You are on page 1of 15

Available online at www.sciencedirect.

com

Journal of Vocational Behavior 73 (2008) 3751


www.elsevier.com/locate/jvb

Workfamily conict and psychological well-being: Stability


and cross-lagged relations within one- and six-year follow-ups
Johanna Rantanen a,*, Ulla Kinnunen b, Taru Feldt a, Lea Pulkkinen a
a

Department of Psychology, University of Jyvaskyla, P.O. Box 35, FIN-40014, Finland


b
Department of Psychology, University of Tampere, Finland
Received 20 June 2007
Available online 26 January 2008

Abstract
The rank-order stability and cross-lagged relations between work-to-family conict (WFC), family-to-work conict
(FWC), and psychological well-being were examined in two longitudinal studies with full two-wave panel designs. In Study
1 (n = 365), the time lag was one year, and in Study 2 (n = 153), six years. The Structural Equation Modeling showed that
the stability for WFC was .69 over one and .73 over six years. The respective stabilities for FWC were .57 and .48. Crosslagged relations were not detected between WFC/FWC and low psychological well-being (job exhaustion, marital adjustment, parental stress, and psychological distress), expected to exist on the basis of the integrative model of workfamily
interface [Frone, M. R., Yardley, J. K., & Markel, K. S. (1997). Developing and testing an integrative model of the
workfamily interface. Journal of Vocational Behavior, 50, 145167.]. Instead, high job exhaustion preceded high psychological distress, both within one- and six-year time lags, and within a one-year time lag: low marital adjustment preceded
high psychological distress, and high psychological distress preceded high parental stress. These ndings supported the
spillover perspective on workfamily interface [Edwards, J. R., & Rothbard, N. P. (2000). Mechanisms linking work
and family: Clarifying the relationship between work and family constructs. Academy of Management Review, 25,
178199.].
2008 Elsevier Inc. All rights reserved.
Keywords: Workfamily conict; Psychological well-being; Longitudinal; Rank-order stability; Cross-lagged; Spillover; Gender

1. Introduction
Workfamily conict refers to an experience in which the demands from work and family domains collide
with each other and exceed the resources available for an individual (Greenhaus & Beutell, 1985). Most work
family conict studies (94%) have been cross-sectional (Casper, Eby, Bordeaux, Lockwood, & Lambert, 2007)
and, therefore, little is known about the stability and change of workfamily conict over time (Ford, Heinen, &
Langkamer, 2007; Geurts & Demerouti, 2003), and causal relations between workfamily conict and
*

Corresponding author. Fax: +358 14 2602841.


E-mail address: johanna.rantanen@jyu. (J. Rantanen).

0001-8791/$ - see front matter 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.jvb.2008.01.001

38

J. Rantanen et al. / Journal of Vocational Behavior 73 (2008) 3751

psychological well-being (Allen, Herst, Bruck, & Sutton, 2000). Furthermore, the question of gender as a moderator of workfamily conict experiences deserves more attention (Ford et al., 2007; Voydano, 2002). In the
present study, workfamily conict originating either from the work (work-to-family conict, WFC) or from
the family (family-to-work conict, FWC) (Greenhaus & Beutell, 1985; Mesmer-Magnus & Viswesvaran,
2005), and psychological well-being in general as well as in the areas of work and family (cf. Allen et al.,
2000) were studied longitudinally among men and women.
1.1. Rank-order stability vs. instability of workfamily conict experience
Westman, Etzion, and Gortler (2004) found that the testretest correlation was .86 for WFC within the
time lag of a few weeks, and Kinnunen, Geurts, and Mauno (2004) found that it was .71 within the time
lag of a year. For FWC, the testretest correlations were .51 across three months (Leiter & Durup, 1996)
and .75 across six months (Kelloway, Gottlieb, & Barham, 1999). The testretest correlations support the view
that workfamily conict is a relatively stable experience, although one could be expect that critical incidents
within work, family or individual circumstances (e.g., job change, judicial separation, onset of severe illness)
might generate instability in workfamily conict experiences. A time lag of some weeks or months may be too
short to demonstrate the stable vs. unstable nature of WFC and FWC experiences, as major changes in work,
family, and other areas of life are likely to happen only to a minority of working adults in that interval. Therefore, further investigation with longer time lags was needed to provide missing information about the longterm stability of WFC and FWC, and to facilitate the denition of an optimal time lag in future longitudinal
studies (cf. de Lange, Taris, Kompier, Houtman, & Bongers, 2003).
In the present study, the stability of WFC and FWC was studied across one- and six-year time lags (Study 1
and Study 2, respectively). With these time lags it was possible to get more information about the stability of
workfamily conict experiences and to assess the role of interim eects (i.e., eects of unobserved events
during the time, such as job change) (de Lange et al., 2003, p. 285) on the stability of WFC and FWC. Within
six years, major changes in work and family characteristics that aect workfamily conict experiences (e.g.,
work schedule, age and number of children; Byron, 2005) are more plausible than within one year. Therefore,
it was hypothesized that despite the relatively stable nature of workfamily conict, the stability of WFC and
FWC would be lower in the six-year time lag than in the one-year time lag (Hypothesis 1).
1.2. Cross-lagged relations between workfamily conict and psychological well-being
The relationship between workfamily conict and psychological well-being has been conceptualized in
three ways. First, Conict ? Strain approach: workfamily conict is seen as an antecedent of high psychological strain, and thought to cause a state where an individuals mental resources are threatened and depleted,
leading to a possible decrease in well-being (e.g., Eby, Casper, Lockwood, Bordeaux, & Brinley, 2005;
Voydano, 2002). Second, Strain ? Conict approach: high psychological strain is seen as an antecedent
of workfamily conict, as a consequence of which individuals suering from psychological strain might have
less mental resources to manage simultaneous work- and family-related duties and the incidence of workfamily conict increases (cf. Westman et al., 2004). Third, Reciprocity approach: work distress is both an antecedent of WFC and an outcome of FWC, and similarly, family distress is both an antecedent of FWC and an
outcome of WFC (Frone, Yardley, & Markel, 1997). Through these work- and family-related strain experiences, the two directions of workfamily conict are reciprocally related to each other.
The Conict ? Strain approach has received wide support but mostly from cross-sectional studies which
do not warrant true antecedentoutcome interpretations (Allen et al., 2000; Casper et al., 2007). Schnabel
(1996) emphasizes that a prerequisite for antecedentoutcome interpretations is a longitudinal study with a
full cross-lagged panel design where every concept to be studied is measured at each time point, and data
are analyzed in a way that considers the stability and cross-lagged relations between all variables simultaneously. In the area of workfamily conict and psychological well-being, the few studies fullling these
requirements support the Reciprocity approach: WFC was both an antecedent of and an outcome of job
exhaustion over three months (Leiter & Durup, 1996) and over six- and twelve-week periods (Demerouti,
Bakker, & Bulters, 2004). Leiter and Durup (1996) also found that conicts with family members predicted

