Gender and Marital Satisfaction in Early Marriage

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

LAWRENCE A.

KURDEK Wright State University

Gender and Marital Satisfaction Early in Marriage:


A Growth Curve Approach

The purpose of this study is to assess differences cal levels. For example, Kiecolt-Glaser and
between husbands and wives (N ¼ 526 couples Newton (2001) reviewed evidence showing that
at the first assessment) on (a) growth curves although men are more physiologically sensi-
over the first 4 years of marriage for psycholog- tive than women to acute stressors (e.g., men
ical distress, marriage-specific appraisals, show larger changes in blood pressure and epi-
spousal interactions, social support, and mari- nephrine to laboratory stressors such as doing
tal satisfaction; (b) the strength of intraspouse mental arithmetic in front of an audience),
links and cross-spouse links involving the initial women show stronger and more durable physio-
assessment of the first four variables and the logical changes to marital conflict than men do.
growth curve for marital satisfaction; and (c) Other biologically oriented theories propose
differences between spouses headed for divorce that men and women differ in areas in which
versus spouses in stable marriages on the they have faced distinct adaptive challenges in
growth curves for all five variables. On the the course of their evolutionary past (see review
basis of evidence that interspouse differences by Simpson & Gangestad, 2001). For example,
were largely nonsignificant, there was little sup- in a meta-analysis of evidence regarding strate-
port for the view that there are his and her ver- gies used to attract a mate, Schmitt (2002)
sions of the processes that affect marital found that whereas appearance-related tactics
outcomes. (e.g., enhancing one’s appearance) were judged
to be more effective for women than for men (a
moderate effect), resource-related tactics (e.g.,
Based primarily on the pioneering work of Ber-
demonstrating one’s financial security) were
nard (1982), family scholars often regard mar-
judged to be more effective for men than for
riage as coming in his and her versions. Indeed,
women (a large effect). This pattern of findings
two classes of theories support the position
is consonant with the idea that men value physi-
that men and women experience their close rela-
cal attractiveness as a response to their distinct
tionships in different ways. The class of bi-
evolutionary challenge of finding a mate who
ological theories posits that men and women
can bear children, and that women value re-
process events in their relationships differently
source prospects as a response to their distinct
at the cardiovascular, endocrinological, immu-
evolutionary challenge of finding a mate who
nological, neurosensory, and neurophysiologi- will provide for a family.
The class of social psychological theories
posits that men and women differ in ways di-
Department of Psychology, Wright State University, rectly relevant to relationship functioning, such
Dayton, OH 45435-0001 (larry.kurdek@wright.edu). as the content and structure of how the self is
Key Words: gender differences, growth curves, longitudinal construed (e.g., Cross & Madson, 1997), per-
study, marital satisfaction. sonality traits (Costa, Terracciano, & McCrae,

68 Journal of Marriage and Family 67 (February 2005): 68–84


Gender and Marital Satisfaction 69

2001), and socialized roles (Eagly, 1987). For interest than gender differences in average
example, Cross and Madson reviewed results levels of individual variables because the for-
indicating that whereas women represent their mer set of differences could mean that the pro-
self-attributes and preferences within the narrow cesses that regulate marital functioning differ
context of dyadic relationships, men represent for husbands and for wives.
them within the broad context of group mem- Social psychological theories in particular
berships. In a meta-analysis of studies con- could be used to argue that relationships are
ducted within the framework of the Big Five more rule-governed for women than they are
model of personality, Costa et al. reported that for men. For example, Eagly and Wood (1999)
women are higher than men in neuroticism (a speculated that as one consequence of accom-
large effect), agreeableness (a large effect), and modating to domestic roles and to female-
extraversion (a medium effect). Finally, Eagly dominated occupations, women have acquired
argued that the underlying cause of gender- communal characteristics that highlight the
typed social behavior is the division of labor adaptive value of being interpersonally skilled.
between women and men. As a consequence of The assertion that wives function as relationship
this division, women are expected to behave experts often has been used to explain findings
communally, and these expectations get trans- that outcomes for married couples are predicted
lated into behaviors such as cooking and child better by information from wives than by infor-
care, which equip them for fulfilling domestic mation from their husbands (e.g., Bentler &
roles. Men, on the other hand, are expected to Newcomb, 1978; Floyd & Markman, 1983;
behave agentically, and these expectations get Karney, Bradbury, Fincham, & Sullivan, 1994).
translated into behaviors relevant to earning This assertion also has been used to explain
capacity that equip them for fulfilling occupa- a pattern of cross-spouse effects in which hus-
tional roles. bands’ marital outcomes are more strongly
The biological and social psychological theo- influenced by their wives’ predictors than
ries might lead one to expect large differences wives’ marital outcomes are influenced by their
between men and women in average values of husbands’ predictors (e.g., Davila, Bradbury, &
marriage-relevant variables. Indeed, as noted Fincham, 1998; Karney & Bradbury, 1997;
earlier, large effect sizes have been found in Sabatelli & Bartle-Haring, 2003).
meta-analytic reviews of gender differences in Despite claims that women function as rela-
both biological variables (Schmitt, 2002) and tionship experts, in a meta-analysis of studies in
social psychological variables (Costa et al., which information was collected from both
2001). However, because most of the evidence spouses from the same couple, Karney and
pertinent to marriage involves comparisons be- Bradbury (1995) found that demographic, per-
tween married men and married women from sonality, and behavioral interaction variables
different marriages, it cannot be used to draw were linked to marital satisfaction to the same
conclusions about the experiences of men and degree for husbands as they were for their
women in the same marriage. Given that wives. In addition, information from husbands
spouses within the same marriage are often sim- predicted marital stability as well as information
ilar to each other as a result of assortative mat- from their wives did. Although Karney and
ing (Buss, 1984; Watson, Hubbard, & Wiese, Bradbury concluded that gender differences in
2000), comparisons between husbands and wives marriage have been exaggerated, it is now
from the same marriage may yield smaller differ- widely acknowledged that analyses with couple-
ences than comparisons between husbands and level data require special statistical techniques
wives from different marriages. (e.g., Schulz, Cowan, Cowan, & Brennan,
Although much of the literature on gender 2004). Because these techniques were not
and marriage has focused on differences be- employed in most of the studies reviewed by
tween husbands and wives on average levels of Karney and Bradbury, the issue of gender differ-
variables relevant to marriage, gender differ- ences in marital processes warrants further
ences also can be examined with regard to study.
whether links between marriage-relevant varia- Differences between husbands and wives in
bles are of the same strength for husbands and the correlates of marital outcomes traditionally
for wives. Gender differences in patterns of have been assessed by doing a separate analysis
such linkages may be of greater theoretical for each spouse on data collected at only one
70 Journal of Marriage and Family

