [daughters'] happiness in subsequent heterosexual
relationships."24
that there are "some fairly long-term associations between the quality of the relationship that
young children have with their fathers and the way that they interact as young adults with
their own peers."25
In addition, Dr. Lamb conceded at trial that there are indeed differences between men and
women that have implications for child development, admitting:
that mothers and fathers are different in a number of respects, that those differences may
be the result of their different genders, and that being raised by people with such
differences is beneficial for children.
2
6
that it was an accurate summary of the literature to say "[r]esearch clearly demonstrates
that children growing up with two continuously married parents are less likely than other
children to experience a wide range of cognitive, emotional, and social problems not only
during childhood but also in adulthood" and that this "distinction is even stronger if we
focus on children growing up with two happily married biological parents."
27
that it was an accurate summary of the literature to say "[w]e conclude that in practice the
kind of mother-father relationship most conducive to responsible fathering in
contemporary U.S. society is a caring, committed, collaborative marriage. Outside of this
arrangement, substantial barriers stand in the way of active, involved fathering."
2
B
that it was true that children in average intact married families do better than children in
average single and stepparent families.29
-3-
001032
that there are differences between men and women-that men are more likely than
women to become incarcerated, be involved in violent altercations, perpetuate sexual
abuse, become addicted to alcohol, and to engage in aggressive behavior; that men and
women suffer from different diseases at different times; that women tend to live longer
than men; that men are disproportionally represented on the bottom of the bell curve for
cognitive abilities; that men are less likely to graduate high school than women; that men
cannot breastfeed; and that men earn more money than women.30
that he had stated, in regards to parenting, that men and women are not "completely
interchangeable with respect to skills and abilities."3
1
that gender "is one of those variables that can have ripple effects in a variety of different
ways on the way in which people behave, and can in a variety of ways affect the way they
behave with their children."32
It is hard to reconcile these statements with the conclusions of the court noted above that are, in
turn, based on expert opinions. That is, unless we consider the very real possibility that the
conclusions are influenced by factors other than scientific research.
Perhaps it should come as no surprise that individual experts and professional organizations'
opinion statements on same-sex parenting and same-sex marriage questions so significantly
illustrate political and cultural pressures. Sociologist Norval Glenn of the University of Texas,
Austin explained: "Given the widespread support for same-sex marriage among social and
behavioral scientists, it is becoming politically incorrect in academic circles even to suggest that
arguments being used in support of same-sex marriage might be wrong."
33
Similarly, two of the
more enthusiastic supporters of family change (including redefining marriage) admit: "[T]he
political stakes of this body of research are so high that the ideological 'family values' of scholars
play a greater part than usual in how they design, conduct, and interpret their studies."34 They
note further: "[T]oo many psychologists who are sympathetic to lesbigay parenting seem hesitant
to theorize at all" and are apt to "downplay the significance of any findings of differences."3s
What is disconcerting, however, is the endorsement of these opinions as legally significant "facts"
that are "beyond dispute," as the district court found in striking down Proposition 8. Such a
conclusion ignores the role of ideology in social sciences and is itself an ideological statement, not
a scientific one.
Notes
1
Perry v. Schwarzenegger, 704 F. Supp. 2d 921 (N.D. Cal. 2010).
2
American Academy of Child & Adolescent Psychiatry, Gay Lesbian, Bisexual, or Transgender
Parents Polict) Statement (October 2008) at
http://www.aacap.org/cs/root/policy statements/gay lesbian transgender and bisexual parents
policy statement.
3 The American Anthropological Association has positions on boycotting the state of Arizona
(http://www.aaanet.org/issues!press/Arizona-Immigration.cfm) and the American Medical
Association on abortion, global warming, recycling, gun control and others (http://www.ama-
assn.org/amal/pub/upload/mm/15/digest of actions.pdf).
4
See Peter Wood, The Marriage Debate Goes Multicultural, NATIONAL REVIEW ONLINE, April 26,
2005 at http://old.nationalreview.com/commentlwood200504260810.asp (criticizing the American
-4-
001033
Anthropological Association statement); Letters to the Editor 110 PEDIATRICS 419 (2002) & Letters to
the Editor 109 PEDIATRICS 1192 (2002) (publishing letters critical of the American Academy of
Pediatrics' 2002 statement).
5 Trial Transcript, 1052-1054, 62:15-64:3.
6 Daniel N. Robinson, Ethics and Advocactj 39 AMERICAN PSYCHOLOGIST 787, 788 (1984).
7 Id. at 792.
8 Id.
9 Id. at 790.
1
0 William Saletan, When Kagan Played Doctor, SLATE, July 3, 2010 at
http://www.slate.com/id/2259495/pagenum/all/#p2.
II Id.
12
Robinson, 39 AMERICAN PSYCHOLOGIST at 793.
13
American Psychological Association, Answers to Your Questions For a Better Understanding of
Sexual Orientation & Homosexuality, 5 at http://www.apa.org/topics/sexualitv/sorientation.pdf.
14 Trial Transcript, 1071:7-10.
18
It should be noted that this statement is actually not supported by the evidence the court cites
since the three sources the court notes actually say that a person attracted to another person of
the same-sex are not necessarily bad parents which is a far cry from the conclusion that a parent's
gender is entirely unrelated to child outcomes.
19 Trial Transcript, 1073: 6-9.
22 Trial Transcript, 1074-75: 25-3.
27 Trial Transcript, 1098:3-8, 14-16.
29 Trial Transcript, 1102-03:23-4.
30 Trial Transcript, 1057-58:22-19.
33 Norval D. Glenn, The Struggle For Same-Sex Marriage 41 Soc'Y 25, 27 (2004).
34 Judith Stacey & Timothy Biblarz, (How) Does the Sexual Orientation of Parents Matter? 66
AMERICAN SOCIOLOGY REVIEW 159, 161 (2001), accord Rober Lerner & Althea K. Nagai, No Basis:
What the Studies Don't Tell Us about Same-Sex Parenting.Washington, DC: Marriage Law Project 56
n.27 (2001), and Walter R. Schumm, What Was Really Learned from Tasker & Golombok's (1995) Study
of Lesbian & Single Parent Mothers?, 94 Psycho!. Rep. 422, 423 (2004).
35 Id.
-5-
001034