J. Rantanen et al. / Journal of Vocational Behavior 73 (2008) 3751

39

WFC, which in turn predicted marital dissatisfaction and dysphoric mood. Over six months FWC was an
antecedent of, and WFC an outcome of, psychological stress symptoms (Kelloway et al., 1999).
The cross-lagged relations between WFC/FWC and psychological well-being were examined in the present
study with a full two-wave panel design. Both in Study 1 (one-year time lag) and in Study 2 (six-year time lag)
time-based workfamily conict was studied (excluding strain- and behavior-based conict; e.g., Carlson,
Kacmar, & Williams, 2000), because the psychological well-being variables were strain-based: job exhaustion,
marital adjustment, parental stress, and psychological distress. In this way, an overlap in item contents was
avoided between workfamily conict and well-being variables, which could have caused overestimated associations (Hurrell, Nelson, & Simmons, 1998). Study 1 was an elaboration of the earlier study by Kinnunen
et al. (2004). This study has now been completed, rst, with a full cross-lagged panel design and using Structural Equation Modeling (SEM) instead of regression analysis, and second, with both time-based WFC and
FWC instead of only WFC including time- and strain-based items.
Studies of workfamily conict having longer than a one-year time lag are truly rare. In the present study,
the one- and six-year time lags were expected to produce knowledge about the long-term relationship between
WFC/FWC and psychological well-being. The hypotheses concerning WFC/FWC and psychological wellbeing were built on the integrative model of workfamily interface (Frone, Yardley et al., 1997). According
to this model, role-related distresses, such as work and family distress, are associated with concurrent
workfamily conict, and reciprocally, workfamily conict is associated with these experiences. Frone and
colleagues (1997, p. 149) argue that work distress weakens both an individuals capability and enthusiasm
to meet the demands of family domain, because his or her resources are drained and used in the work domain,
which leads to the experience of WFC. Family distress, in turn, hinders the management of work duties and
obligations, because an individuals resources are bounded to the family domain, which leads to the experience
of FWC. Moreover, the experience of WFC simultaneously generates family distress, and the experience of
FWC generates work distress. The argument for this is that, because WFC recurrently hinders the full participation in family life, and FWC hinders the full participation in working life, the quality of life in these
domains is challenged, leading to feelings of family and work distress, respectively (Frone, Yardley et al.,
1997, p. 152). Frone and colleagues (1997) tested the integrative model of workfamily interface with a
cross-sectional study design, which as a whole supported their model. They brought up, however, two major
limitations and, consequently, two directions for future research: rst, conducting longitudinal studies testing
dierent parts of the model, and second, integrating general health-related outcomes into their model (Frone,
Yardley et al., 1997, pp. 164165).
Based on the integrative model of workfamily interface, it was hypothesized that high job exhaustion at
Time 1 would precede high WFC at Time 2, and that low marital adjustment and high parental stress at Time
1 would precede high FWC at Time 2 (Hypothesis 2). High WFC at Time1, in turn, was hypothesized to precede low marital adjustment and high parental stress at Time 2, and high FWC at Time 1 was hypothesized to
precede high job exhaustion at Time 2 (Hypotheses 3). We additionally assumed that, on the one hand, context-free psychological distress is likely to hinder performance both in work and family roles, and on the other
hand, the diculties of meeting the demands of work and family domains are likely to increase psychological
distress (cf. Frone, Yardley et al., 1997). Therefore, it was hypothesized that high psychological distress at
Time 1 would precede high WFC and FWC at Time 2, and high WFC and FWC at Time 1 would precede
high psychological distress at Time 2 (Hypothesis 4). These associations were expected to be more numerous
over one year time lag than over six years. Frone, Russell, and Cooper (1997) found that across four years,
only FWC had an eect on psychological well-being (predicting depression), and suggested that the detrimental eect of WFC on health might occur on a shorter time lag.
1.3. Cross-lagged relations between work-related, family-related and general psychological well-being
The idea of workfamily conict as a mediator between antecedents and outcomes in work and family
domains (Frone, Russell, & Cooper, 1992; Frone, Yardley et al., 1997) has governed the workfamily interface
research since its early publications (Ford et al., 2007). The spillover perspective on workfamily interface represents a dierent view according to which work and family domains can aect each other also directly (Edwards
& Rothbard, 2000; Lambert, 1990). Both Edwards and Rothbard (2000) and Lambert (1990), however,

40

J. Rantanen et al. / Journal of Vocational Behavior 73 (2008) 3751

emphasize that the dierent mechanisms and processes linking work and family life are likely to co-exist. Consequently, in the present study, the spillover perspective on workfamily interface was considered as complementing rather than competing with the integrative model of workfamily interface (Frone, Yardley et al., 1997).
Edwards and Rothbard (2000) present that spillover as a linking mechanism between work and family
domains exists when there is a positive link between a work construct and a distinct, but related construct
in the family domain. They further specify that spillover is mainly unintentional process between work and
family domains, where the eective performance within the receiving role is either enhanced or weakened
by the positive or negative mood and experiences within the originating role. The quality of performance
(good vs. poor), in turn, aects the intrinsic and extrinsic rewards obtained within the receiving role and produces similar emotional, aective, and cognitive states experienced within the originating role.
Longitudinal studies of workfamily spillover that consider both spillover eects from work-to-family and
from family-to-work are scarce (reviews by Casper et al., 2007; Eby et al. 2005). Heller and Watson (2005)
showed using a diary study design that daytime job satisfaction was positively related to marital satisfaction
in the evening and vice versa. Leiter and Durup (1996) found that conicts with family members predicted
emotional overload at work three months later. In addition, decrease in job and marital role quality was associated with increase in psychological distress over one-year time period (Barnett, Raudenbush, Brennan,
Pleck, & Marshall, 1995). In the present study, the concept of spillover was applied to psychological well-being
in work, family and generally in life, which all have been shown to be moderately stable (r > .50) in studies
with at least a one-year time lag (e.g., Frone, Russell et al., 1997; Frone, Yardley et al., 1997; Kinnunen
et al., 2004). Due to their stable nature, it was assumed that job exhaustion, marital adjustment, parental
stress, and psychological distress would produce long-term and cumulative eects on role performance across
life domains aecting the well-being and adjustment in each life domain.
The specic hypotheses concerning the cross-lagged spillover relations were as follows: High job exhaustion
at Time 1 precedes low marital adjustment and high parental stress at Time 2, and low marital adjustment and
high parental stress at Time 1 precede high job exhaustion at Time 2 (Hypothesis 5). High psychological distress at Time 1 precedes high job exhaustion, low marital adjustment and high parental stress at Time 2, and
high job exhaustion, low marital adjustment and high parental stress at Time 1 precede high psychological
distress at Time 2 (Hypothesis 6).
1.4. The role of gender in the stability and cross-lagged relations of workfamily interface
Few longitudinal studies have addressed the role of gender in the rank-order stability of WFC and FWC or
in the cross-lagged relations between workfamily conict and psychological well-being (Eby et al., 2005). In
the studies available, methodological dierences may explain conicting results. Demerouti and colleagues
(2004), who applied SEM with a full cross-lagged panel design, did not nd gender dierences in the relations
between work pressure, WFC, and job exhaustion. On the contrary, Kinnunen and colleagues (2004), made
separate regression analyses for men and women and found gender dierences: for women, WFC was an antecedent of job dissatisfaction, parental distress, and psychological stress symptoms, whereas for men, WFC was
an outcome of marital dissatisfaction, parental distress, and psychological as well as physical stress symptoms.
In the present study, the stability and cross-lagged relations of workfamily interface were examined and
compared between men and women in order to nd out whether or not these relations would be moderated
by gender. The social-role hypothesis emphasizing gender similarities in the workfamily interface experiences
(Voydano, 2002) was considered plausible in the cultural context of the present study. According to this perspective, the key issue is a comparable role involvement, not gender: if both men and women invest the same
amount of time, attention, and energy on work and family roles, these involvements should have similar eects
on the experiences of men and women (Voydano, 2002). As the participants were Finnish men and women
being both employed and having family, it is likely that they had comparable role involvements. Finland is one
of the Nordic welfare states where the roles of women and men are relatively equal in working life. For one
example, women constitute almost as a large proportion of the workforce as men, and in most cases being in
full-time employment (Statistical Oce of the European Communities, 2006). Therefore, gender dierences in
the stability and cross-lagged relations between workfamily conict and psychological well-being were not
expected (Hypothesis 7).