point in time, and then determining whether the years of marriage (Bramlett & Mosher, 2001).
effect (e.g., a correlation or a partial regression Because marriages are complex systems, multi-
coefficient) is significant for one or both ple factors are needed to understand how they
spouses. There are at least four problems with begin, are maintained, and end. In a recent
this procedure. First, although an effect may account of these factors, Huston (2000) argued
reach a critical value for statistical significance that relationships need to be studied at three
for one spouse and not the other, the actual levels. The first level involves the individual
magnitude of the effect may not differ statisti- partner and refers to the personal characteristics
cally between spouses. Second, the effects for that each partner brings to the relationship and
husbands and wives may not be accurately esti- the ways in which each partner actively filters
mated because separate analyses assume that information about the relationship. The second
spouses’ scores are independent of each other. level involves partner interactions and refers to
In a structural equation modeling framework in how partners behave toward each other. The
which there are husband and wife versions of final level involves societal forces and under-
the same outcome score (e.g., ratings of marital scores the fact that relationships exist within the
satisfaction), the residuals associated with context of other social relationships.
spouses’ outcome scores are typically moder- Based on Huston’s (2000) social ecological
ately correlated (e.g., Davey, Fincham, Beach, model, four domains of variables were of inter-
& Brody, 2001). Correlated residuals can af- est in the current study. The level of the individ-
fect the estimation of partial regression coeffi- ual partner is represented by two domains of
cients and their standard errors (Kurdek, 2003; variables. The first domain is that of individual
Verbeke & Lesaffre, 1997). differences variables that represent stable or
Third, the interdependence between spouses’ enduring characteristics (Karney & Bradbury,
scores is reflected not only in the presence of 1995) that influence the extent to which partners
correlated errors in the outcome score for each adapt to the challenges posed by the marriage.
spouse, but also in the fact that the predictors of Psychological distress is selected based on evi-
those outcome scores are likely to operate in dence that it represents a stable individual
both an intraspouse fashion and a cross-spouse differences variable (e.g., McCrae, 1993) and
fashion. For example, a husband’s predictor that spouses high in psychological distress are
variable may affect his own marital outcome at risk for poor marital outcomes (Karney &
and that of his wife, and a wife’s predictor vari- Bradbury, 1995). The second domain is that of
able may affect her own marital outcome and marriage-specific appraisal variables that repre-
that of her husband. Because the variables used sent beliefs and attitudes regarding either the
as predictors for husbands and for wives may self or the spouse in the marriage. Because of
themselves be related to each other through the multidimensional nature of these beliefs and
a process of assortative mating (Watson et al., attitudes (e.g., Baldwin, 1992), four variables
2000), the unique effects associated with either widely used in past research were selected.
intraspouse linkages or cross-spouse linkages These include love for one’s spouse (Rubin,
need to be determined by assessing both types 1970), liking of one’s spouse (Rubin), intrinsic
of linkages simultaneously (Kenny, Mohr, & motives for being in the marriage (Rempel,
Levesque, 2001). Finally, because marital pro- Holmes, & Zanna, 1985), and perceptions of
cesses are dynamic rather than static (Huston, levels of trust in the marriage (Rempel et al.).
2000; Karney & Bradbury, 1995), single assess- Given that these marriage-specific appraisals
ments do not capture the mechanisms by which are interrelated (e.g., Attridge, Berscheid, &
marriage changes over time. Simpson, 1995), a single latent variable was
On the basis of the premise that change is used to represent them.
a core marital process (Huston, Caughlin, Houts, The level of partner interactions is repre-
Smith, & George, 2001; Karney & Bradbury, sented by the domain of spousal interaction
1995, 1997), the focus of the current study is variables that represent what happens between
on assessments of both spouses made annually spouses in the marriage. Because these interac-
over the first 4 years of marriage. The early tions are multidimensional in nature (Spanier,
years of marriage were of interest because 20% 1976), three variables commonly used in past
of all first marriages and 23% of all second research were selected. These included the fre-
marriages end in separation or divorce by 5 quency with which spouses agree on important
Gender and Marital Satisfaction 71

issues, the frequency of behaviors indicative of studies have assessed interspouse differences
affection/sexual interest, and the frequency of within a growth curve context, predictions based
shared activities. Given the high overlap among on biological and social psychological theories
these kinds of interactions (Sabourin, Lussier, were more easily advanced for average levels
Laplante, & Wright, 1990), a single latent vari- than for change. Because women are more likely
able is used to represent them. than men to amplify negative moods by ruminat-
Finally, the level of societal forces is repre- ing about them and their causes as a result of
sented by the domain of social support varia- genetic or hormonal predispositions (Nolen-
bles that represent the fact that couples are Hoeksema, 2003), the average levels for psycho-
embedded in social contexts that involve family logical distress were expected to be higher for
members and friends, and that the forces from wives than for their husbands. Because women
these social contexts affect the stability of rela- are socialized more strongly than men to define
tionships (Milardo & Helms-Erikson, 2000). themselves in terms of their connections with
Satisfaction with perceived levels of social sup- others (Cross & Madson, 1997), and because
port was selected based on evidence that it ac- women have faced the adaptive challenge of
counts for variability in marital outcomes above using social resources (e.g., groups of other
and beyond that accounted for by personality women) to protect themselves and their children
variables (e.g., Kurdek, 1989). (Buss, 1998), the average levels for social sup-
Because four annual assessments on the same port were expected to be higher for wives than
variables were obtained, each spouse’s trajec- for their husbands. Finally, because women may
tory of change on each variable was described be more strongly motivated than men to con-
with regard to two aspects of a growth curve. struct positive evaluations about themselves,
The first aspect was the average level of that their spouses, and their marital interactions
variable over all assessments (the intercept of (Murray, Holmes, & Griffin, 1996), average
the growth curve), and the second aspect was levels for marriage-specific appraisals, spousal
the rate of linear change in that variable over all interactions, and marital satisfaction were
assessments (the slope of the growth curve). expected to be higher for wives than for their
Estimating husbands’ and wives’ trajectories of husbands.
change within the same model has several The second purpose of this study is to com-
distinct advantages (Newsom, 2002). First, be- pare differences between husbands and their
cause models were estimated using full infor- wives with regard to the strength with which
mation maximum likelihood techniques, all variables from the individual differences, mar-
available data were used, including those from riage-specific appraisals, spousal interaction,
couples with only one assessment. Second, in- and social support domains were linked to the
terdependencies within each spouse’s scores growth curves for marital satisfaction. Given
over time were handled by allowing the error strong practical interest in identifying early risk
terms for husbands’ outcome scores and the factors for later marital distress (Karney &
error terms for wives’ outcome scores to be cor- Bradbury, 1995), the assessments of the varia-
related across time. Third, interdependencies bles from the individual differences, marriage-
between spouses’ scores were handled by esti- specific appraisals, spousal interaction, and
mating intraspouse links and cross-spouse links social support domains taken at the beginning
simultaneously, and by allowing the error terms of the marriage were used to predict both aver-
between spouses’ outcome scores to be corre- age level (the intercept) and rate of linear
lated at each assessment. Fourth, because change (the slope) for marital satisfaction over
spouses’ intercepts and slopes were estimated in the first 4 years of marriage. Both intraspouse
the same model, differences between spouses’ links (one’s own predictor linked to one’s own
intercepts and differences between spouses’ growth curve for marital quality) and cross-
slopes could be tested statistically (Schulz et al., spouse links (one’s own predictor linked to the
2004). growth curve for marital quality of one’s
The first purpose of this study is to assess dif- spouse) were examined as one reflection of the
ferences between husbands and their wives in interdependent processes underlying marital
average levels (intercepts) and rates of change functioning (Bond & Kenny, 2003).
(slopes) for the growth curves for the variables Despite Karney and Bradbury’s (1995) find-
from the five domains of interest. Because few ings that intraspouse and cross-spouse effects
72 Journal of Marriage and Family