TAB 61

John R. Searle
The Construction
of Social Reality
001035
THE FREE PRESS
A Division of Simon &. Schuster Inc.
1230 Avenue of the Americas
New York, NY 10020
Copyright 1995 by John R. Searle
All rights reserved,
including the right of reproduction
in whole or in part in any form.
First Free Press Edition 1995
THE FREE PREss and colophon are trademarks
of Simon &. Schuster Inc.
Designed by
Manufactured in the United States of America
10
Library of Congress Cataloging-in-Publication Data
Searle, John R.
The construction of social reality I John R. Searle.
p. em.
Includes bibliographical references.
JSBN-13: 978-0-02-928045-4
ISBN-I 0: 0-02-928045-1
JSBN-13: 978-0-684-83179-4 (Pbk)
ISBN-I 0: 0-684-83179-1 (Pbk)
1. Social epistemology. 2. Philosophy of mind. I. Title.
BD175.S43 1995 94-41402
121-dc20 CIP
001036
The Building Blocks of Social Reality 5
function, no answer to the question, What's it for? we are left with
a harder intellectual task of identifying things in terms of their in-
trinsic features without reference to our interests, purposes, and
goals.
The invisibility of the structure of social reality also creates a
problem for the analyst. We cannot just describe how it seems to
us from an internal "phenomenological" point of view, because
money, property, marriages, lawyers, and bathtubs do not seem to
have a complex structure. They just are what they are, or so it
seems. Nor can we describe them from the external behaviorist
point of view, because the description of the overt behavior of peo-
ple dealing with money, property, etc., misses the underlying
structures that make the behavior possible. Nor, in turn, can we
describe those structures as sets of unconscious computational
rules, as is done by contemporary cognitive science and linguis-
tics, because it is incoherent to postulate an unconscious follow-
ing of rules that is inaccessible in principle to consciousness. And
besides, computation is one of those observer-relative, functional
phenomena we are seeking to explain.
2
If neither the internal phenomenological nor the external be-
haviorist point of view is adequate, what then is the correct stance,
the correct methodology, for describing the structure of social
reality? To start with, in this chapter and the next, I will use a first-
person intentionalistic vocabulary to try to lay bare certain ele-
mentary features of social ontology. Later, in Chapter 6, I will
show how some, though not all, of the intentionalistic apparatus
can be explained in terms of, and ultimately eliminated in favor of,
what I have elsewhere called the "Background" of capacities, abil-
ities, tendencies, and dispositions.
Fundamental Ontology
Since our investigation is ontological, i.e., about how social facts
exist, we need to figure out how social reality fits into our overall
ontology, i.e., how the existence of social facts relates to other
001038
The Building Blocks ofSocial Reality 29
3. Bills issued by the Bureau of Engraving and Printing count as
the root of all evil in the United States.
As always the discovery of referential opacity is a crucial point. In
this case it provides a clue that there is a mental component
in institutional facts. The intentionality-with-an-s of the verbal
formulation is a clue that the phenomena represented are inten-
tional-with-a-t. A great deal hangs on this, as we will see in subse-
quent chapters.
Various social theorists have attacked my account of the dis-
tinction between regulative and constitutive rules,
10
but I think my
account is right as far as it goes. The problem is that for our pre-
sent purposes it does not go far enough. We still need a more thor-
ough account of rules and institutions. And we need to answer a
lot of questions. Are all social facts institutional facts? Are there
constitutive rules of, for example, wars and cocktail parties? What
makes something into a "constitutive rule" anyway? Hardest of all,
how do we make the connection between the fundamental ontol-
ogy of conscious biological beasts like ourselves and the appara-
tus of social facts and human institutions?
1 will have more to say later about the form of constitutive rules
and how they relate to the ontology of institutional facts. My aim in
this chapter is to assemble the pieces, and I now have the three I
need: the imposition of function on entities that do not have that
function prior to the imposition, collective intentionality, and the
distinction between constitutive and regulative rules. With these in
hand we can now turn to the construction of institutional reality.
001041
Creating Institutional Facts 33
even if no one ever knew that it was counterfeit, not even the
counterfeiter. In such a case everyone who used that particular
token would think it was money even though it was not in fact
money. About particular tokens it is possible for people to be sys-
tematically mistaken. But where the type of thing is concerned,
the belief that the type is a type of money is constitutive of its being
money in a way we will need to make fully clear.
For some institutional phenomena, such as money, what I say
applies more to types than tokens, for others, such as cocktail par-
ties, it applies to each individual token. For the sake of simplicity I
will assume that the reader is aware of the distinction, and I will
speak of the self-referentiality of institutional concepts in generat
without making the distinction in every case. Later I will try to ex-
plain the difference between self-referentiality as applied to types
and as applied to tokens.
But if the type of thing in question is money only because peo-
ple believe it to be money, if "money" implies "regarded as, used
as, or believed to be money," then philosophers will get worried,
because the claim seems to produce either a vicious infinite
regress or a vicious circle. If part of the content of the claim that
something is money is the claim that it is believed to be money,
then what is the content of that belief? If the content of the belief
that something is money contains in part the belief that it is
money, then the belief that something is money is in part the be-
lief that it is believed to be money; and there is, in turn, no way to
explain the content of that belief without repeating the same fea-
ture over and over again. Later on, I will try to show how to avoid
this infinite regress. At this point, I am just calling attention to a
peculiar logical feature that distinguishes social concepts from
such natural concepts as "mountain" or "molecule." Something
can be a mountain even if no one believes it is a mountain; some-
thing can be a molecule even if no one thinks anything at all about
it. But for social facts, the attitude that we take toward the phe-
nomenon is partly constitutive of the phenomenon. If, for exam-
ple, we give a big cocktail party, and invite everyone in Paris, and if
001044
Creating lnslilutional Facts 35
function as money, but money must come in some physical form
or other.
What is true of money is true of chess games, elections, and
universities. All these can take different forms, but for each there
must be some physical realization. This suggests what I think is
true, that social facts in general, and institutional facts especially,
are hierarchically structured. Institutional facts exist, so to speak,
on top of brute physical facts. Often, the brute facts will not be
manifested as physical objects but as sounds coming out of
peoples' mouths or as marks on paper-or even thoughts in their
heads.
4. Systematic Relationsllips Among
lnslilulional Facts.
An institutional fact cannot exist in isolation but only in a set of
systematic relations to other facts. Thus, for example, in order
that anybody in a society could have money, that society must have
a system of exchanging goods and services for money. But in
order that it can have a system of exchange, it must have a system
of property and property ownership. Similarly, in order that soci-
eties should have marriages, they must have some form of con-
tractual relationships. But in order that they can have contractual
relationships, they have to understand such things as promises
and obligations.
Furthermore, quite apart from the logical or conceptual re-
quirement of interrelationships of institutional facts, it just turns
out that in any real life situation one will find oneself in a complex
of interlocking institutional realities. The restaurant scene de-
scribed in Chapter 1 illustrates this: at any instant in the scene,
one is (at least) a citizen, an owner of money, a client, a bill payer;
and one is dealing with property, a restaurant, a waiter, a bill.
It might seem that games are counterexamples to this general
principle, because, of course, games are designed to be forms of
activity that do not connect with the rest of our lives in a way that
001046
56 The Construction ofSocial Reality
one has to reach a rock bottom of something that is not itself any
form of status-function. So, for example, as I said earlier, all sorts
of things can be money, but there has to be some physical realiza-
tion, some brute fact-even if it is only a bit of paper or a blip on a
computer disk-on which we can impose our institutional form
of status function. Thus there are no institutional facts without
brute facts.
This discussion anticipates a discussion of realism I will pre-
sent in Chapters 7 and 8. It could not be the case, as some antireal-
ists have maintained, that all facts are institutional facts, that there
are no brute facts, because the analysis of the structure of institu-
tional facts reveals that they are logically dependent on brute
facts. To suppose that all facts are institutional would produce an
infinite regress or circulari1y in the account of institutional facts.
In order that some facts be institutional, there must be some
other facts that are brute. This is a consequence of the logical
structure of institutional facts.
Systematic Relations and the Primacy
of the Act over the Object
Our fourth question was, Why are there always certain sorts of
systematic relations among institutional facts? And the fifth was,
Why do institutional acts seem prior to institutional objects?
The most obvious reason why there are systematic relation-
ships among the various sorts of social facts of the 1ype that I tried
to describe is that the facts in question are designed for precisely
that purpose. Goven1ments are designed to impact on our lives in
all sorts of ways; money is designed to provide a unit of value in all
kinds of transactions. Even games, which are explicitly designed
to be insulated from the rest of our lives, nonetheless employ an
apparatus-of rights, obligations, responsibilities, etc.-that, as I
remarked earlier, is intelligible only given all sorts of other social
facts.
The explanation for the apparent primacy of social acts over so-
001050
Creating Institutional Facts 57
cial objects is that the "objects" are really designed to se1ve agen-
tive functions, and have little interest for us otherwise. What we
think of as social objects, such as governments, money, and uni-
versities, are in fact just placeholders for patterns of activities. I
hope it is clear that the whole operation of agentive functions and
collective intentionality is a matter of ongoing activities and the
creation of the possibility of more ongoing activities.
Unconsciously, we have throughout this discussion been ac-
