Professional Documents
Culture Documents
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
National Council on Family Relations is collaborating with JSTOR to digitize, preserve and extend access to
Journal of Marriage and Family.
http://www.jstor.org
631
632
633
4. When heterogeneity is controlled, the relationship between age at first marriage and
marital instability is negative and linear.
5. The higher the age at first marriage, the
greater the effect of heterogeneity on marital
instability.
DATA AND PROCEDURES
Data for this study come from a national probability sample of married persons under 55 years
of age. The data were originally gathered to assess
the effects of female labor-force participation on
marital stability. All interviews were done by telephone; a random digit dialing procedure was used
to locate households with eligible respondents.
The response rate among eligible households was
65%.
The sample characteristics were compared by
the original researchers with the estimates of the
U.S. Census and the sample was found to be highly representative of young and middle-aged married people within the continental United States
with respect to age, race, household size, presence
of children, region, and female labor-force participation (Booth, Johnson, and White, 1981). To
control for the possible bias resulting from the
usual overrepresentation of women and more
highly educated persons, sex and education were
introduced as control variables. Since the effect of
age at first marriage can be expected to decrease
with duration of the marriage, the exclusion of
married persons over 55 years of age does not
seem to be a problem either.
In studies of marital instability, both the inclusion and the exclusion of divorced people from
the sample can introduce methodological problems. Inclusion poses the problem of retrospective
data, whereas exclusion may cause bias and
underestimation. In this research the explanatory
analysis excludes divorced people and is based on
a sample of spouses in their first marriage, consisting of 691 men and 1,018 women. The expected underestimation of effects and possible
bias resulting from the attrition of divorced
couples is assessed by comparing the subsample of
persons in their first marriage to the total sample
of currently married persons (814 men and 1,213
women).
Dependent Variable
In the literature on the effects of the timing of
marriage, either dissolution by the spouses
(through divorce or separation) or marital satisfaction is treated as the dependent variable. However, methodological and theoretical problems are
associated with both of these variables. If divorce
or separation is taken as the dependent variable,
634
the focus is in fact on the consequence of instability rather than on instability itself. Moreover, in
such cases the data source almost always prevents
the inclusion in the analysis of characteristics such
as social class, race, religion, or premarital pregnancy as control variables. On the other hand,
treating marital satisfaction as the dependent variable introduces conceptual confusion. Research
has shown that many dissatisfying marriages remain intact because of religious commitments or
unemployment and also that high satisfaction
does not always preclude voluntary dissolution
(Landis, 1963; Booth and White, 1980). In other
words, the contingencies associated with marital
satisfaction and stability may well be very different.
A marital instability index has been developed
by Booth, Johnson, and Edwards (1983) to
remedy these problems. In this scale, marital instability is treated as the propensity to divorce,
which includes both a cognitive state (thinking the
marriage is in trouble and considering the idea of
getting a divorce) and actions related to these
thoughts. The scale is based on the following
items:
1. Have you ever thought your marriage
might be in trouble?
2. Have you ever talked with family members
or friends about problems in your marriage?
3. As far as you know, has your spouse talked
with relatives or friends about problems
either of you were having with your marriage?
4. As far as you know, has your spouse ever
thought your marriage was in trouble?
5. Has the thought of getting a divorce or
separation crossed your mind?
6. As far as you know, has the thought of getting a divorce or separation crossed your
spouse's mind?
7. Have you or your spouse ever seriously suggested the idea of divorce?
8. Have you talked about the problems of living apart?
9. Have you talked about consulting an attorney?
10. Have you talked about filing?
11. Have you or your spouse filed a divorce or
separation petition?
12. Have you discussed a divorce or separation
with members of your family?
13. Have you discussed a divorce or separation
with a close friend?
14. Have you ever left home because of marital
problems-either for a short time or as a
trial separation?
635
(a)
(b)
d*m-
4-
()
(0
Age
Y=a+bl x+b2x2
Age
y=a+b
x+b2x2
(c)
4-
(d)
4-.