did not differ between husbands and wives, gen- a growth curve profile indicative of greater dis-
der differences in intraspouse and cross-spouse tress for each outcome variable. This profile
links were examined because many of the intra- included not only higher average levels of dis-
spouse and cross-spouse effects that were part tress (intercepts), but also larger increases in
of Karney and Bradbury’s meta-analysis were distress over time (slopes). Thus, spouses from
not obtained with adequate statistical proce- couples who would eventually divorce were ex-
dures. On the basis of the assertion from social pected to show higher average levels of psycho-
psychological theories that wives function as logical distress and lower average levels of
relationship experts, it was expected that the positive marriage-specific appraisals, positive
links between the various predictors and marital spousal interactions, satisfaction with social sup-
satisfaction would be stronger for wives than port, and marital satisfaction as compared with
for husbands, and that wives’ predictors would spouses from stable couples. Further, spouses
be linked to their husbands’ marital satisfaction from couples who would eventually divorce
more strongly than husbands’ predictors would were expected to show larger increases in psy-
be linked to their wives’ marital satisfaction. chological distress and larger decreases in levels
Because previous work involving the link of positive marriage-specific appraisals, positive
between earlier assessments and change in mar- spousal interactions, satisfaction with social sup-
ital satisfaction is limited, and because cross- port, and marital satisfaction as compared with
spouse effects have not been routinely obtained, spouses from stable couples.
the most confident predictions were made for Although showing that the growth curves for
the intraspouse links involving average levels. spouses from couples headed for divorce dif-
Specifically, high levels of psychological dis- fered from those of spouses from stable couples
tress at the start of the marriage were expected is of interest in its own right, of particular inter-
to be related to low average levels of mar- est in this study was whether husbands and
ital satisfaction (Karney & Bradbury, 1995), wives differed in the strength of these effects.
whereas favorable marriage-specific appraisals, That is, for both average levels and rates of lin-
positive spousal interactions, and high levels of ear change, was the difference between hus-
satisfaction with social support at the start of the bands from divorced couples and husbands from
marriage were each expected to be related to stable couples equal to the difference between
high average levels of marital satisfaction wives from divorced couples and wives from
(Kurdek, 1988, 1992; Rempel et al., 1985; stable couples? If wives serve as the barometers
Rubin, 1970). of marital health (Floyd & Markman, 1983),
The final purpose of this study is to compare then differences between wives from divorced
whether husbands headed for divorce differ versus stable couples should be larger than those
from wives headed for divorce with regard to between husbands from divorced versus stable
the growth curves for psychological distress, couples.
marriage-specific appraisals, spousal interac-
tion, social support, and marital satisfaction.
METHOD
Identifying the distinctive features of divorce-
prone spouses was of practical interest given
Participants
the stress associated with divorce (Amato,
2000), and was of methodological interest given Participants were recruited from the marriage
that previous prospective longitudinal studies licenses published in the local newspaper of
(Gottman, Coan, Carrere, & Swanson, 1998; a midwestern city from May 1986 through
Huston et al., 2001; Karney & Bradbury, 1997) January 1988. Of the 7,899 couples who re-
have identified only a relatively small number ceived a letter explaining the study, 1,407 indi-
of couples who provided information from mul- cated an interest in the study. This response rate
tiple assessments shortly before the divorce of 18% is similar to that obtained from other
occurred. studies recruiting unpaid participants from pub-
On the basis of previous studies using growth lic records (e.g., 17% by Spanier, 1976, and
curve methods (e.g., Huston et al., 2001; 18% by Davila, Karney, & Bradbury, 1999).
Karney & Bradbury, 1997), it was expected Completed surveys were returned by 538 cou-
that, compared with spouses from stable cou- ples at the first assessment (Year 1), shortly
ples, spouses headed for divorce would show after their wedding. This return rate of 38% is
Gender and Marital Satisfaction 73

similar to the rate of 33% obtained by Kurdek TABLE 1. DESCRIPTIVE DATA FOR HUSBANDS’ AND
and Schmitt (1986) in a study that involved WIVES’ DEMOGRAPHIC VARIABLES AT EACH OF THE
a survey of similar length, but required anony- FOUR ANNUAL ASSESSMENTS
mous participation of both partners. For the cur-
Year 1 Year 2 Year 3 Year 4
rent study, 12 couples (2% of the sample) were
dropped from the sample because one or both Mean age in years
spouses were previously widowed, reducing the Husbands 29.19 30.22 31.47 32.91
base sample size to 526. Wives 27.14 28.29 29.54 30.89
Data were also obtained at three annual Mean level of annual income
follow-up assessments (Year 2, Year 3, and Husbands 5.80 6.77 7.31 7.88
Year 4). Over this time period, sample size Wives 4.02 4.58 4.95 5.21
decreased because couples separated, withdrew % White
from the study because they no longer wished Husbands 95 96 96 96
to participate, or were dropped from the study Wives 95 97 97 97
because they did not return their surveys. Over % with baccalaureate degree
the 4 years, 265 couples (50% of the sample) Husbands 41.00 49.70 54.90 55.80
provided data at all assessments (396 at Years 1 Wives 40.10 48.00 55.80 58.50
and 2, and 310 at Years 1–3); 57 couples (11% % employed
of the sample) separated (23 at Year 2, 21 at Husbands 92 92 95 94
Year 3, and 13 at Year 4); 64 couples (12% of Wives 79 79 80 77
the sample) withdrew from the study (41 at % in first marriage
Year 2, 17 at Year 3, and 6 at Year 4), and 140 Husbands 65 66 66 66
couples (27% of the sample) were dropped from Wives 64 66 68 69
the study (66 at Year 2, 48 at Year 3, and 26 at Mean # children
Year 4). Husbands and wives are described in Husbands 0.13 0.22 0.34 0.48
Table 1 for each of the four assessments with Wives 0.41 0.48 0.56 0.67
regard to mean age, mean annual personal Mean # stepchildren
income (on a 12-interval scale), percent Husbands 0.35 0.30 0.27 0.23
White, percent with at least a baccalaureate Wives 0.07 0.04 0.05 0.04
degree, percent employed, percent in first Mean years of cohabitation 0.75 1.69 2.72 3.76
marriage, mean number of children, mean Note: For annual income, 1 ¼ less than $5,000, 12 ¼
number of stepchildren, and mean years of $50,000 or more.
cohabitation. The large percentage of White
respondents and the large percentage of re-
spondents with at least a baccalaureate degree
indicate that the sample is not a representa- Measures
tive one. Demographic variables. Spouses provided in-
formation about age, personal annual income
(1 ¼ less than $5,000, 12 ¼ $50,000 or more),
Procedure level of education (1 ¼ less than seventh grade,
8 ¼ doctoral degree), race, whether they were
At each assessment, both partners from each employed, divorce history, the number of
couple were mailed identical surveys. Spouses months they had lived with their partners, and
were directed to respond to their surveys pri- whether they had (step)children.
vately and not to discuss their answers with
each other until the surveys were completed, Psychological distress. Psychological distress
although no checks were made to ensure that was assessed by the global severity index
these directions were followed. Completed sur- derived from the Symptom Checklist 90-R
veys were returned in separate postage-paid en- (Derogatis, 1994). Spouses indicated how
velopes. In the few instances in which returned much discomfort (0 ¼ not at all, 4 ¼ extremely)
surveys had missing data, they were returned to each of 90 problems caused them during the
the spouse with a request that the missing infor- past 7 days. Problem areas included somatic
mation be provided. All spouses complied with complaints, obsessions-compulsions, interper-
this request. sonal sensitivity, depression, anxiety, hostility,
74 Journal of Marriage and Family

phobias, paranoia, and psychoticism. Across tors (four for each spouse) were allowed to cor-
spouses and assessments, Cronbach’s alpha for relate, correlated errors were estimated for each
the summed composite score was .96. The aver- of the four appraisals at 1-year lags (e.g., Year 1
age interspouse correlation (as determined by and Year 2, Year 2 and Year 3) for each spouse,
the correlation between the intercepts of and correlated errors between spouses were esti-
spouses’ growth curves) was .23, p , .01. mated at each year for each of the four apprais-
als. The fit of this measurement model was very
Marriage-specific appraisals. Appraisals of the good (Comparative Fit Index [CFI] ¼ .96,
marriage were assessed by four scores. The love Tucker Lewis Index [TLI] ¼ .95, Root Mean
score and the liking score were derived from Square Error of Approximation [RMSEA] ¼
Rubin’s (1970) measure of romantic love. The .05; see Hu & Bentler, 1999).
13-item love score tapped affiliative and depen-
dent need, predisposition to help, exclusiveness, Spousal interactions. Interactions between
and absorption (e.g., I would do almost any- spouses were assessed by three scores derived
thing for my partner), whereas the 13-item lik- from Spanier’s (1976) Dyadic Adjustment
ing score tapped favorable evaluations of one’s Scale. The consensus score tapped level of
partner and the perception that the partner is agreement between partners and involved par-
similar to one’s self (e.g., I think that my partner ticipants indicating the extent of such agreement
is unusually well adjusted). For all items, re- (0 ¼ always disagree, 5 ¼ always agree) for 13
spondents indicated how true each item was for areas (e.g., handling family finances). The
them (1 ¼ not at all true, 9 ¼ definitely true). affectional expression score tapped level of
The intrinsic motives score was part of Rempel demonstrated affection between partners and
et al.’s (1985) Motivation Scale and tapped rea- involved participants using one of two response
sons for being in the relationship that were formats (0 ¼ always disagree, 5 ¼ always
inherent to the relationship itself. Participants agree, or 0 ¼ no, 1 ¼ yes) to assess four items
indicated how much of a role (0 ¼ plays no role (e.g., whether not showing love caused a prob-
at all, 8 ¼ plays a major role) each of six items lem during the past few weeks). Finally, the
(e.g., We are close and intimate) played in dyadic cohesion score tapped the number or the
forming the basis for the relationship. Finally, frequency of shared activities and involved par-
the trust score was derived from Rempel et al.’s ticipants using two response formats (0 ¼ none
Trust Scale and tapped predictability of the part- of them, 5 ¼ all of them or 0 ¼ never, 5 ¼ more
ner, dependability of the partner, and faith in often) to rate five items (e.g., How often did you
the future of the relationship. Participants indi- and your partner laugh together?). Across
cated how strongly (1 ¼ strongly disagree, 7 ¼ spouses and assessments, Cronbach’s alpha for
strongly agree) they agreed with 18 items (e.g., the consensus, affectional expression, and
Though times may change and the future is dyadic cohesion summed composite scores
uncertain, I have faith that my partner will (unit-weighted) was .85, .67, and .77, respec-
always be ready and willing to offer me strength tively. The average interspouse correlation for
and support, come what may) that described these scores was .69, .72, and .61, respectively,
their relationship. Across spouses and assess- ps , .01.
ments, Cronbach’s alpha for the summed com- Because the three spousal interaction varia-
posite love, liking, intrinsic motives, and trust bles at each year were expected to be correlated
scores was .84, .89, .75, and .88, respectively. for each spouse, their common variance was
The average interspouse correlation for these assumed to be captured by a single latent vari-
scores was .44, .34, .60, and .52, respectively, able for each spouse. This assumption was
ps , .01. tested with a confirmatory factor analysis in
Because the four marriage-specific appraisals which each spouse’s three interaction variables
at each year were expected to be correlated for at each year defined a single factor, the resulting
each spouse, their common variance was six factors (three for each spouse) were allowed
assumed to be captured by a single latent vari- to correlate, correlated errors were estimated for
able for each spouse. This assumption was each of the three interaction scores at 1-year
tested with a confirmatory factor analysis in lags for each spouse, and correlated errors
which each spouse’s four appraisals at each year between spouses were estimated at each year
defined a single factor, the resulting eight fac- for each of the three interaction scores. The fit
Gender and Marital Satisfaction 75