knowledging this point by our talk of institutional facts rather than
institutional objects. Such material objects as are involved in insti-
tutional reality, e.g., bits of paper, are objects like any others, but
the imposition of status-functions on these objects creates a level
of description of the object where it is an institutional object, e.g., a
twenty dollar bill. The object is no different; rather, a new status
with an accompanying function has been assigned to an old object
(or a new object has been created solely for the purpose of serving
the new status-function), but that function is manifested only in ac-
tual transactions; hence, our interest is not in the object but in the
processes and events where the functions are manifested.
The priority of process over product also explains why, as sev-
eral social theorists have pointed out, institutions are not worn
out by continued use, but each use of the institution is in a sense
a renewal of that institution. Cars and shirts wear out as we use
them but constant use renews and strengthens institutions such
as marriage, property, and universities. The account I have given
explains this fact: since the function is imposed on a phenomenon
that does not perform that function solely in virtue of its physical
construction, but in terms of the continued collective intentional-
ity of the users, each use of the institution is a renewed expression
of the commitment of the users to the institution. Individual dol-
lar bills wear out. But the institution of paper currency is rein-
forced by its continual use.
The sixth and final feature we need to explain concerns the role
of language in institutional reality, and to that topic I devote the
next chapter.
001051
3
Language and Social Reality
The primary aim of this chapter is to explain and justifY my claim
that language is essentially constitutive of institutional reality. I
have made this claim in general terms but I now want to make
fully explicit what I mean by it, and to present arguments for it. At
the end of the chapter I will mention some other functions of lan-
guage in institutional facts.
I said in the last chapter that it seems impossible to have insti-
tutional structures such as money, marriage, governments, and
property without some form of language because, in some weird
sense I have not yet explained, the words or other symbols are
partly constitutive of the facts. But this will seem puzzling when
we reflect that social facts in general do not require language.
Prelinguistic animals can have all sorts of cooperative behavior,
and human infants are clearly capable of interacting socially in
quite complex ways without any words. Furthermore, if we are
going to say that institutional reality requires language, what
59
001052
Thus we can refer 01' fail to refer to "the touchdown we scored at
the end of the fourth quarter" or "the President of the Uniled
States" in the same way we can succeed m fail to refer to "The
Evening Star," but the difference is that the creation of thecate-
gory of touchdowns and presidents is already achieved by the
structures according to which we attach status-functions to the X
terms, because the existence of these features is created by at-
tachment of the status-functions.
Think of it this way: What stands to the sound "cat" as its mean-
ing is what stands to the piece of papm as its function as a dollar
bill. However, the sound "cat" has a referential function that the
piece of paper does not have. For example, the sound can occur in
sentences where the speaker in uttering the sentence refers to a
cat. Pieces of paper, even pieces of paper construed as dollar bills,
are not in that way used to refer. But the practice of using pieces
of paper as dollar bills creates a class of entities that cannot exist
without the p1actice. It creates the class of entities: dollar bills. In
order that the practice should exist, people must be able to think
the thought "This piece of paper is a dollar bill," and that is a
thought they cannot think without words or other symbols, even if
the only symbol in question is the object itself.
Other F'unctions of Language in
Institutional Facts
This discussion has been vmy abstract and has concerned the
conditions of the possibility of the creation of institutional reality,
linguistic or otherwise. But if we consider actual natural lan-
guages such as French 01' German and the actual complexity of
social institutions, we can see several other reasons why institu-
tional facts require language.
First, language is epistemically indispensable.
I said that in the structure of institutional facts we impose a Y
status-function on the X term, which it does not perform solely in
001054
Language and Social Reality 77
virtue of its physical constitution. But now how are we to tell
which entities have this status function imposed on them? For
many causal agcntive functions-not all-it is reasonably easy to
tell which objects are chairs, tables, hammers, and screwdrivers
because you can read off the function from the physical structure.
But when it comes to money, husbands, university professors, and
privately owned real estate, you cannot read off the function or
status from the physics. You need labels. In order that we can rec-
ognize bits of paper as money, for example, we must have some
linguistic or symbolic way of representing the newly created facts
about functions, because they cannot be read off from the physics
of the objects themselves. The recognition of the fact that some-
thing is money requires that it be linguistically or symbolically
represented. I will have more to say about this feature in the next
chapter, when we discuss what I call "status indicators."
Second, the facts in question, being inherently social, must be
communicable.
If the systems are to function, then the newly created facts must
be communicable from one person to another, even when invisi-
ble to the naked eye. You must be able to tell people that you are
married, that you are the chairman, that the meeting is adjourned
if the system is to function. Even in simple cases of institutional
facts, this communicability requires a means of public communi-
cation, a language.
Third, in real life the phenomena in question are e(dremely
complel(., and the representation of such complel(. information
requires language.
Even the most apparently simple act of buying and selling has
great complexity, as we saw in our example of ordering beer in a
cafe at the beginning of the book. Because the structure of the
facts exists only to the extent that it is represented, complex facts
require a complex system of representation for their existence;
and such complex systems of representation are languages.
001055
Fourth, the facts in question persist through time independently of
the duration ofthe urges and inclinations ofthe participants in the
institution.
This continued existence I'equires a means of representation of
the facts that is independent of the more primitive prelinguistic
psychological slates of the participants, and such representations
are linguistic.
001056
4
The General Theory
of Institutional Facts
Part 1: Iteration, Interaction, and Logical Structure
Generalizing the Analysis
So far I have given a preliminary account of institutional facts,
using the example of money more than any other sort and em-
phasizing the special role of language in institutional reality. I
will use the tools we have assembled to give an account that de-
scribes the structure not only of money but also of marriage,
property, hiring, firing, war, revolutions, cocktail parties, govern-
ments, meetings, unions, parliaments, corporations, laws, restau-
rants, vacations, lawyers, professors, doctors, medieval knights,
and taxes, for example. I do not know how to tell the story for each
of these with the simplicity of the story about money. To general-
79
001057
The General Theory of Institutional Facts (Part I) Bl
the italicized expressions in the previous sentence express institu-
tional concepts, and the facts reported all presuppose systems of
constitutive rules operating through time.
To develop the analysis further, let us try to tell a story about
marriage and property analogous to the one we told about money.
Such institutions originate in the sheer physical and intentional
facts involved in cohabitation and physical possession, respec-
tively. Property begins with the idea that I have got this, it is mine.
Marriage begins with people simply living with each other, and in
the case of monogamous marriage, having a sexual monopoly on
each other. Why are we not satisfied with these arrangements?
Why is it not enough that I possess this in the sense that I have
physical control over it and why is it not enough that we just live
together? Well, for some people and perhaps for some simple so-
cieties it is enough; but many of us think we are better off if there
is a system of collectively recognized rights, responsibilities, du-
ties, obligations, and powers added onto-and in the end able to
substitute for-brute physical possession and cohabitation. For
one thing, we can have a much more stable system of expectations
if we add this deontic apparatus; for another, we don't have to rely
on brute physical force to sustain the arrangements; and for a
third, we can maintain the arrangements even in the absence of
the original physical setup. For example, people can remain mar-
ried even though they have not lived with each other for years, and
they can own property even though the property is a long way
away from them.
Whatever the advantages and disadvantages, the logically more
primitive arrangements have evolved into institutional structures
with collectively recognized status-functions. Just as in the case of
money, we have imposed, by collective intentionality, new status-
functions on things that cannot perform those functions without
that collective imposition. However, one special feature of these
cases is that often the function is imposed by way of performing
explicit speech acts. In such cases the speech act itself is an in-
stance of a status-function imposed on a status-function; and it is
001059
Tile General Theory of Institutional Facts (Part I) 83
tutions, then it is a consequence of the account that there will be
a hierarchical structure of the creation of a large number of insti-
tutional facts. Thus, going through the example of marriage: Mak-
ing certain noises counts as uttering an English sentence, uttering
a certain sort of English sentence in certain circumstances counts
as making a promise, making a promise in certain circumstances
counts as entering into a contract, entering into certain sorts of
contracts counts as getting married. The marriage ceremony cre-
ates a new institutional fact, the marriage, by imposing a special
function on a set of speech acts. But the creation of the marriage
imposes a new status and therefore a new function on the indi-
viduals involved. They are now "husband" and "wife." And the fact
that they are husband and wife, like the marriage itself, is an insti-
tutional fact.
I hope it is clear from these examples that a pattern is emerg-
ing. The crucial questions to ask are, On what exactly are the
status-functions imposed and what exactly are the imposed
status-functions? In the case of language and money the answers
are relatively simple: for language the statuses are imposed on
types of sounds and marks; and though the functions of language
