Anda di halaman 1dari 13

Journal of Family Psychology 2004, Vol. 18, No.

2, 383395

Copyright 2004 by the American Psychological Association 0893-3200/04/$12.00 DOI: 10.1037/0893-3200.18.2.383

Social Networks, Structural Interdependence, and Marital Quality Over the Transition to Marriage: A Prospective Analysis
Jill N. Kearns and Kenneth E. Leonard
State University of New York at Buffalo
This study examined the relationship between couples network interdependence and marital quality in a sample of 347 couples that were recruited at the time of their rst marriage. Husbands and wives completed separate, self-administered questionnaires at home. Analyses are based on data collected at the time of marriage, at the rst anniversary, and at the second anniversary. Results indicated that after marriage, husbands and wives friend and family networks became increasingly interdependent. Moreover, the interdependence of the friend and family networks at marriage predicted wives marital quality at the rst anniversary, whereas wives marital quality at the rst anniversary predicted the interdependence of the friend network at the second anniversary. No signicant longitudinal relations were observed for husbands.

Despite the fact that most newlyweds report being relatively happy at the time of marriage, more than 60% of rst marriages in the United States end in divorce or permanent separation (Bumpass, 1990). Research indicates that marital quality decreases over the early years of marriage and that one third of all annual divorces occur for couples married 4 years or less (MacDermid, Huston, & McHale, 1990). Hence, it is important to identify the factors that may cause an initially satisfying relationship to deteriorate or remain stable over time. Most of the existing research that investigates marital quality and marital stability has examined how individual attributes (e.g., personality traits) or dyadic attributes (e.g., patterns of interaction between partners) inuence relationship growth or deterioration. As important as this work is, it overlooks how spouses marital quality may be related to external factors, such as the demands of family and friends. As a consequence of marriage, couples have access to and demands from two sets of social ties: their own and their spouses. Couples must determine the degree to which they will maintain separate networks of friends, balance their own and their spouses family, and engage in activities as partners. Although these tasks are often begun prior to marriage and continue throughout marriage, the early years are usually the time when major conicts are rst revealed

and confronted (Leonard & Roberts, 1998). The major goal of this study is to contribute to the literature on the role of social networks in the development, maintenance, and prevention of marital dysfunction over the early years of marriage.

Structural Interdependence and Marital Adjustment


Several researchers (e.g., Milardo, 1986; Surra, 1988) have suggested the concept of structural interdependence as a theoretical basis for linking network structure to marital outcomes. Structural interdependence, which is related to the social exchange theory concept of interdependence (e.g., Kelley & Thibaut, 1978) refers to the properties that reect the form and the degree to which network members are interlinked, such as interconnectedness, density, and network overlap. Structural interdependence refers to the pattern of ties among the members of a couples social network by virtue of the partners dyadic tie. This perspective suggests that as a couple becomes more involved and interdependent, they also develop increasingly interdependent networks. The structural interdependence theory implies that the pattern of ties in the couples social networks inuences the ow of resources that affect the potential for the couples adjustment. In dyadic relationships, it is thought that shared, interconnected network ties are benecial to the development of the relationship because they tend to reinforce the identity of partners as a couple (Jones, 1980; Milardo, 1986). Partners joint exposure to network members raises the probability of being socially visible and perceived, dened, and supported as a couple (Lee, 1979). Members of shared networks tend to be invested in the maintenance of the dyadic relationship and provide partners with sources of relationship support and social comparison (Bott, 1971; Milardo, 1982). In contrast, separate ties, or less connected networks, are thought to provide support for individuation from the dyad and underscore personal rather than couple
383

Jill N. Kearns, Research Institute on Addictions, State University of New York at Buffalo; Kenneth E. Leonard, Research Institute on Addictions, State University of New York at Buffalo, and Department of Psychiatry, State University of New York at Buffalo Medical School. This research was supported by Grant R01-AA09922 from the National Institute on Alcohol Abuse and Alcoholism. Correspondence concerning this article should be addressed to Kenneth E. Leonard, Research Institute on Addictions, 1021 Main Street, Buffalo, NY 14203. E-mail: Leonard@ria .buffalo.edu