J. Rantanen et al. / Journal of Vocational Behavior 73 (2008) 3751

41

2. Methods
2.1. Participants
2.1.1. Study 1
The data were derived from the research project titled Economic Crisis, Job Insecurity and the Household carried out during the years 1999 and 2000. The original sample (n = 1878) was drawn randomly from
the database of the Population Register Centre of Finland and it was limited to 25- to 59-year-old citizens. The
mailed questionnaire was returned by 851 and 655 persons in 1999 and 2000, respectively. It has been found
that the respondents represent well the Finnish adult population regarding gender, age, marital status, and
geographical location (Kinnunen et al., 2000). In the present study, respondents who were employed and a
had family (spouse/partner and/or child(ren) living at home) both in 1999 (Time 1) and 2000 (Time 2) were
included. Altogether 365 participants (189 men, 176 women) met these criteria.
The demographic characteristics of the participants did not change signicantly between Time 1 and 2; 73%
had the same work situation (same employer, occupational status, and full-time vs. part-time weekly working
hours) and 84% had the same family situation (same spouse and number of children). Thus, only Time 1
demographics are described. At Time 1, the mean age of the participants was 43 years for both men and
women (range 2559 years, SD 9 years). More men than women were blue-collar (39% vs. 25%) and upper
white-collar (38% vs. 27%) workers, whereas more women than men were lower white-collar workers (48%
vs. 23%) (p < .001). The majority of men (95%) and women (88%) worked full-time (>34 h per week), and
the average number of weekly working hours was 44 for men and 39 for women (p < .001). Of the men,
98%, and of the women, 90% were married or cohabiting, and of these participants spouses 80% were working. Over half of the men (61%) and women (62%) had children living at home. Among those with children,
the number of children was typically one or two and the median age of the youngest child was nine years.
2.1.2. Study 2
The data were gathered in 1995 and 2001, as a part of the Jyvaskyla Longitudinal Study of Personality
and Social Development, that has been conducted in Finland since 1968 (Pulkkinen, 2006). The original
sample consisted of all 369 pupils (196 boys and 173 girls born mostly in 1959) in 12 urban and suburban,
randomly selected second-grade school classes. The sample in adulthood (at ages 36 and 42) has proven to
be representative of the age cohort born in 1959 according to gender, marital status, number of children,
and employment status (Pulkkinen, 2006). Those participants who had participated both at age 36 in
1995 and at age 42 in 2001, and who were employed and had a family (spouse/partner and/or child(ren)
living at home) were included in the present study. These criteria yielded 153 participants (78 men, 75
women).
Of these 153 participants, 38% had the same work situation (same employer, occupational status, and fulltime vs. part-time weekly working hours) and 56% had the same family situation (same spouse and number of
children) at the age of 42 than at the age of 36. At age 36 (Time 1), the participants divided into occupational
groups as follows: 43% (men)/3% (women) were blue-collar workers, 41%/72% lower white-collar workers,
and 16%/25% upper white-collar workers. At age 42 (Time 2), the respective percentages were: 47% (men)/
8% (women) blue-collar workers, 17%/60% lower white-collar workers, and 36%/32% upper white-collar
workers. Men in this sample were more often than women blue-collar workers, whereas women were more
often than men lower white-collar workers at both Times 1 (p < .001) and 2 (p < .001). Similarly, the average
weekly working hours was higher for men than for women, the respective means being 46 and 40 h both at
Time 1 (p < .001) and at Time 2 (p < .01). Of the men, 96% (Time 1) and 95% (Time 2) worked full-time
(>34 h per week); the respective percentages for women were 81% and 85%.
The majority of the participants (100% of the men and 81% of the women) were either married or cohabiting at Time 1, and of these participants spouses 81% were working. At Time 2, 92% of the men and 82% of
the women had a spouse, and of these, 89% were working. The majority of the participants had children living
at home (90% of the men and 88% of women at Time 1; 92% of men and 91% of women at Time 2), the
common number of children being two at both times. The median age of the youngest child was four years
at Time 1 and ten years at Time 2.

42

J. Rantanen et al. / Journal of Vocational Behavior 73 (2008) 3751

2.2. Measures
The measures for job exhaustion, parental stress, and psychological distress were the same in both studies.
The measures for workfamily conict and marital adjustment diered between the studies; but the content of
these measures corresponded highly between the studies. The measures were carefully translated from original
sources into Finnish, the native language of the participants.
2.2.1. Workfamily conict
In Study 1, WFC was measured with two items from the scale by Stephens and Sommer (1996) (e.g., My
work keeps me from my family more than I would like). FWC was measured with two items from the scale
developed by Netemeyer, Boles, and McMurrian (1996) (e.g., I have to put o doing things at work because
of the demands on my time at home). The response scale ranged from 1 (strongly disagree) to 5 (strongly
agree). The Cronbachs alphas for WFC at Time 1 were .87 (men) and .76 (women), and for FWC .77/.67
(respectively). At Time 2, they were .83/.84 for WFC, and .68/.64 for FWC.
In Study 2, workfamily conict was measured with a four-item scale developed by Frone et al. (1992). The
scale consists of two items measuring WFC and two items measuring FWC. The response scale ranged from 1
(never) to 5 (always). The Cronbachs alphas at Time 1 for WFC were .79 (men) and .87 (women), and for
FWC .77/.80 (respectively). At Time 2 they were .78/.58 for WFC, and .68/.74 for FWC.
2.2.2. Job exhaustion
Psychological strain at work was measured with a four-item job exhaustion scale developed by Maslach and
Jackson (1986). In Study 1, the response scale ranged from 1 (never) to 5 (always) and the Cronbachs alphas
for this measure (men and women, respectively) were .86/.91 (Time 1) and .87/.89 (Time 2). In Study 2, the
response scale ranged from 1 (never) to 6 (always) and the Cronbachs alphas were .88/.89 (Time 1) and
.90/.89 (Time 2).
2.2.3. Marital adjustment
In Study 1, the functionality and the quality of partner relationship were assessed with nine items from the
dyadic adjustment scale (Busby, Crane, Larson, & Christensen, 1995; Spanier, 1976) (e.g., How often have
you regretted your marriage or cohabitation? (reversed)). The response scale ranged from 1 (never) to 5
(always) and the Cronbachs alphas for this measure were .85/.85 (Time 1) and .85/.84 (Time 2).
In Study 2, marital adjustment at Time 1 was assessed with 15 questions (e.g., How much conict, tension,
and discord is there in your relationship at the moment?). On the basis of these questions, ve value categories describing marital quality were formed with (1) denoting a poor and (5) a very good relationship (see
Kinnunen & Pulkkinen, 2003). At Time 2, marital adjustment was assessed with the same nine items (from
Busby et al., 1995; Spanier, 1976) as in Study 1. For most of the items the response scale ranged from 1 (never)
to 6 (always). Because there were two items with a response scale from 1 (never) to 5 (every day), all of the
items were rst standardized before forming the composite variable. The Cronbachs alphas for this measure
at Time 2 were .81/.83.
2.2.4. Parental stress
The diculties and sense of inadequacy experienced as a parent were assessed with three items from the
measure by Abidin (1990) (e.g., When I think about the kind of parent I am, I often feel guilty or bad about
myself). In Study 1, the response scale ranged from 1 (strongly disagree) to 5 (strongly agree) and the Cronbachs alphas for this measure were .79/.78 (Time 1) and .80/.84 (Time 2). In Study 2, the response scale ranged from 1 (I do not agree at all) to 4 (I agree totally) and the Cronbachs alphas were .77/.75 (Time 1), and
.74/.72 (Time 2).
2.2.5. Psychological distress
Psychological strain experienced in everyday life was measured with the 12-item General Health Questionnaire (GHQ: Goldberg, 1972). The response scale ranged from 1 to 4. The Cronbachs alphas in Study 1 were
.87/.90 (Time 1) and .89/.91 (Time 2), and in Study 2 they were .83/.89 (Time 1), and .88/91 (Time 2).