0m-
(0
V)
.C
Age
ya+bl x+b2x2
used. This technique is suitable for the examination of curvilinear relationships and allows for the
simultaneous introduction of many control variables. In addition, this method permits the treatment of age at first marriage as a continuous variable, thus eliminating the need for an arbitrary
categorization.1
The test of the main hypothesis of this research
occurs as follows. The relationship between age at
marriage and marital instability is expected to
look like the curve in Figure la. If heterogeneity
Age
y=a+bx
accounts as much for the instability of early marriages as it does for the instability of late marriages, the introduction of heterogeneity into the
regression equation will not change the shape of
the curve depicting the relationship (Fig. lb).
However, if heterogeneity is helpful in explaining
the instability of late marriages but not of the
early ones, statistical control for heterogeneity
will "flatten" the curve (Fig. lc), possibly to the
extent that it becomes a straight line (Fig. ld). As
the equations for these curves show, this means
636
flexibility is evaluated by introducing multiplicative interaction terms into the regression equation.
FINDINGS
Tables 1 and 2 compare those variables associated with previous divorce in the total sample of
married people with the predictors of marital instability in the subsample of persons in their first
marriage. Because the total sample consists of
married people only, the variables in the first
equation do not simply predict divorce but divorce and subsequent remarriage. Thus they also
Regression
Coefficient
Variablea
Sex of respondent (m = 1, f = 2)
Length of marriage
Polynomial
Religion
None (omitted category)
Protestants
Catholics
Jews
Other
Age at first marriage
Polynomial
(Constant)
Beta
Standard
Error of b
-.060***
-.020
.0003**
**
-.082
-.515
.239
.016
.003
.000
.049
.003
.120*
-.025
-.156
.003***
3.407
.068
.004
.053
-.016
-1.492
1.240
.030
.031
.054
.041
.018
.000
R2 = .167; N = 2,012
aNot significant: Race, social-desirability response tendency, social class of respondent, education of respondent,
social class of respondents' parents, and interactions between sex and race, religion, education, social-desirability
response tendency, length of marriage, age at first marriage, and the polynomial term of age at first marriage.
*p < .05. **p < .01. ***p < .001.
TABLE
2. VARIABLES ASSOCIATED WITH MARITAL INSTABILITY IN A SAMPLE OF PEOPLE IN THEIR FIRST INTACT MARRIAGE (HETEROGENEITY VARIABLES EXCLUDED)
Variablea
Cohabited? (y = 1, n = 2)
Length of courtship (months)
Polynomial
Social-desirability response tendency
Length of marriage (years)
Polynomial
Race
Caucasian (omitted category)
Hispanics
Blacks
Others
Age at first marriage
Polynomial
(Constant)
Regression
Coefficient
Beta
Standard
Error of b
-3.315***
-.117
.001**
-1.062***
.226
-.010***
-.107
-.211
.203
-.110
.207
-.293
.794
.039
.001
.236
.097
003
.003
.064
-.041
-.523
.447
1.096
1.112
1.771
.615
.013
.133
2.949**
-3.030
-1.554
.028*
35.667
R2 = .058; N= 1,659
Note: The polynomial term of age at first marriage and the variable "sex of respondent," both significant by
themselves, share variance to such an extent that neither one is significant if the other variable is included in the
regression. On theoretical grounds, sex of respondent was excluded from the regression.
aNot significant: Social class of spouses and social class of spouses' parents.
*p < .05. **p < .01. ***p < .001.
637
Heterogeneity in:
Age (in years)
Religious affiliation
when dating
Previous marital
status
Years of education
Social class of
parents
Age at First
Marriage
.154
p=.000
(1,723)
-.014
p = .280
(1,720)
.176
p = .000
(1,725)
.124
p=.000
(1,722)
.105
p= .000
(1,372)
Marital
Instability
.060
p =.000
(1,716)
.067
p = .003
(1,713)
.124
p= .000
(1,717)
.021
p =.195
(1,714)
.056
p = .019
(1,372)
Heterogeneity with respect to years of education shows a reverse pattern: here the relationship
with age at marriage follows the prediction, but
there is no significant impact of dissimilar education on marital instability. However, if the direction of the heterogeneity is taken into accountthat is, if a distinction is made between the cases
in which the husband and those in which the wife
has more education, a very different picture
emerges. Such a dichotomy of the variable measuring educational differences is presented in
Table 4, in which all cases are represented by
either the value of the difference in years of
education if the category is appropriate or by a 0
in case the category does not apply.