of this measurement model was very good spouse’s average level was obtained by weight-
(CFI ¼ .97, TLI ¼ .95, RMSEA ¼ .05). ing scores at Year 1, Year 2, Year 3, and Year 4
with a value of 1, and the latent variable repre-
Social support. Level of satisfaction with per- senting each spouse’s rate of linear change was
ceived social support was assessed by Sarason, obtained by weighting these scores with values
Sarason, Shearin, and Pierce’s (1987) Social of 3, 1, 1, and 3, respectively. The weights
Support Scale. Respondents indicate how satis- for linear change are those used in orthogonal
fied (1 ¼ very dissatisfied, 6 ¼ very satisfied) polynomial analyses in which four assessments
they were with the support they received in six are equally spaced and the intercept is to be in-
areas (e.g., Think of the people you can count terpreted as the value of the outcome score aver-
on to be dependable when you need help). aged over all assessments (Cohen, Cohen,
Across spouses and assessments, Cronbach’s West, & Aiken, 2003).
alpha for the summed composite score was .88. Because the analyses for the set of marriage-
The average interspouse correlation was .32, specific appraisal scores and the set of spousal
p , .01. interaction scores involved multiple observed
indicators at each year, spouses’ average levels
Marital satisfaction. Marital satisfaction was as- and rates of linear change for these scores were
sessed by four items from the Dyadic Satisfac- derived with second-order latent growth curve
tion subscale of Spanier’s (1976) Dyadic models (Sayer & Cumsille, 2001). For each set
Adjustment Scale. Each item involved making of scores, four first-order latent variables and
a global evaluation of the marriage. Respondents two second-order latent variables were created.
used two response formats. The first format (1 ¼ The first-order latent variables represented each
all the time, 6 ¼ never) was used to respond to relevant construct (i.e., appraisal or interaction)
the following items: How often do you discuss at each of the 4 years of assessment. Because
or have you considered divorce, separation, or the metrics for the multiple observed indicators
terminating your relationship? In general, how differed, the observed scores were first con-
often do you think that things between you and verted to z scores over year of assessment and
your partner are going well? Do you ever regret spouse. For the set of appraisal scores, four
that you married? The second format (1 ¼ observed indicators (love, liking, intrinsic moti-
extremely unhappy, 7 ¼ perfect) was used to vations, and trust) defined a latent appraisal
assess degree of happiness in the relationship. score at each year, whereas for the set of inter-
Across spouses and assessments, Cronbach’s action scores, three observed indicators (con-
alpha for the summed composite score (unit- sensus, affective expression, and cohesion)
weighted) was .86. The average interspouse cor- defined a latent interaction score at each year.
relation was .81, p , .01. The two second-order latent variables repre-
sented the average levels and rates of linear
change, respectively, for the first-order latent
RESULTS
variables at Year 1–Year 4. For the second-
order latent variable that represented average
Estimating Husbands’ and Wives’ Growth
level, the first-order latent variables at Year
Curves Within a Latent Variable Framework
1–Year 4 were each weighted with a 1. For the
For each variable, average levels (intercepts) second-order latent variable that represented
and rates of linear change (slopes) for husbands rate of linear change, the first-order latent varia-
and wives were estimated in the same model bles at Year 1–Year 4 were weighted with
with a latent variable approach using version values of 3, 1, 1, and 3, respectively.
3.00 of the Mplus program (Muthén & Muthén, Preliminary analyses indicated that attrition
2004). General information regarding conduct- in the sample was not random. Compared with
ing growth curve analyses with both spouses husbands who completed all four assessments,
from married couples can be found in Kurdek those who did not had higher average levels of
(2003). For the analyses involving psychologi- psychological distress, less positive average
cal distress, social support, and marital satisfac- levels of and stronger declines in marriage-
tion, there was only one observed indicator per specific appraisals, less positive average levels
spouse for each of the 4 years. For these var- of and stronger declines in spousal interactions,
iables, the latent variable representing each and less positive average levels of and stronger
76 Journal of Marriage and Family

declines in marital satisfaction, all ps , .01. Positive values for rate of change indicated that
Wives showed the same pattern. Thus, dummy the outcome score increased over time, whereas
variables indicating the three reasons for miss- negative values for rate of change indicated that
ing data (i.e., divorced, withdrawn from the the outcome score decreased over time. Differ-
sample, or dropped from the sample) were used ences between spouses’ average levels and dif-
as control variables in which spouses from ferences between spouses’ rates of linear change
couples who provided data at all assessments were assessed by comparing the v2 value for
served as the reference group. a model in which spouses’ average levels (or
Preliminary analyses also indicated that error rates of linear change) were freely estimated
covariances needed to be modeled. For all anal- with the v2 value for a model in which spouses’
yses, separate error variances were estimated at average levels (or rates of linear change) were
each assessment for each spouse, errors between constrained to be equal. The difference between
spouses at each assessment were allowed to cor- the two v2 values is itself a v2 value with 1
relate, and errors between adjacent assessments degree of freedom, and is presented next to the
(i.e., Year 1 and Year 2, Year 2 and Year 3, and unstandardized coefficients in Table 2 as the
Year 3 and Year 4) were allowed to correlate Husband vs. wife v2 value. Following Rosenthal
for each spouse. Because dummy variables indi- (1994), the effect size for the v2 difference value
cating whether each spouse had been divorced was expressed as r, where r ¼ Ov2/N. These
(0 ¼ no, 1 ¼ yes) did not improve the fit of any effect-size r’s are presented next to the v2 values
model, these variables were not considered fur- in Table 2. Following Cohen (1988), cut-off
ther. Following Hox (2002), the fit of each values for small, medium, and large effects were
growth curve model was assessed with the represented by r’s of .10, .30, and .50, respec-
RMSEA for which values less than .08 indicate tively. Unless indicated otherwise, all effect
acceptable fit, and values less than .05 indicate sizes reported in this study were small.
good fit (all CFIs and TLIs were above .95). With regard to psychological distress, wives
reported higher average values than their hus-
bands did. Although the decrease in psychologi-
Differences Between Husbands’ and Wives’
cal distress was significant only for husbands,
Growth Curves
the difference between spouses’ rates of change
The RMSEA associated with the model for each was not significant. With regard to marriage-
of the five outcome scores of interest is pre- specific appraisals, wives’ average values were
sented in Table 2. Each model showed at least higher than those of their husbands. Although
an acceptable level of fit. The unstandardized both spouses showed significant decreases in
coefficients for average level and rate of linear appraisals over time, this decrease was stronger
change obtained from each model are also pre- for wives than it was for their husbands. With
sented for husbands and for wives in this table. regard to spousal interactions, wives’ average