J. Rantanen et al. / Journal of Vocational Behavior 73 (2008) 3751

43

2.3. Data analysis


The method of analyses was SEM using multigroup procedure and the analyses proceeded in three stages.
In the rst stage, the stability of WFC and FWC was compared between Study 1 and Study 2 (Hypothesis 1).
Therefore, the data from Studies 1 and 2 concerning these two constructs were combined into a common
covariance matrix. First, the equality of the stability coecients of WFC and FWC between genders within
Studies 1 and 2 was tested. After that, the equality of the stability coecients of WFC and FWC between
Studies 1 and 2 was examined.
In the second stage, the hypothesized cross-lagged relations of workfamily interface were investigated
within Study 1 (Hypotheses 26). First, the stability coecients and cross-lagged relations between study variables were freely estimated for men and women. Then non-signicant cross-lagged relations (t-value <1.96)
were removed and the gender-equality of the remaining signicant cross-lagged relations were tested (Hypothesis 7). In Study 1, the observed items were used as latent factor indicators for WFC, FWC, job exhaustion,
and parental stress. However, the use of nine and twelve observed items as latent factor indicators for marital
adjustment and psychological distress, respectively, would have resulted in an unacceptably high ratio of estimated parameters, which was not feasible. Therefore, a parceling procedure for the latent factor indicators of
marital adjustment and psychological distress was chosen (Little, Cunningham, Sharar, & Widaman, 2002).
The latent factors of marital adjustment and psychological distress were based on three parceled scales.
In the third stage, the hypothesized cross-lagged relations of workfamily interface were investigated in
Study 2 (Hypotheses 26). Path analysis with observed variables was used instead of latent factors, as the sample size did not allow full SEM analyses with a measurement model component. The stability coecients and
cross-lagged relations were estimated rst for the whole sample (men and women combined), after which the
gender-equality of the nal model with only signicant cross-lagged relations (t-value <1.96) was tested
(Hypothesis 7).
The SEM and path analyses were conducted by using Mplus statistical package (Version 4.0: Muthen &
Muthen, 19982006) in which the robust full information maximum likelihood (MLR) estimation method
was used. The goodness-of-t of the SEM models was evaluated using a chi-square value, with a non-significant p-value indicating a good t (Kelloway, 1998), a root mean square error of approximation (RMSEA)
with values 0.05 or less indicating a good t, values 0.060.08 a reasonable t (Steiger, 1990), and a
comparative t index (CFI) and TuckerLewis Index (TLI) with values 0.90 or above indicating a good t
(Hu & Bentler, 1999). When comparing models with alternative equality constraints, the signicance of change
in the chi-square index of t was evaluated by using a scaled chi-square dierence test (Satorra & Bentler,
1999). The constraints are supported if the chi-square dierence test produces a non-signicant loss of t
for the constrained model, when compared to the unconstrained model.
3. Results
3.1. Descriptive analyses
The correlations were mostly similar in direction and magnitude between genders both in Study 1 and 2
(Table 1). The testretest correlations for WFC and FWC were signicant (p < .001). Signicant (p < .001)
cross-lagged correlations between workfamily conict and psychological well-being indicators were found
only in Study 1, whereas signicant (p < .001) cross-lagged spillover relations from work to general psychological well-being were found both in Study 1 and 2.
3.2. Rank-order stability of WFC and FWC: Study 1 and Study 2 compared
A base model (Model A), consisting of a correlated two-factor measurement model at Time 1 and Time 2
and stability coecients between the latent factors for WFC and FWC, was rst freely estimated both across
gender and study samples: men in Study 1, women in Study 1, men in Study 2, and women in Study 2. The t
of Model A was reasonable: v2 (88) = 130.95, p < 0.01, RMSEA = 0.06, CFI = 0.96, TLI = 0.95. To see
whether or not the rank-order stability of WFC and FWC diered rst, between genders, and second, between

44

Table 1
Means (M), standard deviations (SD), and pearson intercorrelations between study variables
Variables

10

11

12

.17*

.25**

.14

.07

.35**m
.08f
.10
.34***
.12

.54***

.10

.18*

.05

.01

.04

.09
.18*
.08

.39***
.12
.07

.01
.67***
.09

.01
.13
.34***

.11
.28***
.21*

.02
.39***
.08

.03
.23**

.20*
.16

.23***
.30***

.24**
.06

.57***
.03

.20*
.23**

.32***

.18m
.47***f
.16*

.13

.01

.14

.25**m
.08f
.17*

.17*

.19*

.32***

.64***m
.40***f
.19**

2. FWC, family-to-work conict


3. Job exhaustiona
4. Marital adjustmenta

1.43 (0.64) 1.81 (0.74)


2.19 (0.85) 2.51 (0.92)
3.96 (0.52) 3.94 (0.95)

.32***
.34***
.18**
.09
.18**

5. Parental stressa
6. Psychological distress

2.30 (0.85) 1.57 (0.56)


1.95 (0.43) 1.89 (0.41)

.11
.19***

.21***
.22***

Time 2
7. WFC, work-to- family conict 2.57 (1.22) 2.57 (0.91)

.59***

.26***

.28***

.10

.13*

.13*

8. FWC, family-to-work conict

1.41 (0.63) 1.78 (0.69)

.28***

.48***

.14**

.06

.25***

.15**

.38***

9. Job exhaustiona

2.15 (0.84) 2.44 (0.91)