are numerous, the primary functions are those of representing
the world in the various speech act modes.
1
For money, the sta-
tuses have typically been imposed on bits of metal and paper, and
the function are those of serving as a medium of exchange, repos-
itory of value, etc. In the case of marriage, the situation is a little
more complicated. The status is initially imposed on a set of
speech acts, those that constitute the marriage ceremony, but
those speech acts function to create a new institutional fact, the
marriage. But tl1e marriage itself imposes new status-functions on
the parties involved, the status-functions of being husband and
wife, which carry specific rights and obligations. Now this pattern,
the creation of a new institutional fact, usually by the performance
of a speech act, where the speech act itself imposes a function on
people, buildings, cars, etc., is characteristic of a large number of
social institutions. Property, citizenship, licensed drivers, cathe-
001061
The General Theory of Institutional Facts (Part I) 85
queathing, partial transfer, mortgaging, etc., of property. The
characteristic devices used are speech acts-deeds, bills of sale,
registration papers, wills, etc.; and it is no accident that these are
usually called legal "instruments." All are cases of status-functions
imposed on speech acts. And, of course, the original speech act is
already a case of imposed status-function. So, for example, a bill of
sale simply records the fact that I sold you, for example, my car. It
is an assertive speech act, but it now can count as your title to the
car pending the issuance of new registration papers.
Once a society has the institution of property, new property
rights are usually created by speech acts, as when I give some-
thing to someone, or by speech acts accompanied by other sorts
of acts, as when I exchange property for money. Suppose I give my
watch to my son. I can do this by saying, "It's yours," "You can have
it," or more pompously with the performative, "I hereby give you
my watch." I have now imposed a new status-function on these
speech acts, that of transferring ownership. These speech acts in
turn impose a new status-function on the watch, that of belonging
to my son, that of being his property.
I said that the institutional structures enable brute physical
possession in the case of property, or brute physical proximity in
the case of marriage, to be replaced by a recognized set of rela-
tionships whereby people can be married even though they are
not living with each other, and people can own property even
though the property is far away from them. To achieve this re-
markable intellectual feat, we must have what I have called status
indicators. Just as the paper certificates, when they were re-
deemable in gold, were status indicators for value, so we have an
acknowledged system of legally recognized marriages and prop-
erty rights. And we have status indicators in the form of marriage
certificates, wedding rings, and title deeds, for example. Even
when I am a long way from my house or my wife, the institutional
structures enable me to remain an owner or a husband, and, if
need be, to demonstrate that position to others through the use of
status indicators. In such cases, the institutional facts substitute
001063
the new functions, of marriage and knighthood, to the degree that
paper is arbitrarily related to the function of money; but at the
same time they are not matters of necessity either. One can imag-
ine and even construct all sorts of perfectly acceptable ways of get-
ting married or becoming a knight. And because of this slack
between the conditions specified by the X term and the function
specified by the Y term, cultures differ in the qualifications they
require for the performance of the same or similar functions. For
example, in most Anlerican states, the status of "attorney" requires
the possession of a graduate law degree, passing a state bar exam,
and swearing in. In Britain, on the other hand, no graduate law
degree is required, but such things as being articled to a solicitor
for a certain period of time or dining regularly at the Inns of Court
count towards qualification. It is not at all obvious how these two
different sets of conditions are supposed to enable the possessor
to serve the same function, that of legal counsel. Nonetheless, the
respective accrediting agencies apparently think they do.
The bifurcation of the imposition of status-functions into the X
andY components has some important consequences for our in-
vestigation. First, the status expressions admit of two definitions,
one in terms of the constitution (the X term) and one in terms of
the imposed agentive function (theY term). Thus currency can be
defined in terms of its origin and structure: Certain sorts of paper
issued by the Bureau of Engraving and Printing (X term) are U.S.
currency. But currency can also be in part defined as, and indeed
is described on the face of U.S. currency as, "legal tender for all
debts, public and private" (Y term). A touchdown is when you
break the plane of the goal line with the ball in your possession
while the play is in progress (X term), and a touchdown is six
points (Y term).
Codification
A test for the presence of genuine institutional facts is whether or
not we could codify the rules explicitly. In the case of many insti-
001065
tutional facts, such as property, marriage, and money, these in-
deed have been codified into explicit laws. Others, such as friend-
ship, dates, and cocktail parties, are not so codified, but they
could be. If people believe that a certain set of relationships in
which they are involved is a case of friendship/date/cocktail party,
then the possession of each such status is constituted by the belief
that the relationship does in fact possess that status, and the pos-
session of the status carries with it certain functions. This is
shown by the fact that the people involved have certain sorts of
justified expectations from a friendship/date/cocktail party, which
they do not have from an identical set of arrangements about
which they do not believe that it is a friendship/date/cocktail party.
Such institutional patterns could be codified if it mattered tre-
mendously whether or not something was really a cocktail party
or only a tea party. If the rights and duties of friendship suddenly
became a matter of some grave legal or moral question, then we
might imagine these informal institutions becoming codified ex-
plicitly, though of course, explicit codification has its price. It de-
prives us of the flexibility, spontaneity, and informality that the
practice has in its uncodified form.
It should be clear from these examples that there is a gradual
transition and not a sharp dividing line between social facts in
general and the special subclass of institutional facts. In my soci-
ety "going for a walk with someone" names a social fact but not
an institutional fact, because the label assigns no new status-
functions. It just labels the intentionality and its manifestation.
The characteristic institutional move, however, is that form of col-
lective intentionality that constitutes the acceptance, recognition,
etc., of one phenomenon as a phenomenon of a higher sort by im-
posing a collective status and a corresponding function upon it.
The function is always internally related to the status in the sense
that it could not be that status if it did not have that function. The
criterion is always this: Does the assignment of the label carry
with it the assignment of some new functions, for example, in the
form of rights and responsibilities, which can be performed only
001066
if there is collective acceptance of the function? By this criterion,
"husband," "leader," and "teacher" all name status-functions; but
"drunk," "nerd," "intellectual," and "celebrity" do not. And, to re-
peat, it should be obvious that there is no sharp dividing line.
A fascinating test case for this account is war. War is always a
form of collective intentionality; hence it is a war only if people
think it is a war. But in typical wars, the sheer events count as hav-
ing a certain legal or quasi-legal status that is supposed to impose
certain responsibilities and rights on the participants; and in such
cases the war is more than just a social fact; it is an institutional
fact. Furthermore, as with marriage, there are ways in which the
institutional status is supposed to be imposed. Thus in the case of
the war in Korea, the American authorities at the time were very
anxious that it not be called "the Korean War" (it was called "the
Korean conflict") because it did not satisfy the legal definition of
war, since no war had been legally declared in accordance with
the Constitutional provision for a declaration of war. They had a
choice: If it was a "war," it was unconstitutional; so it was not a
"war"; it was a "United Nations police action," a different status-
function altogether. Since the phenomenon did not satisfy the X
term for imposing the status-function, the Y term "war" was not
applied. By the time of the Vietnam War, these sorts of evasions
had been abandoned and the sheer physical and intentional facts
warranted the application ofthe term "war," even though the legal
situation was no more that of a declared war than had been the
case in Korea.
"War" thus oscillates between naming a type of large-scale so-
cial fact and a type of institutional fact. The test for the distinction
is whether the term "war" is used to label an existing set of rela-
tions or whether it implies further consequences that derive from
its recognized status as a "war." This is related to how the war
came to exist. War as social fact can exist no matter how it came
about, but under the U.S. Constitution, war as an institutional fact
exists only if it is created by an act of Congress, a type of speech act
I call a Declaration. Perhaps after Vietnam and the Persian Gulf
001067
The General Theory of Institutional Facts (Part 1) 91
erty while the police pointed their guns at them and ordered them
to stop. The looters simply ignored the police, with no further con-
sequences. "Why are you doing this?" asked one reporter. "It's
free," the thief replied. All this was watched by millions on televi-
sion. The police power of the government is usable only against
very small numbers, and even then on the assumption that nearly
everyone else accepts the systems of status-functions. Once the
number of lawbreakers is more than tiny, the police typically re-
treat to the station house, or put on a ceremonial show of acting as
if they were enforcing the law, as in Los Angeles, or quite often ar-
rest the law-abiding citizenry. In Berkeley during the same period
of rioting and looting, a store owner was arrested because he had
armed himself with the intent of defending his store, and this ar-
rest occurred while looters robbed nearby stores unhindered by
the police. In many democratic societies, once the number of law-
breakers reaches critical mass, the police force is largely for
show.*
The point for our present discussion is that we cannot assume
that the system of acceptance is backed by a credible system of
force. For one thing the system of force is itself a system of accep-
tance. Police forces and armies, for example, are systems of
status-functions. But more important for our present purposes,
the system of force presupposes the other systems of status-