TABLE 2. ROOT MEAN SQUARE ERROR OF APPROXIMATION (RMSEA), UNSTANDARDIZED COEFFICIENTS FOR AVERAGE
LEVEL AND LINEAR CHANGE FOR HUSBANDS AND WIVES, v2 TEST FOR INTERSPOUSE DIFFERENCE IN COEFFICIENTS,
2
AND EFFECT-SIZE R FOR THE v TEST FOR INTERSPOUSE DIFFERENCE FOR DISTRESS, MARRIAGE-SPECIFIC APPRAISALS,
SPOUSAL INTERACTION, SOCIAL SUPPORT, AND MARITAL SATISFACTION VARIABLES

Average Level Linear Change

Coefficients Husband vs. Wife Coefficients Husband vs. Wife

Variable RMSEA Husband Wife v2 r Husband Wife v2 r

Distress .001 0.35 0.44 12.34** .15 0.01* 0.00 1.83 .05
Marriage-specific appraisals .04 0.06 0.04 4.27* .09 0.07** 0.09** 4.00* .08
Spousal interaction .04 0.08 0.04 8.90** .13 0.07** 0.08** 2.40 .06
Social support .06 31.23 31.75 3.47 .08 0.35** 0.38** 0.17 .01
Marital satisfaction .01 17.52 17.60 0.44 .02 0.17** 0.25** 6.51* .11
*p , .05. **p , .01.
Gender and Marital Satisfaction 77

values were higher than those of their husbands, version of the predictor of interest. The other
and both spouses showed equivalent decreases spouse’s version of the same predictor was used
over time. With regard to satisfaction with social as a control variable. For analyses involving
support, husbands and wives had equivalent psychological distress and social support, the
average values and showed equivalent decreases relevant predictor was the single observed indi-
over time. Finally, with regard to marital satis- cator at Year 1, whereas for analyses involving
faction, spouses showed equivalent average marriage-specific appraisals and spousal inter-
values. Although both spouses showed signifi- actions, the relevant predictor was the latent var-
cant decreases in marital satisfaction over time, iable derived from the set of scores at Year 1.
this decrease was stronger for wives than it was The unstandardized coefficients for the intra-
for their husbands. spouse links are presented for husbands and for
wives in the left portion of Table 3 for average
level and rate of linear change by each predictor
Differences Between Husbands’ and
variable. For descriptive purposes, the effect
Wives’ Intraspouse Links
sizes associated with these coefficients are also
The next issue addressed whether one’s own given, where r ¼ Z/O N (Rosenthal, 1994).
levels of psychological distress, marriage- Tests of differences between husbands and their
specific appraisals, spousal interaction, and so- wives in the size of these coefficients were
cial support assessed at the beginning of the made by comparing the difference between the
marriage predicted both one’s own average v2 value associated with a model in which coef-
level of marital satisfaction over the first 4 years ficients for each spouse were freely estimated,
of marriage (intercept) and change in one’s own and the v2 value associated with a model in
marital satisfaction over this same time period which coefficients for each spouse were con-
(slope) equally well for husbands and their strained to be equal. This v2 difference value is
wives. These intraspouse links were estimated presented in Table 3, along with its associated
with growth curve analyses using latent varia- effect-size r.
bles (as described earlier) in which each With regard to psychological distress, for
spouse’s latent variable representing average each spouse, high levels at the start of the mar-
level and each spouse’s latent variable repre- riage were linked to low average levels of mari-
senting change were predicted by that spouse’s tal satisfaction, and the strength of these links

TABLE 3. UNSTANDARDIZED COEFFICIENTS FOR INTRASPOUSE AND CROSS-SPOUSE LINKS BETWEEN YEAR 1
LEVELS OF DISTRESS, MARRIAGE-SPECIFIC APPRAISALS, SPOUSAL INTERACTION, AND SOCIAL SUPPORT VARIABLES
2
AND AVERAGE LEVEL OF AND LINEAR CHANGE IN MARITAL SATISFACTION FOR HUSBANDS AND WIVES, v TEST FOR
2
INTERSPOUSE DIFFERENCE, AND EFFECT-SIZE R FOR THE v TEST

Intraspouse Cross-Spouse

Husband Wife Husband vs. Wife Husband Wife Husband vs. Wife
2
Variable Coefficient r Coefficient r v r Coefficient r Coefficient r v2 r

Distress
Average 1.65** .25 1.33** .21 0.39 .03 0.89** .12 0.95** .16 0.03 .00
Change 0.01 .00 0.03 .02 0.12 .01 0.00 .00 0.03 .02 0.07 .01
Appraisals
Average 1.43** .36 1.90** .41 2.41 .02 0.30 .07 0.60** .15 1.18 .05
Change 0.06 .06 0.02 .01 0.32 .00 0.01 .01 0.02 .01 0.12 .04
Interaction
Average 1.27** .27 1.48** .27 0.33 .06 0.28 .05 0.64** .16 1.22 .04
Change 0.04 .03 0.16* .10 1.52 .02 0.05 .03 0.01 .01 0.21 .01
Support
Average 0.13** .19 0.15** .22 0.27 .02 0.04 .05 0.09** .15 1.60 .04
Change 0.01 .02 0.01 .03 0.02 .05 0.01 .06 0.00 .00 1.19 .01
*p , .05. **p , .01.
78 Journal of Marriage and Family

did not differ between spouses. With regard to regard to satisfaction with marriage-specific ap-
marriage-specific appraisals, for each spouse, praisals, spousal interactions, and social sup-
high levels at the start of the marriage were port, although cross-spouse effects were not
linked to high average levels of marital satisfac- significant for husbands, wives’ high levels for
tion, effect sizes were medium in strength for these variables at the start of the marriage were
each spouse, and the strength of these links did linked to high average levels of marital satisfac-
not differ between spouses. With regard to tion for their husbands. Nonetheless, husbands
spousal interactions, for each spouse, high and their wives did not differ in the strength of
levels at the start of the marriage were linked to any of these cross-spouse effects. Because no
high average levels of marital satisfaction, and variable showed cross-spouse links involving
the strength of these links did not differ between change in marital satisfaction, no interspouse
spouses. Finally, with regard to satisfaction with differences in these effects occurred. In sum,
social support, for each spouse, high levels at analyses involving the cross-spouse links failed
the start of the marriage were linked to high to reveal interspouse differences in the strength
average levels of marital satisfaction, and the of these links.
strength of these links did not differ between
spouses. Overall, initial levels of psychological
Differences Between Husbands From Divorced
distress, marriage-specific appraisals, spousal
Versus Stable Couples and Wives From
interactions, and social support failed to predict
Divorced Versus Stable Couples
change in marital satisfaction for either spouse.
In sum, analyses involving the intraspouse links The final issue focused on whether comparisons
failed to reveal interspouse differences in the between the growth curves for spouses who
strength of these links. divorced and those who did not divorce differed
between spouses. For this issue, spouses from
57 couples who were known to have divorced
Differences Between Husbands’ and Wives’
were compared with spouses from 469 couples
Cross-Spouse Links
who did not divorce. Although this latter group
The next issue addressed whether one’s own included spouses from couples who withdrew
levels of psychological distress, marriage- from the study, were dropped from the study, or
specific appraisals, spousal interaction, and completed all assessments, all spouses within
social support assessed at the beginning of the this group came from stable couples. Latent var-
marriage predicted both the average level of iables representing each spouse’s average level
marital satisfaction of one’s spouse and change and each spouse’s change were derived as in the
in the marital satisfaction of one’s spouse equally first set of analyses, and only a dichotomous
well for husbands and for their wives. As already divorce variable (0 ¼ no, 1 ¼ yes) was used as
noted, these cross-spouse links were estimated in a predictor of each latent variable.
latent variable growth curve analyses with con- The unstandardized coefficient for this
trols for intraspouse links. The unstandardized dummy variable is presented for husbands and
coefficients for the cross-spouse links, the asso- for their wives in Table 4 for each of the five
ciated effect-size r’s, the v2 associated with the variables of interest. For average values, posi-
test of differences between spouses in the tive coefficients indicate that spouses from
strength of the effects, and the associated effect- divorced couples had higher values than spouses
size r’s are presented in the right portion of from stable couples, whereas negative coeffi-
Table 3. The entries for husbands refer to the cients indicate that spouses from divorced cou-
link between husbands’ predictors and their ples had lower values than spouses from stable
wives’ marital satisfaction, whereas the entries couples. For change, positive coefficients indi-
for wives refer to the link between wives’ pre- cate that spouses from divorced couples had
dictors and their husbands’ marital satisfaction. steeper increases than spouses from stable cou-
With regard to psychological distress, for ples, whereas negative coefficients indicate that
both husbands and their wives, high levels at spouses from divorced couples had steeper
the start of the marriage were linked to low decreases than spouses from stable couples.
average levels of marital satisfaction for one’s Differences between effects for husbands and
spouse, and the strength of these links did not effects for wives were assessed by comparing
differ between husbands and their wives. With the v2 values from models in which spouse
Gender and Marital Satisfaction 79