.19***

.13*

.65***

.22***

.26***

.36***

.31***

.22***

10. Marital adjustmenta

3.93 (0.51) 0.04b (0.61)

.09

.12*

.80***

.17**

.22***

.09

.05

.16**

11. Parental stressa


12. Psychological distress

2.29 (0.91) 1.48 (0.53)


1.93 (0.42) 1.94 (0.47)

.12
.14**

.23***
.19***

.01m
.17f
.26**
.41***

.22***
.30***

.65***
.20***

.26***
.55***

.21***
.20***

.31***
.23***

.30***
.54***

.07
.11
.11

.10
.30**
.07m

.25**
***
.30 f
.27***
.26***
***
.52
.14m
.18**
.48***f

.11

.20**
.25***

.30***

.28**
**
.21 m
.46***f

Note. Correlations for Study 1 (n = 293365) are below and for Study 2 (n = 129153) above the diagonal. If two correlations are given there was signicant (p < .05) gender dierence
in the correlations between m men and f women.
*
p < .05.
**
p < .01.
***
p < .001.
a
The response scale is not comparable between Study 1 and 2.
b
Calculated from standardized scores due to dierences in the response scales of the items.

J. Rantanen et al. / Journal of Vocational Behavior 73 (2008) 3751

Study 1
Study 2
M (SD)
M (SD)
1. WFC, work-to- family conict 2.57 (1.23) 2.55 (1.04)

Time 1

J. Rantanen et al. / Journal of Vocational Behavior 73 (2008) 3751

45

Fig. 1. Stability of work-to-family and family-to-work conict within one- and six-year time lags. In the factor loadings, the rst count
refers to men and the second to women. *Item autocovariance between Time 1 and Time 2 has been released.

one- and six-year time lags (Hypothesis 1), Model A was compared with two alternative models. In Model B,
the stability coecients for WFC and FWC were constraint equal between genders within each study sample.
In Model C, these equality constraints were extended over both study samples as well.1 Both Models B and C
had reasonable t with the data: v2 (102) = 148.36, p < 0.01, RMSEA = 0.06, CFI = 0.96, TLI = 0.96, and v2
(109) = 187.28, p < 0.001, RMSEA = 0.06, CFI = 0.93, TLI = 0.93, respectively.
Model B was supported over Model A: the chi-square dierence test was non-signicant (v2dierence
(14) = 17.97, p = n.s.). Thus, in comparison to freely estimated Model A, the equality constraints in stability
coecients included in Model B did not produce a signicant loss of model t, which means that the rankorder stabilities of WFC and FWC did not dier statistically signicantly between genders within Study 1
and Study 2. Instead, there was signicant dierence in the rank-order stabilities of WFC and FWC between
one- and six-year time interval (comparison between Models C and A: v2dierence (21) = 52.56, p < 0.001). As
only Model B was supported over Model A, it provided better t with the data than Model C, and the results
of Model B are illustrated in Fig. 1. Overall, the stability for WFC was higher than the stability for FWC in
both studies. When the two studies were compared, the stability of WFC was statistically signicantly higher
and the stability of FWC was lower in Study 2 than in Study 1.
3.3. Cross-lagged relations of workfamily interface across one year
In Study 1, a correlated six-factor measurement model at Time 1 and Time 2 with stability coecients
between the latent factors for WFC, FWC, job exhaustion, marital adjustment, parental stress, and psycho1
A prerequisite for valid group comparisons in SEM and regression coecients is the measurement invariance of the scales between the
groups (Little, 1997; Ryan, Chan, Ployhart, & Slade, 1999). In the present study, the measurement invariance tests were executed before
the examination of gender- and sample-equality in the stability coecients of WFC and FWC. The results showed that the measurement
structure of these scales was statistically highly similar both across time and between genders and samples. Thus, time, gender and sample
invariance constraints in factor loadings and factor correlations were included in the Models A, B, and C. The detailed evidence was
provided during the review process, but was omitted from the nal manuscript, as the structural invariance of the measurement scales was
not the main focus of the present study. If interested, more information can be obtained from the corresponding author.

46

J. Rantanen et al. / Journal of Vocational Behavior 73 (2008) 3751

Fig. 2. Stability and cross-lagged relations between workfamily conict and four indicators of psychological well-being in men and
women in Study 1. The model includes the synchronous correlations between latent factors at Time 1 and residual covariances between
latent factors at Time 2, although these are not shown here for the sake of clarity of presentation. In the factor loadings, the rst count
refers to men and the second to women. *Item/parcel autocovariance between Time 1 and Time 2 has been released.

logical distress and cross-lagged relations (Hypotheses 26) were freely estimated for both genders (Model D).
The t of Model D with the data was satisfactory (v2 (989) = 1299.40, p < 0.001, RMSEA = 0.04, CFI = 0.96,
TLI = 0.95), but it contained only three signicant cross-lagged relations: high job exhaustion at Time 1 preceded high psychological distress at Time 2, low marital adjustment at Time 1 preceded high psychological
distress at Time 2, and high psychological distress at Time 1 preceded high parental stress at Time 2. The
non-signicant cross-lagged relations were removed and the resulting model was rst freely estimated for both
genders (Model E). Then, the signicant cross-lagged relations and stability coecients were constrained equal
between men and women (Model F, Hypothesis 7) and compared to freely estimated Model E in order to
examine the invariance across genders.2
Although both Models E and F provided satisfactory t with the data (v2 (1023) = 1349.11, p < 0.001,
RMSEA = 0.04, CFI = 0.96, TLI = 0.95, and v2 (1058) = 1397.76, p < 0.001, RMSEA = 0.04, CFI = 0.96,
TLI = 0.95, respectively), the v2 dierence test supported the gender-equal Model F over the freely estimated
Model E (v2dierence (35) = 48.02, p = n.s.). Thus, the stabilities and cross-lagged relations were the same
between men and women. These nal results of Model F are illustrated in Fig. 2. Despite highly signicant
cross-lagged correlations (Table 1), workfamily conict and psychological well-being constructs were not
longitudinally related to each other when the stability and concurrent correlations between each construct

The measurement model component in Models D, E, and F was found to be time- and gender-invariant before gender-equality
comparisons in cross-lagged relations and stability coecients were performed. Hence, time and gender invariance constraints in factor
loadings and factor correlations were included in Models D, E, and F. If interested in these measurement invariance analyses, more
information can be obtained from the corresponding author.