TABLE 4. DIFFERENCES IN EDUCATION, AGE AT FIRST
MARRIAGE, AND MARITAL INSTABILITY
Difference in Years
of Education
Husband superior
Wife superior
Age at First
Marriage
.071
p=.002
(1,722)
Marital
Instability
-.033
p = .085
(1,714)
.086
p=.000
.083
p =.002
(1,722)
(1,714)
638
TABLE
Beta
Standard
Error of b
-2.624**
-.093
.001*
.250
-.011***
-1.120***
*
-.085
-.170
.176
.228
-.311
-.116
.799
.039
.001
.097
.003
.235
-.037
2.250*
-2.936
-1.315
.021
-.001
.049
-.040
-.443
.337
1.088
1.111
1.762
.617
.013
.051
.075
.042
.083
R2 = .007; N = 1,659
.100
1.042
.532
.207
Regression
Coefficient
Variable
Cohabited? (y = 1, n = 2)
Length of courtship (months)
Polynomial
Length of marriage (years)
Polynomial
Social-desirability response tendency
Race
Caucasian (omitted category)
Hispanics
Blacks
Others
Age at first marriage
Polynomial
Heterogeneity
Age (in years)
Previous marital status
Social class parents
Wife superior education
.199*
3.052**
.934
.710***
639
FOOTNOTES
1. In multipleregressiona normaldistributionof the
dependentvariableis assumedfor the unbiasedcomputation of significantlevels (Cohen and Cohen,
1975:48). Becausethe scoreson the maritalinstability index show considerableheteroscedasticityand
the directionof the resultingbias is unknown,variables that do not quite reachstatisticalsignificance
were not excluded from the presentedequations.
of the instabilityscores
Logarithmictransformations
wouldhaveproduceda normaldistributionof the independent variable, but such a transformation
obscuresthe substantivemeaningsof the results.In
addition, such a transformation automatically
makesthe relationshipbetweenage at first marriage
and instabilitylinear,sinceall joint normaldistributions have linearrelationships(Blalock, 1979:388).
640
REFERENCES
Adams, Bert N. 1979. "Mate selection in the United
States: A theoretical summation." Pp. 259-267 in
Wesley R. Burr, Reuben Hill, F. Ivan Nye, and Ira L.
Reiss (eds.), Contemporary Theories about the Family (Vol. 1). New York: Free Press.
Bauman, Karl E. 1967. "The relationship between age
at first marriage, school dropout, and marital instability: An analysis of the Glick effect." Journal of
Marriage and the Family 29: 672-680.
Berger, Peter L., and Hansfried Kellner. 1980. "Marriage and the construction of reality." Pp. 392-402 in
Howard Robboy and Candace Clark (eds.), Social Interaction. New York: St. Martin's Press.
Bernard, Jessie Sherley. 1982. The Future of Marriage.
New Haven, CT: Yale University Press.
Blalock, Herbert M. 1979. Social Statistics. New York:
McGraw-Hill.
Booth, Alan, and John N. Edwards. 1984. Age at Marriage and Marital Instability. Unpublished manuscript.
Booth, Alan, David R. Johnson, and John N. Edwards.
1983. "Measuring marital instability." Journal of
Marriage and the Family 45: 387-393.
Booth, Alan, David R. Johnson, and Lynn K. White.
1981. Female Labor Force Participation and Marital
Instability: Methodology Report. Unpublished manuscript.
Booth, Alan, and Lynn K. White. 1980. "Thinking
about divorce." Journal of Marriage and the Family
29: 672-680.
Bumpass, Larry L., and James A. Sweet. 1972. "Differentials in marital instability: 1972." American Sociological Review 37: 754-767.