TABLE 4. UNSTANDARDIZED COEFFICIENTS FOR HUSBANDS AND WIVES FOR LINK BETWEEN DUMMY VARIABLE
FOR DIVORCED COUPLES (N = 57) VERSUS STABLE COUPLES (N = 469), AND AVERAGE LEVEL OF AND LINEAR CHANGE
FOR DISTRESS, MARRIAGE-SPECIFIC APPRAISALS, SPOUSAL INTERACTION, SOCIAL SUPPORT, AND MARITAL
SATISFACTION VARIABLES, v2 TEST FOR INTERSPOUSE DIFFERENCE, AND EFFECT-SIZE R FOR THE v2 TEST

Average Level Linear Change

Coefficient Husband vs. Wife Coefficient Husband vs. Wife

Variable Husband Wife v2 r Husband Wife v2 r

Distress 0.16** 0.15** 0.01 .00 0.01 0.03 1.54 .05


Appraisals 0.78** 1.12** 3.55 .08 0.18** 0.29** 4.00* .08
Interaction 0.60** 0.55** 0.17 .01 0.14** 0.15** 0.04 .00
Support 1.90** 0.14 4.32* .09 0.16 0.40 2.96 .07
Satisfaction 3.29** 3.85** 1.83 .05 0.70** 0.86** 1.25 .04
*p , .05. **p , .01.

effects were freely estimated with those in cerning differences between husbands and their
which they were constrained to be equal. The wives in the strength of these effects, only one
resulting v2 difference values are presented in difference was found such that the effect for
Table 4, along with the associated effect-size r. marriage-specific appraisals was stronger for
Findings are presented first for average value wives than it was for their husbands. Thus, de-
and then for change. With regard to psychologi- creases in positive marriage-specific appraisals
cal distress, for both husbands and their wives, was linked to eventual divorce more strongly
spouses from divorced couples had higher aver- for wives than it was for their husbands. In sum,
age levels than did spouses from stable couples. because only 2 out of 10 differences between
With regard to marriage-specific appraisals, husbands and their wives were found, analyses
spousal interactions, and marital satisfaction, involving comparisons between spouses from
for both husbands and their wives, spouses from divorced couples and spouses from stable cou-
divorced couples had lower average levels than ples generally did not reveal consistent inter-
did spouses from stable couples. With regard to spouse differences.
social support, the comparison between average
levels was significant only for husbands such
that husbands from divorced couples had lower
DISCUSSION
average levels than did husbands from stable
couples. Concerning differences between hus- The major goal of this study is to reassess the
bands and their wives in the strength of these ef- position—based, in part, on biological and
fects, only one difference was found such that social psychological theories—that there are his
the effect involving social support was signifi- and her versions of every marriage. This reas-
cantly stronger for husbands than it was for sessment was warranted because most of the
their wives. Thus, low average levels of social authors of previous studies who collected lon-
support were linked to eventual divorce for hus- gitudinal data from husbands and wives from
bands but not for wives. the same marriage (see review by Karney &
Regarding the slopes for psychological dis- Bradbury, 1995) did not employ appropriate sta-
tress and social support, comparisons between tistical procedures to assess differences between
spouses from divorced couples and spouses spouses’ scores. In the current study, gender dif-
from stable couples were not significant for ferences in marriage were assessed with longitu-
either husbands or their wives. Regarding the dinal data collected over the first 4 years of
slopes for marriage-specific appraisals, spousal marriage, and couple-level statistical procedures
interactions, and marital satisfaction, decreases were employed that allowed not only for accu-
were stronger for spouses from divorced cou- rate estimates of effects for each spouse, but
ples than they were for spouses from stable also for direct tests of differences between
couples for both husbands and their wives. Con- spouses’ effects.
80 Journal of Marriage and Family

Three issues were addressed with sets of vari- appraisals and marital satisfaction, but not for
ables derived from the individual partner, part- psychological distress, spousal interactions, and
ner interactions, and societal forces levels of social support. Relative to their husbands, wives
Huston’s (2000) social ecological model. Psy- showed stronger declines in both marriage-
chological distress and marriage-specific ap- specific appraisals and marital quality. Once
praisals represented the level of the individual again, however, the effect size associated with
partner, spousal interactions represented the each of these differences was small.
level of partner interactions, and satisfaction The findings that husbands and wives dif-
with social support represented the level of soci- fered in average levels and rates of change on
etal forces. Both marital satisfaction and marital variables relevant to marital functioning is con-
stability were used as outcome variables. The sistent with predictions from both biological
first issue was whether husbands and their wives and social psychological theories. Because all
differed in either average level (i.e., the inter- effects were small in size, however, the findings
cept of growth curves) or rate of linear change run counter to popular accounts of substantial
(i.e., the slope of growth curves) for psychologi- differences in his and her versions of marriage
cal distress, marriage-specific appraisals, spou- (e.g., Gray, 2004). Further, the current study
sal interaction, social support, and marital cannot address which theory is better able to
satisfaction. The second issue was whether hus- account for the obtained differences. Such evi-
bands and their wives differed in either intra- dence could come from studies that specifically
spouse or cross-spouse effects involving the target the identification of processes that medi-
link between the growth curves for marital satis- ate interspouse differences. For example, a lon-
faction and values for psychological distress, gitudinal study that included the variables
marriage-specific appraisals, spousal interaction, studied here, as well as sets of physiological
and social support at the start of the marriage. variables (e.g., Kiecolt-Glaser & Newton, 2001)
The final issue was whether husbands headed and gendered family role variables (e.g., Eagly,
for divorce differed from husbands in a stable 1987), would be ideal to examine whether inter-
marriage as much as wives headed for divorce spouse differences in growth curve parameters
differed from wives in a stable marriage on the were fully accounted for by either biological or
growth curves for psychological distress, mar- social psychological processes, or some combi-
riage-specific appraisals, spousal interaction, nation of both types of processes.
social support, and marital satisfaction.
Do Husbands and Wives Differ
Do Husbands and Wives Differ in Average in Intraspouse Links?
Levels and in Rates of Linear Change?
Given strong practical interest in predicting
Spouses differed in average levels over the first marital outcomes from information gathered at
4 years of marriage for psychological distress, the start of the marriage (Huston et al., 2001),
marriage-specific appraisals, and spousal inter- there was reason to examine interspouse differ-
actions, but not for social support and marital ences in the extent to which levels of psycho-
satisfaction. Consistent with previous evidence logical distress, marriage-specific appraisals,
based on single assessments, relative to their spousal interactions, and social support at the
husbands, wives showed higher levels of psy- beginning of the marriage forecasted both aver-
chological distress (Derogatis, 1994), more pos- age level and the rate of linear change in marital
itive marriage-specific appraisals (Cross & satisfaction over the early years of marriage. In
Madson, 1997), and more positive spousal in- recognition of the interdependence in marital
teractions (Kiecolt-Glaser & Newton, 2001). processes (Bond & Kenny, 2003), both intra-
The effect size for each of these differences was spouse and cross-spouse effects were assessed
small. In line with findings from other prospec- within the same analysis. In contrast to previous
tive longitudinal studies (Huston et al., 2001; studies using growth curve methods (e.g.,
Karney & Bradbury, 1997), there was evidence Karney & Bradbury, 1995; Huston et al., 2001),
of change over the first 4 years of marriage for these intraspouse and cross-spouse effects were
at least one spouse in each of the five sets of assessed with controls for both types of
variables of interest. Spouses differed, however, attrition and correlated errors within spouses
in the rate of linear change for marriage-specific and between spouses.
Gender and Marital Satisfaction 81