functions. We cannot assume that Leviathan will come to our aid
in a genuine crisis; on the contrary, we are in a state of nature all
the time, but the state of nature is precisely one in which people
do in fact accept systems of constitutive rules, at least nearly all the
time.
More spectacular examples are provided by the collapse of the
Soviet empire in the annus mirabilis, 1989. Anyone who visited the
'I originally became aware of this during my first term as an undergraduate at
Oxford, when I attended the annual Guy Fawkes riots of that era. The Proctors
and Bulldogs apprehended me, a passive spectator, rather than confront the ac-
tual participants, who were much too dangerous.
001069
countries of the Soviet empire over the generation prior to 1989
could see that the whole thing was maintained only by a system
of terror. Most people did not think that the system of status-
functions was morally acceptable, much less socially desirable.
But there did not seem to be anything that anyone could do about
it because the whole system was maintained by an elaborate ap-
paratus of police powers backed by the armed might of the Soviet
military forces. Efforts at reform such as the Czechoslovak
"Prague Spring" of 1968 were brutally crushed by the Soviet Army
with the help of the domestic secret police. In Czechoslovakia,
every tenth person was made to spy on the other nine and report
any sign of disaffection to the secret political police. In the GDR
the system of police surveillance was even more thorough and
ruthless, to the point that even husbands and wives were forced to
report on each other. No one-no qualified expert on the Soviet
system, no diplomat, no journalist, and no tourist-could predict
in the mid-1980s that the whole system would collapse suddenly
and within a few years. It collapsed when the system of status-
functions was no longer accepted. The fear of Soviet intervention
was no longer credible, and the indigenous police and military
were unwilling to attempt to maintain the system. In the GDR the
army refused to fire on the opposition even when ordered to do so.
I do not believe there is any single motivation for the continued
acknowledgment of institutional facts. It is tempting to some to
think that there must be some rational basis for such acknowledg-
ment, that the participants derive some game theoretical advan-
tage or get on a higher indifference curve, or some such, but the
remarkable feature of institutional structures is that people con-
tinue to acknowledge and cooperate in many of them even when it
is by no means obviously to their advantage to do so. When institu-
tions are maintained largely by habit, they can also collapse quite
suddenly, as when people lose confidence in their currency or
cease to recognize their government as a government.
Marx, believing that the most fundamental interests were class
interests, said that all history is the history of class struggle. But
001070
The General Theory of Institutional Facts (Part l) 93
the surprising thing is how little of history is about class struggles.
In the great upheavals of the twentieth century, for example, na-
tionalloyalties proved much more powerful than class solidarity,
and conationals of all classes slaughtered enemy nationals of all
classes with passion and enthusiasm. International class solidar-
ity counted for next to nothing. And in most of these great up-
heavals, the systems of constitutive rules that sustained the class
distinctions were preserved, even though all sorts of other institu-
tional changes took place; and in places where the institutional
structures sustaining the class structure were destroyed-for
example, Russia in the first war, China after the second-their de-
struction was not one of the war aims of their enemies. Imperial
Germany was not out to create a Bolshevik state in Russia, nor was
Maoism an objective of the Greater East Asia Coprosperity Sphere.
The point I am trying to illustrate is that there is no simple set of
relations among motivation, self-interest, institutional structure,
and institutional change.
Perhaps the most amazing form of status-function is in the cre-
ation of human rights. Prior to the European Enlightenment the
concept of rights had application only within some institutional
structure-property rights, marital rights, droit de seigneur, etc.
But somehow the idea came to be collectively accepted that one
might have a status-function solely by virtue of being a human
being, that the X term was "human" and the Y term was "possessor
of inalienable rights." It is no accident that the collective accep-
tance of this move was aided by the idea of divine authority: "they
are endowed by their Creator with certain unalienable rights, that
among these are Life, Liberty and the pursuit of Happiness." The
idea of human rights has survived the decline of religious belief,
and has even become internationalized. The Helsinki Declaration
on Human Rights is frequently appealed to, with varying degrees
of effectiveness, against dictatorial regimes. Lately there has even
been a movement for the recognition of animal rights. Both
human and animal rights are cases of the imposition of status-
function through collective intentionality.
001071
In general status-functions are matters of power, as we will see
in the rest of this chapter. The structure of institutional facts is a
structure of power relations, including negative and positive, con-
ditional and categorical, collective and individual powers. In our
intellectual tradition since the Enlightenment the whole idea of
power makes a certain lype of liberal sensibility very nervous. A
certain class of intellectuals would rather that power did not exist
at all (or if it has to exist they would rather that their favorite op-
pressed minority had lots more of it and everyone else had lots
less). One lesson to be derived from the study of institutional facts
is this: everything we value in civilization requires the creation
and maintenance of institutional power relations through collec-
tively imposed status-functions. These require constant monitor-
ing and adjusting to create and preserve fairness, efficiency,
flexibility, and creativity, not to mention such traditional values as
justice, liberty, and dignity. But institutional power relations are
ubiquitous and essential. Institutional power-massive, perva-
sive, and typically invisible-permeates every nook and cranny of
our social lives, and as such it is not a threat to liberal values but
rather the precondition of their existence.
Some Types of Imposition of
In order to investigate the logical structure of institutional reality,
I first want to ask: what sorts of new facts, new powers, new causal
structures can people create by creating status-functions, when
status-functions exist only because they are believed to exist?
Where physical functions are concerned the only limitations
are provided by the sheer physical possibilities. The history of
technology is the history of how accumulated knowledge and or-
ganized desires have utilized technical possibilities. But when it
comes to institutional facts, improvements in technology do not
change the possibilities. We cannot impose an electrical charge
just by deciding to count something as an electrical charge, but
we can impose the office of the Presidency just by deciding what
001072
we will count as becoming President, and then making those peo-
ple President who meet the conditions we have decided on. The
intensionality-with-an-s of the sentence form "X counts as Y in C"
is a clue to the intentionality-with-a-t of the phenomena. Because
neither the X term nor theY term permits substitution of corefer-
ring expressions without loss or change of truth value of the
whole statement, we have good reasons to suppose that the
"counts as" locution specifies a form of intentionality. The possi-
bilities of creating institutional facts by the use of this formula are
limited by the possibilities of imposing new features on entities
just by collectively agreeing that they have those features. Our
question now is, What are the forms and limits of the institutional
imposition of function?
At first sight institutional facts seem utterly bewildering in their
variety. We can make promises, score touchdowns, get tenure, be-
come President, adjourn the meeting, pay our bills, and fire our
employees all by way of institutional facts. But within this enor-
mous variety of subject matter there are actually only a very few
general formal properties of institutional facts.
Because the creation of institutional facts is a matter of im-
posing a status and with it a function on some entity that does
not already have that status-function, in general the creation of a
status-function is a matter of conferring some new power. There
would not be much point to imposing the status-function named
by the Y term if it did not confer some new power on the X term,
and most (not all) creations of institutional facts are precisely
conferring powers on the X term, or performing some truth
functional operation such as negation and conditionalization on
the creation of the power. In the simplest case, theY term names
a power that the X term does not have solely in virtue of its X
structure. In cases where the X term is a person, that person ac-
quires powers that he or she did not already have. In cases
where the X term is an object, the user of that object can do
things with it that he or she could not do solely in virtue of its X
structure. Thus money, passports, driver's licenses, and sen-
001073
Their opposites are matters of negative honors. Thus, if you are
censured for your bad behavior, reprimanded by your superiors,
or voted the least popular in your class, these are all negative
honors. No further powers, positive or negative, need apply.
Our question is, In the formula "X counts as Y in C/' how many
types of "Y"s are there? Because institutional facts are structured
by collective intentionality and because there are strict limitations
on the possibilities of creating institutional facts, we ought to be
able to answer this question. So let us begin naively by listing some
formal features of institutional reality.
The Y status can be imposed on several different ontological
categories of phemomena: People (e.g., chairmen, wives, priests,
professors); objects (e.g., sentences, five dollar bills, birth certifi-
cates, driver's licenses); and events (elections, weddings, cocktail
parties, wars, touchdowns). The people, objects, and events in-
teract in systematic relationships (e.g., governments, marriages,
corporations, universities, armies, churches). Often theY status
is imposed on people and groups of people in virtue of a set of
preexisting preinstitutional relations among them. Thus a collec-
tion of people might constitute a city-state, or a man and woman
might constitute a married couple, but such constitution is not
simply in virtue of being a collection of people of the right size,
but rather in virtue of the relations among the members of the
collection.
What then are the features of objects, events, and people that
are imposed by the new status-functions? My first suggestion is
that the category of people, including groups, is fundamental in
the sense that the imposition of status-functions on objects and
events works only in relation to people. This should not be sur-
prising, since it is a general feature of agentive functions. It is not