384

KEARNS AND LEONARD

identities (Jones, 1980; Milardo, 1982). Additionally, separate ties create opportunities to develop relationships that compete with the conjugal bond (Julien, Chartrand, & Begin, 1999). These processes, while important for relationship development, are also of potential relevance to the maintenance of marital relationships. Consistent with the structural interdependence perspective, research has provided evidence of variations in structural interdependence as a function of time and stage of relationship development. A number of studies have shown that as romantic relationships advance, partners reduce the number of contacts with members in their respective network, and they increase joint social exposure (Johnson & Leslie, 1982; Milardo, 1982; Milardo, Johnson, & Huston, 1983). For example, in a sample of dating couples, Milardo (1982) found that the proportion of network members held in common by both partners (e.g., the joint network) increased for couples advancing in stage and decreased for those regressing. More specically, the percentage of network members common to both members of the dating relationship increased substantially from 30% at casual dating to 58% at exclusive dating to 77% for engaged couples. Other studies have shown that couple members typically become more involved with their partners friends and family members as the relationship progresses. For example, Parks, Stan, and Eggert (1983) found that more romantically involved individuals had greater contact and communication with the partners friends and family. Similarly, Kim and Stiff (1991) reported that partners in more developed relationships communicated with members of each others network with increased frequency and with increased feelings of closeness as compared with partners in less developed relationships. Moreover, there is evidence that social networks can have an impact on relationship dissolution. Using a short-term prospective design, Parks and Adelman (1983) found that romantic partners were less likely to break up when they communicated more often with their partners friends and family. Although there is considerable research demonstrating that couple members social networks become more interdependent over the course of a romantic relationship as the depth of involvement increases, there is little research that examines whether this interdependence impacts the maintenance or decline of marital quality over time. There are a few studies that have shown that spouses who maintain shared networks have more stable and satisfying marriages than those who maintain separate networks (Cotton, Cunningham, & Antill, 1993; Hansen, Fallon, & Novotny, 1991). For example, Hansen et al. (1991) found a strong linear relationship between network overlap and marital adjustment. The greater the reported overlap between spousal networks, the higher was the couples reported marital satisfaction. Cotton et al. (1993) also found that both husbands and wives experienced greater marital satisfaction when they considered more of their partners network members to be their own friends. This research, however, has been limited in several ways. First, most of the studies involved cross-sectional designs, which places serious restrictions on the ndings. Although

greater interdependence between spouses networks is positively associated with marital adjustment (Julien & Markman, 1991), one cannot make any inferences about the temporal ordering of this relationship. It may be that the marital satisfaction facilitates changes in network involvement. Conversely, it may be that preexisting patterns of network involvement may inhibit declines in marital intimacy. A further limitation of this research concerns the heterogeneous nature of the couples with respect to the length of time they had been married. For example, couples in the Cotton et al. (1993) study had been married for an average of 9 years, but the range was 2 to 22 years. As a result, this study included couples who were in the early years of marriage as well as couples with long-term stable marriages. The social network and marital dynamics are likely to be quite different for couples making the transition to marriage than it is for couples who have successfully maintained their relationship for 20 years. From a developmental perspective, it is of critical importance to examine couples as they make their transition to marriage and to follow them through their marriage. A third limitation is that most of the studies that have examined network interdependence and marital adjustment have been restricted to global assessments of networks without specically identifying the particular members. However, all network members are not equal. As Hoyt and Babchuk (1983) pointed out, the family network differs from the friend network in that it is ascribed, more or less permanent, and generally associated with a sense of obligation. Friendship networks, in contrast, are characterized by voluntary involvement, affective bonds, and homophily. In addition, research shows that kin and non-kin differ substantially in the amount of emotional support, services, and companionship that they offer to a couple (Wellman & Wortley, 1989). Therefore, if different categories of network members provide different types of support and fulll different roles, it is possible that they will also have distinctly different inuences on the marital relationship.