Consistent with evidence from previous stud- bles, cross-spouse effects were significant only
ies (e.g., Karney & Bradbury, 1995), intra- for wives, such that low levels of wives’
spouse effects were significant for average marriage-specific appraisals, spousal interac-
levels of marital satisfaction for both spouses. tions, and social support at the start of the mar-
Specifically, high levels of psychological dis- riage were linked to low average levels of their
tress, low levels of marriage-specific appraisals, husbands’ marital satisfaction. The strength of
low levels of spousal interactions, and low the cross-spouse effects obtained for wives,
levels of satisfaction with social support were however, did not differ from the strength of the
each linked to low average levels of marital sat- parallel cross-spouse effects obtained for their
isfaction. Nearly all of these effects were small husbands. Thus, although cross-spouse effects
in size. In contrast to any claims that social psy- occurred more frequently for wives than they
chological processes in particular make marital did for their husbands, the lack of statistically
satisfaction a more rule-governed phenomenon significant interspouse differences in these
for wives than for their husbands, there was no cross-spouse effects provided no support for the
evidence that the correlates of average levels of view derived from social psychological theories
marital satisfaction differed in strength between that wives function as relationship experts
spouses. (e.g., Floyd & Markman, 1983; Sabatelli &
Further, there was little evidence of intra- Bartle-Haring, 2003).
spouse effects regarding how initial levels of
psychological distress, marriage-specific ap-
Do Husbands and Wives Differ in
praisals, spousal interactions, and social support
At-Risk Profiles for Divorce?
forecasted the degree of change in marital satis-
faction for either spouse. In fact, initial levels of The last area in which husbands and their wives
these variables generally failed to show any were compared involved the at-risk profiles for
reliable links to patterns of change in marital divorce. Of particular interest was whether the
satisfaction, making the question of interspouse size of differences between spouses headed for
differences in these effects moot. Karney and divorce and spouses in stable couples in both
Bradbury (1997) reported a similar pattern of average levels and rate of change for the psy-
findings for their self-report measures. In light chological distress, marriage-specific appraisals,
of the dynamic nature of marital processes spousal interactions, social support, and marital
(Huston, 2000), future studies could examine satisfaction variables was the same for husbands
interspouse differences in how patterns of and their wives. Consistent with data from other
change in predictors are linked to patterns of prospective longitudinal studies (Huston et al.,
change in marital satisfaction. 2001; Karney & Bradbury, 1997), for both hus-
bands and their wives, spouses headed for
divorce reported higher average levels of psy-
Do Husbands and Wives Differ
chological distress, lower average levels of and
in Cross-Spouse Links?
stronger declines in marriage-specific apprais-
The evidence regarding cross-spouse effects als, lower average levels of and stronger
showed a similar pattern to that obtained for in- declines in spousal interactions, and lower
traspouse effects in that cross-spouse effects for average levels of and stronger declines in
variables assessed at the start of the marriage marital satisfaction as compared with spouses
occurred more frequently for links to average from stable marriages. For husbands only,
levels of marital satisfaction than for links to lower average levels of social support dis-
change in marital satisfaction. For both hus- criminated between spouses who divorced
bands and their wives, high levels of psycholog- from those who remained married. Of the 10
ical distress in one’s spouse were linked to interspouse differences assessed, only 2 were
one’s own low average marital satisfaction, and significant (each small in size) and might best
the strength of these cross-spouse effects did be attributed to chance. Thus, there was no
not differ between spouses. The major method- consistent evidence that differences between
ological point of the current study is nicely spouses headed for divorce and spouses from
illustrated by cross-spouse effects involving the stable marriages were more pronounced for
marriage-specific appraisal, spousal interaction, wives as relationship experts than they were
and social support variables. For these varia- for their husbands.
82 Journal of Marriage and Family

Limitations that despite slight gender differences in aver-


age levels of variables relevant to marriage, the
The findings from this study must be viewed in
processes that account for variability in both
the context of its limitations. No claim is made
marital satisfaction and marital stability are
that the sample of couples is representative of
similar for husbands and wives. Because couple-
all newlywed couples, the number of couples
level data analyses can now be conducted in
completing all assessments was relatively small
readily available statistical programs, additional
given the size of the sample at the first assess-
work regarding gender differences in marriage
ment, more frequent assessments could have
would do well to employ those analyses so that
been made over the 4 years of study to yield
accurate estimates of effects for husbands and
more reliable evidence of change, nearly all
wives are obtained, statistical tests of differ-
measures were open to biases associated with
ences between effects for husbands and those
self-report methods, other variables from the
for their wives can be conducted, and the effect
domains of interest could have been sampled,
size associated with those tests of interspouse
and other domains (such as sexual behavior)
differences can be reported and used in future
could have been selected. In defense of the
meta-analytic reviews of gender differences in
measures used here, however, they had accept-
marriage.
able psychometric properties, showed gender
differences in average levels similar to those re-
ported in other studies, replicated intraspouse
and cross-spouse linkages found in other stud- NOTE
ies, and reliably discriminated spouses in stable I would like to thank the couples who participated in this
couples from spouses in divorced couples. study.
Nonetheless, it is possible that larger differences
between husbands and wives could have been
obtained with actual observations of husbands REFERENCES
and wives, or with different self-report mea-
Amato, P. R. (2000). The consequences of divorce
sures of different variables.
for adults and children. Journal of Marriage and
the Family, 62, 1269–1287.
Attridge, M., Berscheid, E., & Simpson, J. A. (1995).
CONCLUSION
Predicting relationship stability from both partners
The growth curve findings from this study versus one. Journal of Personality and Social Psy-
nicely supplement the longitudinal evidence chology, 69, 254–268.
summarized by Karney and Bradbury (1995), Baldwin, M. W. (1992). Relational schemas and the
but provide more accurate estimates of effects processing of social information. Psychological
for husbands and wives because of the couple- Bulletin, 112, 461–484.
level statistical analyses employed. In an exten- Bentler, P. M., & Newcomb, M. D. (1978). Longitu-
sion of Karney and Bradbury’s conclusion that dinal study of marital success and failure. Journal
claims derived from biological and social psy- of Consulting and Clinical Psychology, 46, 1053–
chological theories have led to exaggerated gen- 1070.
der effects in marriage, the current findings Bernard, J. (1982). The future of marriage. New
showed that although there were small differen- Haven, CT: Yale University Press.
ces in average levels of growth curves for varia- Bond, C. F., & Kenny, D. A. (2003). The triangle of
bles relevant to marital functioning, there were interpersonal models. Journal of Personality and
no differences between husbands and wives in Social Psychology, 83, 355–366.
the strength of intraspouse and cross-spouse Bramlett, M. D., & Mosher, W. D. (2001). First mar-
predictions of marital satisfaction and in the size riage dissolution, divorce, and remarriage: United
of comparisons that discriminated between States. Advance data from vital and health statis-
spouses headed for divorce and those headed tics (No. 323). Hyattsville, MD: National Center
for stable marriages. Thus, within the confines for Health Statistics.
of the specific set of variables studied here, the Buss, D. M. (1984). Toward a psychology of person-
findings provide no support for the existence of environment (PE) correlation: The role of spouse
his and her versions of marital processes. selection. Journal of Personality and Social Psy-
Rather, the findings support an alternative view chology, 47, 361–377.
Gender and Marital Satisfaction 83