the five dollar bill as an object that matters, but rather that the
possessor of the five dollar bill now has a certain power that he or
she did not otherwise have. Just so, it is not the screwdriver as an
object that matters, but rather that the possessor of the screw-
driver now has a power that he or she did not otherwise have.
001075
This suggests what I think is in fact the case, that the content of the
collective intentionality in the imposition of the status-function
will typically be that some human subject, singular or plural, has
some power, positive or negative, conditional or categorical. This
will be directly the case where the status is imposed on an agent,
as in, e.g., Jones is President, and indirectly the case where the
status is imposed on an object, as in, e.g., this is a five dollar bill.
Another formal feature to note is that the usual distinction be-
tween the internal and the external points of view applies to insti-
tutional facts. In this book we are interested primarily in the
internal point of view, because it is only fmm the internal point of
view ofthe participants that the institution can exist at all. The an-
thropologist from outside the institution may see the potlatch, for
example, as performing functions of which the Kwakiutl partici-
pants are totally unaware, but the whole feast is a potlatch in the
first place only because of the collective intentionality and the im-
position of status-functions by the participants, and this, whether
conscious or unconscious, can exist only fmm the internal first-
person point of view.
Even within the internal point of view there are some formal
distinctions to be made. At the microlevel the individual sees
money as a medium of exchange and store of value and he or she
sees marriage as a collective lifetime promise between a male and
a female partner. But at a macrolevel, planners and organizers,
even from an internal point of view, see the institutions as having
different functions, though the status assigned in individual cases
is the same. The bishop sees the function of marriage as glorifying
God and producing social stability, and the central bank sees the
supply of money as a way of controlling the economy. The impor-
tant point is that the internal microlevel is ontologically primary.
There is no way that the bishop, the head of the Federal Reserve
Board, and the anthropologist can have their points of view with-
out the lowest-level participants in the very trenches of money
and marriage having the basic form of intentionality that consti-
001076
tutes the structure of institutional facts. Furthermore, the mi-
crolevel participants may have all sorts of other functions that
they want the institutional entities to perform for them that are ir-
relevant to the basic ontology. Thus many people want money for
power and prestige, and for them that is a basic function of
money. Among the ruling dynasties of Europe marriage was a tool
of dynastic power. (Alii bella gerunt, tu feli?C. Austria nube: Happy
Austria, other nations wage wars, you marry.) And marriage has
performed such ulterior functions even among the humble. The
point is that all this works only given the basic ontology of the
ordinary collective intentionality in the street, so to speak, impos-
ing a status-function according to the formula.
If we look at institutional facts with these points in mind, it
seems to me that status-functions fall into certain broad cate-
gories. As a first stab at classifying these-we will have to make im-
provements later-! will provisionally divide them into four broad
categories, which I call Symbolic, Deontic, Honorific, and Proce-
dural.
1. Symbolic Powers: The Creation of Meaning
The point of having symbolic powers is to enable us to represent
reality in one or more of the possible illocutionary modes. In such
cases we impose intentionality on entities that are not intrinsically
intentional. And to do this is to create language and meaning in all
its forms. The imposition of the intentionality on a certain type of
physical structure determines both a formal structure-the syn-
tal(.-and a meaningful content-the semantics. Thus, for exam-
ple, the phonetic/graphemic type "II pleut" counts as a sentence of
French, and "Es regnet" counts as a sentence of German. On phys-
ical sounds and marks we impose the statuses word, sentence,
and syntax generally. And on different syntactical objects we have
in these cases imposed the same semantic content. They both
mean "it's raining." Symbolization is essential to the other forms of
001077
quired to do as a result of the assignment of status specified in the
Y term. Here are some examples:
John has one thousand dollars in the bank.
Tom is a citizen of the United States.
Clinton is President.
Sally is an attorney.
Sam owns a restaurant.
Each of these assigns rights and responsibilities. The first example
assigns to John the right to buy things or employ people with his
money and the duty to pay taxes on interest earned by the money.
The second example assigns to Tom the right, among many oth-
ers, to vote in elections and the obligation, among many others, of
getting a Social Security number. The third example assigns to
Clinton the right to veto legislation and the responsibility of deliv-
ering a State of the Union address to Congress, etc. Notice also that
institutional facts that assign rights and responsibilities can also be
destroyed or eliminated in various ways. Here are some examples:
Ann lost all her money.
Ivan's fortune in rubles has become worthless through infla-
tion.
Nixon resigned from office.
Coolidge's term expired.
Sam got divorced.
Sally's husband died.
3. Honor: Status for Its Own Sake
The point of honors (and dishonors) is to have statuses valued (or
disvalued) for their own sake, rather than just for their further
consequences. Examples are victory and defeat in games, and in-
stitutionally sanctioned forms of public honor and disgrace. Here
are some examples:
001079
Mark won the Far West skiing championship.
McCarthy was censured by the U.S. Senate.
Bill was awarded a medal by the College de France.
In addition to these three categorical types of status-functions, we
need also to identify the conditional or procedural features of de-
on tic powers and honors.
4. Procedural Steps on the Way to Power
and Honor
Within institutions we can assign procedural stages toward the
achievement of either rights and responsibilities or honors and
disgraces. Here are some examples:
Bill voted for Reagan.
Clinton was nominated the Democratic candidate for President.
The objection was sustained by the judge.
In the case of voting, though one has a right to vote, the actual
casting of the vote does not create any new rights and responsibil-
ities by itself. Only the accumulated set of votes establishes a win-
ner with the requisite majority and therefore with new rights and
responsibilities. Getting six votes is like getting six points in a foot-
ball game, but unlike getting six dollars. Six votes and six points
are procedural steps on the way to winning, but you can't do any-
thing else with them. With six dollars you can actually buy some-
thing. Again, when a candidate is nominated for the Presidency,
he or she does acquire new rights and responsibilities as a candi-
date, but the whole point of the candidacy is supposed to be that it
is a stage on the way to becoming President.
One and the same institutional fact can involve all four of the pre-
ceding features. Thus becoming a Democratic nominee gives the
person so nominated certain rights and responsibilities, it is a
001080
The General Theory of Institutional Facts (Part 1) 103
great honor, and it is a procedural stage on the road to becoming
President, and the whole thing could not exist without words OI'
other sorts of symbols, as I explained in Chapter 3.
I want to illustrate these points by showing how they apply to
the case of games. Games are especially useful objects of study
for this analysis because they provide a microcosm of larger so-
cial phenomena. Famously, Wittgenstein a1gued that there is no
essence marked by the word "game." But all the same, there are
certain common features possessed by paradigmatic games such
as those in competitive sports-baseball, football, tennis, etc. In
each case the game consists of a series of attempts to overcome
certain obstacles that have been created for the purpose of trying
to overcome them. Each side in the game tries to overcome the
obstacles and prevent the other side from overcoming them. The
rules of the game specify what the obstacles are and what can be
done to overcome them, as well as what must and what must not
be done. Thus in baseball the rules allow the batter to swing at the
ball, but they do not require him to swing. However, after he gets
three strikes he must leave the batter's box and let someone else
bat. Most of the rules of the game have to do with rights and
obligations (feature 2) but the overall aim is winning (feature 3)
and many of the intervening steps are procedural (feature 4). For
example, several of the rights and obligations are conditional.
Thus if a batter has one strike or three balls, that does not so far
give him any further rights or obligations, but it establishes condi-
tional rights and obligations: two more strikes and he is out, one
more ball and he is walked to first base. Such conditional rights
and obligations are typical of institutional structures. For exam-
ple, in American universities, after so many years of service you
are entitled to be considered for a tenure position.
'This answer to Wittgenstein on games was not invented by me. I do not know
who first thought of it or where I first heard it, but it has become part of the oral
tradition.
001081
104 Tile Construction ofSocial Reality
1'he IJogical Structure of Conventional Power
To further explore the issues raised in the tentative taxonomy of
the previous section, I now want to examine the intentional struc-
ture of institutional facts. My aim is to try to state the general form
of the content of the Y status-function when we go from X to Y in
the formula "X counts as Yin C." Because theY content is imposed
on the X element by collective acceptance, there must be some
content to these collective acceptances (recognitions, beliefs, etc.);
and I am suggesting that for a large class of cases the content
involves some conventional power mode in which the subject is
related to some type of action or course of actions. Furthermore,
because there are strict limits on what sorts of powers can be cre-
ated by collective acceptance, we ought to be able to state the gen-
eral forms of the content of the Y term in a very small number of
formulae. Because power is always the power to do something or
constrain someone else from doing something, the propositional
content of power status-functions is always in part that
(S does A)
where "S" is to be replaced by an expression referring either to a
single individual or a group, and ''/\' by the name of an act, action,
or activity, including negatives such as refraining or abstaining.
Following this line of thought, we see that the primitive struc-
ture of the collective intentionality imposed on the X term, where
X counts as Yin C, is
We accept (S has power (S does A)).
Formally speaking, one can perform a number of operations on
this basic structure, and these operations exemplify several dis-