J. Rantanen et al. / Journal of Vocational Behavior 73 (2008) 3751

47

were taken into account. Instead, job exhaustion and marital adjustment had longitudinal eects on psychological distress, and psychological distress had a longitudinal eect on parental stress.
3.4. Cross-lagged relations of workfamily interface across six years
In Study 2, the stabilities and cross-lagged relations between WFC, FWC, job exhaustion, marital adjustment, parental stress and psychological distress (Hypotheses 26) were investigated through path analysis. The
constructed Model G, estimated rst for the whole sample, consisted of stability coecients and cross-lagged
paths with synchronous correlations (at Time 1) and residual covariances (at Time 2). Model G provided good
t with the data: v2 (10) = 13.03, p = n.s., RMSEA = 0.04, CFI = 0.99, TLI = 0.95. However, it contained
only one signicant cross-lagged relation: high job exhaustion at Time 1 preceded high psychological distress
at Time 2. Therefore, all the non-signicant cross-lagged relations were removed (Model H). Model H provided outstanding t with the data: v2 (29) = 28.64, p = n.s., RMSEA = 0.00, CFI = 1.00, TLI = 1.00.
To examine the gender-equality of Model H, it was freely estimated for both genders (Model I) and compared to Model J, in which the stability coecients and the detected cross-lagged relation were set equal
between genders (Hypothesis 7). Both Model I and J provided very good t with the data (v2 (58) = 47.50,
p = n.s., RMSEA = 0.00, CFI = 1.00, TLI = 1.06, and v2 (95) = 97.91, p = n.s., RMSEA = 0.00,
CFI = 0.99, TLI = 0.99, respectively) but the v2 dierence test supported gender-equal Model J over Model
I (v2dierence (37) = 49.69, p = n.s.). In conclusion, in Model J the stability coecients were .53 for WFC,
.41 for FWC, .65 for job exhaustion, .32 for marital adjustment, and .56 for parental stress for both genders.
The stability coecient for psychological distress (.09) turned out to be non-signicant after the cross-lagged
relation from job exhaustion at Time 1 to psychological distress at Time 2 was included in the model. The path
coecient for the latter relation was .32.
4. Discussion
The results of the present study showed, rst, that the experiences of work-to-family conict (WFC) and
family-to-work conict (FWC) where rather stable even across six years. Second, workfamily conict and
psychological strain in job, family, or everyday life in general did not seem to have antecedentoutcome relations that would extend over years. Instead, the adverse eect of job exhaustion on general psychological wellbeing appeared to be far-reaching.
Both WFC and FWC showed rank-order stability over one and six years (coecients ranged between .48
and .73). One explanation for the stability of experiences can be found in the way in which workfamily conict was operationalized. The participants were asked to evaluate WFC and FWC on a general level without
attaching it to any specic time period or episode at work or within family. Therefore, the participants may
have reported their average opinion that has been formed over a longer period of time instead of reporting
their immediate experience at each measurement time. Thus, more specic measures would be needed for better tapping the often assumed uctuating nature of workfamily interface (cf. Ford et al., 2007). Another
explanation for the stability of the experience of the workfamily conict may lie in the role of personality
traits as they have been shown to be related to the experiences of WFC and FWC (e.g., Rantanen, Pulkkinen,
& Kinnunen, 2005; Wayne, Musisca, & Fleeson, 2004). Personality traits show considerable stability in adulthood (Rantanen, Metsapelto, Feldt, Pulkkinen, & Kokko, 2007; Roberts & DelVecchio, 2000), and therefore,
each participants personality induced adaptation to changing work and family environments might have generated the stable nature of WFC and FWC experiences.
The stability was higher for WFC than for FWC. The results suggest that the WFC is less context-bound
than FWC indicating that the interim eects (that is, the eects of unobserved events during measurement
interval; de Lange et al., 2003) are stronger for FWC than for WFC. This nding might be related, however,
also to the fact that FWC was reported less than WFC. Therefore even a mild change in some individuals
FWC experiences (e.g., from never to from time to time) may exert a large eect on the rank-order stability
in a group level. The present study cannot rule out this possibility.
The stability of FWC was lower across six years than across one year as expected (Hypothesis 1). An
explanation for this dierence may be in stronger interim eects over six years compared to one year time

48

J. Rantanen et al. / Journal of Vocational Behavior 73 (2008) 3751

lag. It was obtained that during six years the participants family structure changed more than during one
year (i.e., in Study 2 children got older, new were born and there where also more changes in marital relations than in Study 1). The stability of WFC was signicantly higher across six years than across one year,
which was against expectations (Hypothesis 1). However, as the dierence in stability coecients (.69 in
Study 1 vs. .73 in Study 2) was relatively small, this result is to be considered with caution and needing
a replication.
The ndings both from one-year and six-year follow-ups showed that neither WFC nor FWC had longterm reciprocal associations with job exhaustion, low marital adjustment, parental stress, or psychological distress. Hence, the Hypotheses 2, 3 and 4 predicting these associations were not supported, most likely because
of the long time lag between the measurements. Hypotheses 2, 3 and 4 were based on the integrative model of
workfamily interface which was originally successfully tested with a cross-sectional design (Frone, Yardley
et al., 1997), and which has received support from short-term studies with a time-lag of less than a year
(Demerouti et al., 2004; Kelloway et al., 1999; Leiter & Durup, 1996).
Another reason for the lag of hypothesized associations may be the fact that we used a time-based work
family conict measure. Kelloway and colleagues (1999), who investigated both strain- and time-based work
family conicts separately, found signicant cross-lagged relations only between strain-based WFC/FWC and
stress. On the other hand, the longitudinal study across four years by Frone, Russell and colleagues (1997)
with the same workfamily conict measure used (as in the present study), showed that at least FWC was negatively related to psychological well-being (predicting depression). One further reason for the present zero
ndings might be the fact that, in the SEM analysis we examined simultaneously both conict (Frone, Yardley
et al., 1997) and spillover (Edwards & Rothbard, 2000) perspectives on workfamily interface and psychological well-being, of which only spillover perspective was supported. The simultaneous analysis of workfamily
conict and workfamily spillover eects on psychological well-being was not included in the study of Frone,
Russell and his colleagues (1997).
In the present study, the following cross-lagged spillover relations between domain specic (job exhaustion,
marital adjustment, parental stress) and general well-being (psychological distress) were found. First, job
exhaustion preceded psychological distress both across one- and six-year periods. Second, low marital adjustment preceded psychological distress, and third, psychological distress preceded parental stress across one
year. These ndings supported partially Hypothesis 6, concerning spillover between domain specic and general well-being, but not Hypothesis 5, concerning spillover between work- and family-related well-being. The
chain of cross-lagged relations across a one-year time lag in Study 1 suggests that well-being and adjustment at
work and in marital relationship have a central role in maintaining psychological functioning in everyday life
(cf. Barnett et al., 1995) which, in turn, is a base for sense of adequacy and ability to function as a parent. The
results from Study 1 also support the idea that general psychological well-being and mood may act as a transmitter of psychological strain across work and family domains (cf. Edwards & Rothbard, 2000; Heller &
Watson, 2005; Leiter & Durup, 1996).
As all the ndings in the present study were gender-equal, they lent strong support for Hypothesis 7 predicting that there are no gender dierences either in the stability or in the cross-lagged relations of workfamily interface. Hence, the social-role hypothesis, which emphasizes gender similarities in the workfamily
interface experiences (Voydano, 2002), was also supported. This nding may reect the cultural context in
which the study was conducted. Finland belongs to the Nordic welfare states, where the state has an active
role in reconciling the demands of work and family lives. For one example, in Finland both fathers and mothers benet from a well-functioning communal day-care system with highly qualied childcare sta for children
under school age, and the legal right to reduce their working hours when children are young. These social policy arrangements may diminish the gender dierences in the workfamily interface among Finnish parents.
The present ndings are, however, also in line with a meta-analysis showing considerable symmetry between
genders in the cross-domain workfamily relations (Ford et al., 2007).
The present results are most generalizable into countries where socio-cultural gender expectations and practices are characterized as egalitarian rather than traditional, as in Nordic countries. At the same time, the present study oers insights for cross-cultural comparisons between working adults experiences in dierent
countries because the participants represented a wide array of occupations and socioeconomic backgrounds.
Casper and colleagues (2007) have emphasized that workers in occupations other than managerial or