Buss, D. M. (1998). Sexual strategies theory: Histori- Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit
cal origins and current status. Journal of Sex indexes in covariance structure analysis: Conven-
Research, 35, 19–31. tional criteria versus new alternatives. Structural
Cohen, J. (1988). Statistical power analysis for Equation Modeling, 6, 1–55.
the behavioral sciences. New York: Academic Huston, T. L. (2000). The social ecology of marriage
Press. and other intimate unions. Journal of Marriage
Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. and the Family, 62, 298–320.
(2003). Applied multiple regression/correlation Huston, T. L., Caughlin, J. P., Houts, R. M., Smith,
analysis for the behavioral sciences. Mahwah, NJ: S. E., & George, L. J. (2001). The connubial
Erlbaum. crucible: Newlywed years as predictors of
Costa, P. T., Terracciano, A., & McCrae, R. R. marital delight, distress, and divorce. Journal of
(2001). Gender differences in personality traits Personality and Social Psychology, 80,
across cultures: Robust and surprising findings. 237–252.
Journal of Personality and Social Psychology, 81, Karney, B. R., & Bradbury, T. N. (1995). The longi-
322–331. tudinal course of marital quality and stability: A
Cross, S. E., & Madson, L. (1997). Models of the review of theory, method, and research. Psycho-
self: Self-construals and gender. Psychological logical Bulletin, 118, 3–34.
Bulletin, 122, 5–37. Karney, B. R., & Bradbury, T. N. (1997). Neuroti-
Davey, A., Fincham, F. D., Beach, S. R. H., & cism, marital interaction, and the trajectory of mar-
Brody, G. H. (2001). Attributions in marriage: ital satisfaction. Journal of Personality and Social
Examining the entailment model in dyadic context. Psychology, 72, 1075–1092.
Journal of Family Psychology, 15, 721–734. Karney, B. R., Bradbury, T. N., Fincham, F. D., &
Davila, J., Bradbury, T. N., & Fincham, F. D. (1998). Sullivan, K. T. (1994). The role of negative affec-
Negative affectivity as a mediator of the associa- tivity in the association between attributions and
tion between adult attachment and marital satisfac- marital satisfaction. Journal of Personality and
tion. Personal Relationships, 5, 467–484. Social Psychology, 66, 413–424.
Davila, J., Karney, B. R., & Bradbury, T. N. (1999). Kenny, D. A., Mohr, C. D., & Levesque, M. J.
Attachment change processes in the early years of (2001). A social relations variance partitioning of
marriage. Journal of Personality and Social Psy- dyadic behavior. Psychological Bulletin, 127,
chology, 76, 783–802. 128–141.
Derogatis, L. (1994). SCL 90-R: Administration, Kiecolt-Glaser, J. K., & Newton, T. L. (2001).
scoring, and procedures manual. Minneapolis, Marriage and health: His and hers. Psychological
MN: National Computer Systems. Bulletin, 127, 472–503.
Eagly, A. H. (1987). Sex differences in social behav- Kurdek, L. A. (1988). Perceived social support in
ior: A social-role interpretation. Hillsdale, NJ: gays and lesbians in cohabiting relationships. Jour-
Erlbaum. nal of Personality and Social Psychology, 54,
Eagly, A. H., & Wood, W. (1999). The origins of sex 504–509.
differences in human behavior: Evolved disposi- Kurdek, L. A. (1989). Social support and psychologi-
tions versus social roles. American Psychologist, cal distress in first-married and remarried newly-
54, 408–423. wed husbands and wives. Journal of Marriage and
Floyd, F. J., & Markman, H. J. (1983). Observational the Family, 51, 1047–1052.
biases in spouse observation: Toward a cognitive/ Kurdek, L. A. (1992). Dimensionality of the Dyadic
behavioral model of marriage. Journal of Consul- Adjustment Scale: Evidence from heterosexual and
ting and Clinical Psychology, 51, 450–457. homosexual couples. Journal of Family Psychol-
Gottman, J. M., Coan, J., Carrere, S., & Swanson, C. ogy, 6, 22–35.
(1998). Predicting marital happiness and stability Kurdek, L. A. (2003). Methodological issues in
from newlywed interactions. Journal of Marriage growth-curve analyses with married couples. Per-
and the Family, 60, 5–22. sonal Relationships, 10, 235–266.
Gray, J. (2004). Men are from Mars, women are from Kurdek, L. A., & Schmitt, J. P. (1986). Relationship
Venus: The classic guide to understanding the quality of partners in heterosexual married, hetero-
opposite sex. New York: HarperCollins. sexual cohabiting, and gay and lesbian relation-
Hox, J. (2002). Multilevel analysis: Techniques and ships. Journal of Personality and Social
applications. Mahwah, NJ: Erlbaum. Psychology, 51, 711–720.
84 Journal of Marriage and Family

McCrae, R. R. (1993). Moderated analyses of longi- reconciliation. Psychological Assessment, 2, 333–


tudinal personality stability. Journal of Personality 337.
and Social Psychology, 65, 577–585. Sarason, I. G., Sarason, B. R., Shearin, E. N., &
Milardo, R. M., & Helms-Erikson, H. (2000). Net- Pierce, G. R. (1987). A brief measure of social
work overlap and third-party influence in close support: Practical and theoretical implications.
relationships. In S. S. Hendrick & C. Hendrick Journal of Social and Personal Relationships, 4,
(Eds.), Close relationships: A sourcebook (pp. 33– 497–510.
45). Newbury Park, CA: Sage. Sayer, A. G., & Cumsille, P. E. (2001). Second-order
Murray, S. L., Holmes, J. G., & Griffin, D. W. latent growth models. In L. M. Collins & A. G.
(1996). The benefits of positive illusions: Idealiza- Sayer (Eds.), New methods for the analysis of
tion and the construction of satisfaction in close re- change (pp. 177–200). Washington, DC: American
lationships. Journal of Personality and Social Psychological Association.
Psychology, 70, 79–98. Schmitt, D. P. (2002). A meta-analysis of sex differ-
Muthén, L. K., & Muthén, B. O. (2004). Mplus ences in romantic attraction: Do rating contexts
users’ guide: Statistical analysis with latent varia- moderate tactic effectiveness judgments? British
bles. Los Angeles: Authors. Journal of Social Psychology, 41, 387–402.
Newsom, J. T. (2002). A multilevel structural equa- Schulz, M. S., Cowan, P. A., Cowan, C. P., &
tion model for dyadic data. Structural Equation Brennan, R. T. (2004). Coming home upset: Gen-
Modeling, 9, 431–447. der, marital satisfaction, and the daily spillover of
Nolen-Hoeksema, S. (2003). Women who think too workday experience into couple interactions. Jour-
much. New York: Henry Holt and Company. nal of Family Psychology, 18, 250–263.
Rempel, J. K., Holmes, J. G., & Zanna, M. P. (1985). Simpson, J. A., & Gangestad, S. W. (2001). Evolu-
Trust in close relationships. Journal of Personality tion and relationships: A call for integration. Per-
and Social Psychology, 49, 95–112. sonal Relationships, 8, 341–355.
Rosenthal, R. (1994). Parametric measures of effect Spanier, G. B. (1976). Measuring dyadic adjustment:
size. In H. Cooper & L. V. Hedges (Eds.), The New scales for assessing the quality of marriage
handbook of research synthesis (pp. 231–244). and similar dyads. Journal of Marriage and the
New York: Russell Sage Foundation. Family, 38, 15–28.
Rubin, Z. (1970). Measurement of romantic love. Verbeke, G., & Lesaffre, E. (1997). The effect of
Journal of Personality and Social Psychology, 16, misspecifying the random-effects distribution in
265–273. linear mixed models for longitudinal data.
Sabatelli, R. M., & Bartle-Haring, S. (2003). Family- Computational Statistics & Data Analysis, 23,
of-origin experiences and adjustment in married 541–556.
couples. Journal of Marriage and the Family, 65, Watson, D., Hubbard, B., & Wiese, D. (2000). Self-
159–169. other agreement in personality and affectivity:
Sabourin, S., Lussier, Y., Laplante, B., & Wright, J. The role of acquaintanceship, trait visibility and
(1990). Unidimensional and multidimensional assumed similarity. Journal of Personality and
models of dyadic adjustment: A hierarchical Social Psychology, 78, 546–558.

You might also like