tinctions I have made. As mentioned earlier, there is a distinction
between positive and negative conventional powers, the distinc-
tion between enablements and requirements. There is also a dis-
tinction between the creation and destruction of conventional
powers. Examples of this are the distinctions between marriage
001082
The General Theory of Institutional Facts (Part Il) 117
71w Continued Exislence of Institutional Facls
The secret of understanding the continued existence of institu-
tional facts is simply that the individuals directly involved and a
sufficient number of members of the relevant community must
continue to recognize and accept the existence of such facts. Be-
cause the status is constituted by its collective acceptance, and be-
cause the function, in order to be performed, requires the status,
it is essential to the functioning that there be continued accep-
tance of the status. The moment, for example, that all or most of
tl1e members of a society refuse to acknowledge property rights,
as in a revolution or other upheaval, property rights cease to exist
in that society.
One of the most fascinating-and terrifying-features of the
era in which I write this is tl1e steady erosion of acceptance of
large institutional structures around the world. The breakdown
of national identification in favor of ethnic tribalism occurs in
places as various as Bosnia, Canada, tl1e former Czechoslovakia,
Turkey, and many American universities. In several African coun-
tries there is no way to tell where the army ends and tl1e armed
bands begin or who is a "military leader" and who a "warlord." In
Russia the instability is such tl1at anything one might say with con-
fidence now-about tl1e relations among the state, the military,
the secret police, and organized crime, for example-is likely to
be out of date by the time you read this. The temptation in all these
cases is to think that in the end it all depends on who has the most
armed might, that brute facts will always prevail over institutional
facts. But that is not really true. The guns are ineffectual except to
those who are prepared to use them in cooperation with others
and in structures, however informal, with recognized lines of au-
thority and command. And all of that requires collective inten-
tionality and institutional facts.
One of the great illusions of the era is that "Power grows out of
the barrel of a gun." In fact power grows out of organizations, i.e.,
systematic arrangements of status-functions. And in such organi-
001083
zations the unfortunate person with a gun is likely to be among
the least powerful and the most exposed to danger. The real
power resides with the person who sits at a desk and makes
noises through his or her mouth and marks on paper. Such peo-
ple typically have no weapons other than, at most, a ceremonial
pistol and a sword for dress occasions.
Because institutions survive on acceptance, in many cases an
elaborate apparatus of prestige and honor is invoked to secure
recognition and maintain acceptance. Charles de Gaulle's behav-
ior regarding France, both during and after World War II, was a
continuous illustration of these points. By constantly insisting on
the honor and prestige of France, by pretending that an indepen-
dent French government continued to exist during the war, and
by constantly insisting that other national leaders acknowledge
him as an equal, de Gaulle helped to re-create and maintain the
French nation-state. And the point is completely general. Where
the institution demands more of its participants than it can extract
by force, where consent is essential, a great deal of pomp, cere-
mony, and razzmatazz is used in such a way as to suggest that
something more is going on than simply acceptance of the for-
mula X counts as Y in C. Armies, courtrooms, and to a lesser ex-
tent universities employ ceremonies, insignia, robes, honors,
ranks, and even music to encourage continued acceptance of the
structure. Jails find these devices less necessary because they
have brute force.
One way to create institutional facts in situations where the in-
stitution does not exist is simply to act as if it did exist. The classic
case is the Declaration of Independence in 1776. There was no in-
stitutional structure of the form X counts as Y in C, whereby a
group of the King's subjects in a British Crown Colony could cre-
ate their independence by a performative speech act. But the
Founding Fathers acted as if their meeting in Philadelphia was a
context C such that by performing a certain declarative speech act
X they created an institutional fact of independence Y. They got
away with this, that is, they created and sustained acceptance of
001084
The General Theory of Institutional Facts (Part Il) 119
the institutional fact because of local community support and mil-
itary force, culminating in Cornwallis's surrender at Yorktown.
The formula "X counts as Y" applies to both the creation and the
continued existence ofthe phenomenon, because the constitutive
rule is a device for creating the facts, and in generaC the existence
of the fact is constituted by its having been created and not yet de-
stroyed. Thus going through the ceremony counts as getting mar-
ried, and getting married and not subsequently dying, getting
divorced, or having the marriage annulled counts as being mar-
ried. Saying "I declare the parliament open" counts as opening the
parliament, and for the parliament to have been opened and not
subsequently closed counts as its being in session.
Status Indicators
Since institutional facts exist only by human agreement, in many
cases they require official representations, what I earlier called
status indicators, because the existence of institutional facts can-
not in general be read off from brute physical facts of the situa-
tion. War is an exception for the obvious reason that the brute
facts-people killing one another on a large scale, for example--
usually make official indicators unnecessary. Money does notre-
quire additional documentation, because it is itself a form of
documentation. It says on the bill that it is "One Dollar" or "Ten
Pounds," etc., and all these terms are defined as money. Even in
preliterate societies coins are easily recognizable as such, and
thus such features as shape and size mark the conventional fact
that the object is a coin. Stocks and bonds, as well as credit cards
and checks, also speak for themselves. Likewise, speech acts are
self-identifYing for those who know the language.
In complex societies common status indicators are passports
and driver's licenses. They indicate the status of the bearer as
someone who is legally entitled to travel to and from foreign coun-
tries or is legally qualified to drive. The most common device for
status indication is the written signature. Signing a document may
001085
create a new institutional fact, but the continued existence of the
written signature indicates, other things being equal, the contin-
ued existence of the fact. The signature on the document persists
in a way that the live pel'formative does not and thus is able to play
its role as a status indicator. The function of status indicators is al-
ways epistemic. We need to distinguish the role of language in
constituting the institutional fact, a role I described in Chapter 3,
from the role of language in identifying that which has already
been constituted, even though the same word OJ' symbol may
serve both roles. I am describing this latter role when I speak of
status indicators.
Some status indicators need not be explicitly linguistic, that is,
they need not be actual words. The most obvious examples are
wedding rings and uniforms. But both are nonetheless symbolic
in a way that is just like language, and wearing a wedding ring or
a uniform is performing a type of speech act. Such indicators
serve not only epistemic functions but other functions as well-
expressive, ceremonial, aesthetic, and most importantly, constitu-
tive. Of course, the uniform does not constitute being a
policeman, but it does symbolize a status-function; and that sym-
bolization, in some form or other, is essential to the existence of
the status function. Throughout this book I have tried to empha-
size that in institutional facts language is not only descriptive but
constitutive of reality.
The Hierarchy of li'acts: From Brute
to Institutional
There is an implicit hierarchical taxonomy in the account I have
been giving, and I would now like to try to make it explicit. The
world of Supreme Court decisions and of the collapse of commu-
nism is the same world as the world of the formation of planets
and of the collapse of the wave function in quantum mechanics.
One of the aims of this book is to show how that can be so, how
the world of institutions is part of the "physical" world. A hierar-
001086
Conclusion
One way to get at the underlying thrust of what I have been argu-
ing in this book is this: On my view the traditional opposition that
we tend to make between biology and culture is as misguided as
the traditional opposition between body and mind. Just as mental
states are higher-level features of our nervous system, and conse-
quently there is no opposition between the mental and the physi-
cal, the mental is simply a set of physical features ofthe brain at a
higher level of description than that of neurons; so there is no op-
position between culture and biology; culture is the form that bi-
ology takes. There could not be an opposition between culture
and biology, because if there were, biology would always win. Dif-
ferent cultures are different forms that an underlying biological
substructure can be manifested in. But if that is right, then there
ought to be a more or less continuous story that goes from an on-
tology of biology to an ontology that includes cultural and institu-
tional forms; there should not be any radical break. The thesis I
have been arguing is that there is no radical break. The connect-
227
001087
228 Conclusion
ing terms between biology and culture are, not surprisingly, con-
sciousness and intentionality. What is special about culture is the
manifestation of collective intentionality and, in particular, the
collective assignment of functions to phenomena where the func-
tion cannot be performed solely in virtue ofthe sheer physical fea-
tures of the phenomena. From dollar bills to cathedrals, and from
football games to nation-states, we are constantly encountering
new social facts where the facts exceed the physical features of the
underlying physical reality.
However, though there is a continuum from the chemistry of
neurotransmitters such as seretonin and norepinephrine to the
content of such mental states as believing that Proust is a better
novelist than Balzac, mental states are distinguished from other
physical phenomena in that they are either conscious or poten-
tially so. Where there is no accessibility to consciousness, at least
in principle, there are no mental states. Similarly, though there is
a continuity in collective behavior between lions attacking a hyena
and the Supreme Court making a constitutional decision, institu-
tional structures have a special feature, namely, symbolism. The
biological capacity to make something symbolize-or mean, or
express-something beyond itself is the basic capacity that under-
lies not only language but all other forms of institutional reality as
well. Language is itself an institutional structure because it in-