Purposes of This Investigation


The purpose of this study was twofold. First, we examined how network interdependence changes over the marital transition and into the second year of marriage. We hypothesized that, after marriage, couple members networks would become increasingly interdependent as they begin to form joint networks comprised of network members that are mutually shared by both partners. Second, we examined how network interdependence inuences the subsequent marital quality of husbands and wives. We hypothesized that husbands and wives with highly interdependent networks would report higher levels of subsequent marital quality than husbands and wives with less interdependent networks. The current study addresses limitations of previous studies by using a longitudinal design examining the networks of newly married couples at three occasions: when applying for a marriage license, at the rst anniversary, and at the second year anniversary. In addition, all analyses are

SOCIAL NETWORKS AND MARITAL QUALITY

385

conducted separately for the friend and family networks, as these two sectors may have distinctly different inuences on the marital relationship.

Method Participants
Couples (N 347) were drawn from the Adult Development Study, a four-wave longitudinal study of newlyweds designed to examine changes in alcohol use as it relates to other developmental changes. Couples who indicated that the marriage was the rst marriage for both husband and wife and who were both 18 years of age or older, English-speaking, and literate were eligible for the study. The current analyses are based on self-report data obtained from couple members at the time of marriage (Wave 1) and at their rst (Wave 2) and second (Wave 3) anniversaries. Wave 4 data collection is in progress. This study sample consists of couples that had completed the rst three assessments at the time of the analyses. At the time of marriage, average ages (and standard deviations) of the 347 husbands and wives were 29.0 (6.2) years and 27.0 (5.6) years, respectively. The majority of husbands (65%) and wives (67%) were European American. A relatively large percentage of the sample was African American (husbands, 28%; wives, 27%). There were very small percentages of Hispanic, Asian, and Native American couples. Approximately 8% of the husbands and 5% of the wives did not graduate from high school; 25% of husbands and 25% of wives did not receive any education beyond high school. About 39% of husbands and 40% of wives were college graduates. At the time of marriage, 33% of the husbands and 38% of the wives were parents. Almost 70% of the couples were living together before marriage, but many of these couples had cohabitated for only a short period before marriage (e.g., 1 6 months, 15%; 712 months, 15%; 1318 months, 8%; 19 24 months, 9%; 24 months, 24%).

Procedures
Over a 3-year period spanning 1996 through 1999, couples were recruited for a brief (5 to 10 minutes), paid ($10) interview at the Buffalo City Hall after they had applied for a marriage license. The interview was conducted by a trained research assistant and focused on basic sociodemographic factors, family and relationship factors, and drinking characteristics. For couples (n 62) who were interested but unable to be interviewed at city hall because of time constraints, a phone interview was conducted at a later time. Less than 8% of the rst-time marriage applicants who were approached declined to participate. In total, 970 couples were interviewed. After the interview, couples were asked to participate in the longitudinal study, which was described to them as concerning many aspects of adult life, including attitudes and personality, stress and coping, social relationships, marital satisfaction, and health-related behaviors. Husbands and wives who agreed to participate were each given a main questionnaire booklet, a social network booklet, and a separate postage-paid envelope to return the booklets. They were asked to complete the materials privately within 2 weeks and to not discuss the contents until both had returned the packets. A schedule of reminder calls was made, and letters were sent over the 3 months following initial contact to minimize nonresponse. The questionnaires took 23 hr to complete, and each spouse was paid $40 for his or her participation. Only 7% (n 70) of the eligible couples declined to participate in