J. Rantanen et al. / Journal of Vocational Behavior 73 (2008) 3751

49

professional positions and from outside the United States are clearly underrepresented in the workfamily
research literature.
A limitation of the present study was that the abridged scales of WFC, FWC, marital adjustment and
parental stress were used. The data both in Study 1 and 2 were derived from larger research projects including
workfamily issues as one theme among many others. To avoid exhausting the participants, density and conciseness in the scales were aimed at. The coecient alphas for the workfamily conict scales were not as high
as in the other scales because both WFC and FWC consisted of only two items. However, if a scale with a few
items produces a very high reliability coecient it is likely to be an indication of tautology in items (Brewerton
& Millward, 2001, p. 89). The SEM analyses showed that the construct validity of the workfamily scales was
good: the items used measured the underlying phenomenon equally well both across time, genders, and study
samples (cf. Footnote 1 and Fig. 1) conrming the justied use of these scales.
The other restrictions of the present study were the facts that, it was limited to time-based workfamily
conict (excluding strain- and behavior-based WFC/FWC; see Carlson et al., 2000), the positive side of the
workfamily interface was not included (e.g. Greenhaus & Parasuraman, 1999), and workfamily conict
and well-being were studied from the perspective of the working adults based on self-reports (excluding
spouses and children in the family, and supervisors and co-workers in the workplace as informants; cf.
Hammer, Cullen, Neal, Sinclair, & Sharo, 2005).
There are at least three possibilities maintaining the stability of workfamily conict experiences obtained
in the present study which should be considered for the implications of the study. First, the situational work
and family characteristics may remain stable; second, the way workfamily conict is operationalized may
produce stability, and third, the stability of personality in adulthood may create stability into workfamily
conict experiences. Equally important is to examine whether the moderate rank-order stability of WFC
and FWC are homogeneous phenomena among adults, or is there an unobserved heterogeneity in a given population. If this is the case, for some adults the experiences of WFC/FWC may change dramatically, whereas
for others changes may be very mild or non-existent. Consequently, the future research focusing on stability
and change in workfamily conict would benet from techniques identifying latent classes with dierent
types of workfamily conict trajectories over time (e.g., Factor Mixture Modeling; Lubke & Muthen,
2005). The closer analyses of these latent classes and factors predicting a class membership would be helpful
in planning well allocated, ecient workfamily benets for working adults.
The present results indicated that time-based workfamily conict was neither an antecedent nor a consequence of psychological well-being but job exhaustion did exert a negative spillover eect on general psychological well-being both in men and women, and across one- and six-year follow-ups. These results do not
dispute the importance of cross-sectional and short-term longitudinal ndings showing that workfamily conict and psychological strain are intertwined (e.g., Allen et al., 2000; Demerouti et al., 2004; Kelloway et al.,
1999; Leiter & Durup, 1996). When the present ndings are combined to these earlier ndings they indicate
that if organizations want to take care of their employees, they should pay special attention to the experience
of mental strain among their employees. Oering extime or reduced working hours to meet the responsibilities of work and family life is beneting but it may not be enough, because the more primary correlate of low
psychological well-being may be strain-based than time-based workfamily conict (cf. Kelloway et al., 1999).
This is not to say that time-based workfamily conict does not matter at all, but that its role in the long run
maybe additive in relation to other established antecedents of psychological well-being. Therefore, organizations should ensure that they are focusing on all relevant issues in their human resource policies. Of these policies, the prevention and reduction of job exhaustion in todays competitive working life is one of the most
challenging and, in the light of the present study, the most important tasks.

Acknowledgments
This paper was prepared as a part of the project Human Development and Its Risk Factors, nanced by
the Academy of Finland (Finnish Center of Excellence Program No. 44858 for 20002005). The research
project Economic Crisis, Job Insecurity and the Household was nancially supported by the Academy of
Finland (Grant No. 43553).