valves the imposition of a special kind of function on brute physi-
cal entities that have no natural relation to that function. Certain
sorts of sounds or marks count as words and sentences, and cer-
tain sorts of utterances count as speech acts. The agentive func-
tion is that of representing, in one or other of the possible speech
act modes, objects and states of affairs in the world. Agents who
can do this collectively have the fundamental precondition of all
other institutional structures: Money, property, marriage, govern-
ment, and universities all exist by forms of human agreement that
essentially involve the capacity to symbolize.
'
001088

Appendix Tabs 54-61

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Appendix Tabs 54-61

Uploaded by

Copyright:

Available Formats

Tab

, that is representative of the US population. Members of the KnowledgePanel

are randomly recruited

survey is 65%. Appendix A presents a comparison of age-appro-

) beside it indicate that the groups dichotomous variable estimate

0.41 0.37 0.39

Currently employed full-time 0.49 0.26

0.34 0.41 0.42 0.47

0.15 0.14 0.13

0.07 0.08 0.10 0.05 0.09

0.25 0.20 0.12

0.16 0.12 0.16

Family-of-origin safety/security 4.13 3.12

3.58 3.95 4.03 3.85

1.95 2.01 1.91

Current relationship quality index 4.11 3.83 3.63

3.79 3.95 3.80

2.35 2.43 2.35

3.49 3.31 3.33 3.43

1.19 1.30 1.21

0.55 0.42 0.44

0.49 0.37 0.41 0.27 0.21

0.47 0.63 0.41 0.45

Family received welfare growing up 0.17 0.72

Currently on public assistance 0.10 0.32 0.49

0.17 0.10 0.12 0.17

0.13 0.09 0.13 0.01

Is in a same-sex romantic relationship 0.04 0.12 0.02 0.14 0.05 0.15

0.05 0.01 0.04 0.13

0.23 0.18 0.23 0.18 0.20 0.30

0.12 0.12 0.17

0.17 0.08 0.16 0.12

0.13 0.17 0.17 0.11 0.12 0.10 0.23

2.78 2.92 3.21

Family-of-origin safety/security 4.13 3.23

Family-of-origin negative impact 2.30 3.30

3.88 3.90 3.63 3.50

4.20 3.87 4.17 3.58 3.79

3.53 3.29 2.77

3.54 3.66 3.53 3.54

3.27 3.10 3.08 3.00 3.12

2.84 3.26 3.22

5.96 7.08 7.42 7.46

3.95 3.88 3.94 3.92 3.65 3.66

3.92 3.77 4.36

2.21 2.47 2.43 2.15 2.32

2.60 2.31 1.85 2.35 2.31

1.70 1.50 1.62 1.33 1.50

2.63 1.89 2.55 2.59 2.74 2.84

1.78 1.37 1.73 1.32 1.73 1.68

2.22 2.44 2.25

2.03 2.31 2.38 2.27 2.14 1.90 2.59 2.34

3.46 3.17 3.33 3.21 3.24 3.47 3.98

3.50 3.51 3.37 2.27

1.52 1.38 1.39

Freq pled guilty to non-minor offense 1.10 1.43

1.36 1.30 1.20 1.21

0.94 0.52 0.36 0.47 0.47

N of female sex partners (among men) 2.70 2.37

3.43 3.24 3.60

0.98 0.37 0.10

You might also like