50

J. Rantanen et al. / Journal of Vocational Behavior 73 (2008) 3751

References
Abidin, R. R. (1990). Parenting stress index manual. University of Virginia: Pediatric Psychology Press.
Allen, T. D., Herst, D. E. L., Bruck, C. S., & Sutton, M. (2000). Consequences associated with work-to-family conict: A review and
agenda for future research. Journal of Occupational Health Psychology, 5, 278308.
Barnett, R. C., Raudenbush, S. W., Brennan, R. T., Pleck, J. H., & Marshall, N. L. (1995). Change in job and marital experiences and
change in psychological distress: A longitudinal study of dual-earner couples. Journal of Personality and Social Psychology, 69,
839850.
Brewerton, P., & Millward, L. (2001). Organizational research methods: A guide for students and researchers. London: SAGE Publications.
Busby, D. M., Crane, D. R., Larson, J. H., & Christensen, C. (1995). A revision of the dyadic adjustment scale for use with distressed and
nondistressed couples: Construct hierarchy and multidimensional scales. Journal of Marital & Family Therapy, 21, 289308.
Byron, K. (2005). A meta-analytic review of workfamily conict and its antecedents. Journal of Vocational Behavior, 67, 169198.
Carlson, D. S., Kacmar, K. M., & Williams, L. J. (2000). Construction and initial validation of a multidimensional measure of work
family conict. Journal of Vocational Behavior, 56, 249276.
Casper, W. J., Eby, L. T., Bordeaux, C., Lockwood, A., & Lambert, D. (2007). A review of research methods in IO/OB workfamily
research. Journal of Applied Psychology, 92, 2843.
de Lange, A. H., Taris, T. W., Kompier, M. A. J., Houtman, I. L. D., & Bongers, P. M. (2003). The very best of the millennium:
Longitudinal research and the demand-control-(support) model. Journal of Occupational Health Psychology, 8, 282305.
Demerouti, E., Bakker, A. B., & Bulters, A. J. (2004). The loss spiral of work pressure, work-home interference and exhaustion:
Reciprocal relations in a three-wave study. Journal of Vocational Behavior, 64, 131149.
Eby, L. T., Casper, W. J., Lockwood, A., Bordeaux, C., & Brinley, A. (2005). Work and family research in IO/OB: Content analysis and
review of the literature (19802002). Journal of Vocational Behavior, 66, 124197.
Edwards, J. R., & Rothbard, N. P. (2000). Mechanisms linking work and family: Clarifying the relationship between work and family
constructs. Academy of Management Review, 25, 178199.
Ford, M. T., Heinen, B. A., & Langkamer, K. L. (2007). Work and family satisfaction and conict: A meta-analysis of cross-domain
relations. Journal of Applied Psychology, 92, 5780.
Frone, M. R., Russell, M., & Cooper, M. L. (1992). Antecedents and outcomes of workfamily conict: Testing a model of the work
family interface. Journal of Applied Psychology, 77, 6578.
Frone, M. R., Russell, M., & Cooper, M. L. (1997). Relation of workfamily conict to health outcomes: A four-year longitudinal study
of employed parents. Journal of Occupational and Organizational Psychology, 70, 325335.
Frone, M. R., Yardley, J. K., & Markel, K. S. (1997). Developing and testing an integrative model of the workfamily interface. Journal of
Vocational Behavior, 50, 145167.
Geurts, S. A. E., & Demerouti, E. (2003). Work/non-work interface: A review of theories and ndings. In M. J. Schabracq, J. A. M.
Winnubst, & C. L. Cooper (Eds.), The handbook of work and health psychology (pp. 279312). Chichester: John Wiley & Sons.
Goldberg, D. P. (1972). The detection of psychiatric illness by questionnaire: A technique for the identication and assessment of nonpsychotic psychiatric illness. Oxford, England: Oxford U. Press.
Greenhaus, J. H., & Beutell, N. J. (1985). Sources and conict between work and family roles. Academy of Management Review, 10, 7688.
Greenhaus, J. H., & Parasuraman, S. (1999). Research on work, family, and gender. In G. N. Powell (Ed.), Handbook of gender and work
(pp. 391412). Thousand Oaks, CA: Sage Publications.
Hammer, L. B., Cullen, J. C., Neal, M. B., Sinclair, R. R., & Sharo, M. V. (2005). The longitudinal eects of workfamily conict and
positive spillover on depressive symptoms among dual-earner couples. Journal of Occupational Health Psychology, 10, 138154.
Heller, D., & Watson, D. (2005). The dynamic spillover of satisfaction between work and marriage: The role of time and mood. Journal of
Applied Psychology, 90, 12731279.
Hu, L., & Bentler, P. M. (1999). Cutt-o criteria for t indices in covariance structure analysis: Conventional criteria versus new
alternatives. Structural Equation Modeling, 6, 155.
Hurrell, J. J. J., Nelson, D. L., & Simmons, B. L. (1998). Measuring job stressors and strains: Where we have been, where we are, and
where we need to go. Journal of Occupational Health Psychology, 3, 368389.
Kelloway, E. K. (1998). Using LISREL for structural equation modeling: A researchers guide. Thousand Oaks, CA, US: Sage Publications,
Inc.
Kelloway, E. K., Gottlieb, B. H., & Barham, L. (1999). The source, nature, and direction of work and family conict: A longitudinal
investigation. Journal of Occupational Health Psychology, 4, 337346.
Kinnunen, U., Natti, J., Happonen, M., Kalliolahti, M., Kelhala, A., & Mauno, S. (2000). Kokemuksia tyosta ja perheesta laman
jalkeisessa Suomessa. [Experiences of work and family life in Finland after the economic recession of the early 1990 s.]. University of
Jyvaskyla. Reports from the Family research Unit, 12.
Kinnunen, U., Geurts, S., & Mauno, S. (2004). Work-to-family conict and its relationship with satisfaction and well-being: A one-year
longitudinal study on gender dierences. Work & Stress, 18, 122.
Kinnunen, U., & Pulkkinen, L. (2003). Childhood socio-emotional characteristics as antecedents of marital stability and quality. European
Psychologist, 8, 223237.
Lambert, S. J. (1990). Processes linking work and family: A critical review and research agenda. Human Relations, 43, 239257.
Leiter, M. P., & Durup, M. J. (1996). Work, home, and in-between: A longitudinal study of spillover. Journal of Applied Behavioral
Science, 32, 2947.

J. Rantanen et al. / Journal of Vocational Behavior 73 (2008) 3751

51

Little, T. D. (1997). Mean and covariance structures (MACS) analyses of cross-cultural data: Practical and theoretical issues. Multivariate
Behavioral Research, 32, 5376.
Lubke, G. H., & Muthen, B. (2005). Investigating population heterogeneity with factor mixture models. Psychological Methods, 10, 2139.
Maslach, C., & Jackson, S. (1986). MBI: Maslach Burnout Inventory (2nd ed.). Palo Alto, CA: Consulting Psychologists Press.
Mesmer-Magnus, J. R., & Viswesvaran, C. (2005). Convergence between measures of work-to-family and family-to-work conict: A metaanalytic examination. Journal of Vocational Behavior, 67, 215232.
Muthen, L. K., & Muthen, B. O. (19982006). Mplus users guide (4th ed.). Los Angeles, CA: Muthen n.
Netemeyer, R. G., Boles, J. S., & McMurrian, R. (1996). Development and validation of workfamily conict and family-work conict
scales. Journal of Applied Psychology, 81, 400410.
Pulkkinen, L. (2006). The Jyvaskyla Longitudinal Study of Personality and Social Development (JYLS). In L. Pulkkinen, J. Kaprio, & R.
J. Rose (Eds.), Socioemotional development and health from adolescence to adulthood (pp. 2955). New York: Cambridge University
Press.
Rantanen, J., Metsapelto, R-L., Feldt, T., Pulkkinen, L., & Kokko, K. (2007). Long-term stability in the big ve personality traits in
adulthood. Scandinavian Journal of Psychology, 48, 511518.
Rantanen, J., Pulkkinen, L., & Kinnunen, U. (2005). The big ve personality dimensions, workfamily conict, and psychological distress:
A longitudinal view. Journal of Individual Dierences, 26, 155166.
Roberts, B. W., & DelVecchio, W. F. (2000). The rank-order consistency of personality traits from childhood to old age: A quantitative
review of longitudinal studies. Psychological Bulletin, 126, 325.
Ryan, A. M., Chan, D., Ployhart, R. E., & Slade, L. A. (1999). Employee attitude surveys in a multinational organization: Considering
language and culture in assessing measurement equivalence. Personnel Psychology, 52, 3758.
Satorra, A., & Bentler, P. M. (1999). A scaled dierence chi-square test statistic for moment structure analysis. Available from: <http://
www.stat.ucla.edu/papers/preprints/260/>.
Schnabel, K. (1996). Latent dierence models as alternatives to modeling residuals of cross-lagged eects. In U. Engel & J. Reinecke
(Eds.), Analysis of change: Advanced techniques in panel data analysis (pp. 253278). Berlin: Walter de Gruyter.
Spanier, G. B. (1976). Measuring dyadic adjustment: New scales for assessing the quality of marriage and similar dyads. Journal of
Marriage & the Family, 38, 1528.
Statistical Oce of the European Communities (2006). A statistical view of the life of women and men in the EU25. Eurostat News
Release, 29.
Steiger, J. H. (1990). Structural model evaluation and modication: An interval estimation approach. Multivariate Behavioral Research,
25, 173180.
Stephens, G. K., & Sommer, S. M. (1996). The measurement of work to family conict. Educational and Psychological Measurement, 56,
475486.
Voydano, P. (2002). Linkages between the workfamily interface and work, family, and individual outcomes: An integrative model.
Journal of Family Issues, 23, 138164.
Wayne, J. H., Musisca, N., & Fleeson, W. (2004). Considering the role of personality in the workfamily experience: Relationships of the
big ve to workfamily conict and facilitation. Journal of Vocational Behavior, 64, 108130.
Westman, M., Etzion, D., & Gortler, E. (2004). The workfamily interface and burnout. International Journal of Stress Management, 11,
413428.

You might also like