the longitudinal study. There were no signicant differences in most sociodemographic attributes between those who refused and those who agreed to participate. Women in couples who agreed were, however, more likely to have children than those in couples who refused (p .01). Also, couples who agreed to participate had lower incomes (p .01) than couples who refused. Complete Wave 1 data were collected for both husband and wife for 71% (n 626) of the 886 eligible couples recruited into the longitudinal study (14 of the 900 couples who agreed to participate in the study did not marry). Couples were considered to have completed the assessment if valid data were received from both partners on both the main questionnaire and the social network questionnaire. When compared with regard to the data collected in the initial screening interview, there were several signicant sociodemographic differences between couples who completed the questionnaires at the rst wave and those who did not. Husbands and wives who completed the questionnaires were more likely to be European American: husbands, 2(1, N 886) 3.35, p .05; wives, 2(1, N 886) 4.33, p .05. In addition, husbands in couples who did not complete the questionnaires were more likely to have a higher education level, 2(1, N 886) 6.16, p .05. A greater percentage of the respondent than nonrespondent couples were also living together prior to marriage (70% vs. 62%, respectively, p .05). Around the time of the couples rst and second anniversaries, couple members were mailed questionnaire packets similar to those they had completed at the time of marriage. Procedures were identical to those used at the rst assessment. Complete Wave 2 data were obtained from both partners for 85% (n 532) of the couples in which both spouses participated at Wave 1. There were a few signicant sociodemographic differences between couples who completed both waves and those who did not. Husbands and wives who completed both assessments were also more likely to be European American: husbands, 2(1, N 627) 4.82, p .05; wives, 2(1, N 627) 3.32, p .05, and to not have children: husbands, 2(1, N 627) 4.36, p .05; wives, 2(1, N 627) 9.31, p .01, than were husbands and wives who only completed the Wave 1 assessment. Additionally, wives who completed both assessments were more likely to have a higher level of education, 2(1, N 627) 13.15, p .01, than were those who only completed the Wave 1 assessment. A comparison of the couples who dropped out of the study by Wave 2 with those couples who completed both assessments indicated that attrition was also related to husbands marital quality as well as some social network characteristics for both husbands and wives. Husbands who completed both assessments reported higher levels on the Marital Adjustment Test, t(622) 2.47, p .05, and the Personal Assessment of Intimacy in Relationships Scale, t(624) 2.27, p .05, than those who only completed the Wave 1 assessment. With regard to the social network variables, husbands and wives who completed both assessments were more likely to include their partner in their family interactions: husbands, t(550) 2.24, p .05; wives, t(586) 3.11, p .01, than were husbands and wives who only completed the Wave 1 assessment. In addition, wives who completed both assessments were more likely to include their partner in their interactions with friends, t(592) 2.58, p .05. Complete Wave 3 data were obtained from both partners for 88% (n 472) of the couples in which both spouses participated at Wave 2. At the time of these analyses, data had been coded for 394 of these 472 complete couples. Of these 394 coded couples, complete Wave 3 data were obtained from both partners for 88% (n 347) of the couples. A comparison of the couples who dropped out of the study by Wave 3 with those couples who

386

KEARNS AND LEONARD closeness or intimacy of relationship), social support (e.g., whether member provides emotional support), and alcohol-related variables (e.g., general past year drinking pattern). The proportion of network overlap between husbands and wives was computed by taking the proportion of partners common names to the sum of their respective network sizes minus the size of the common network. To assess the average percentage of time that the respondent reported that his or her spouse was included in his or her network interactions, participants reported the percentage of time in the past year that their partner was also present when they got together with each person in their network (1 none of the time; 2 25% of the time; 3 50% of the time; 4 75% of the time; 5 100% of the time; 6 did not get together with person in the past year). Scores were averaged across network members. If participants reported that they did not get together with the person in the past year, their score was coded as missing/ not applicable. All variables were computed separately for the friend and family network. Friends were dened as close friends, friends, and acquaintances. The family network consisted of parents, siblings, extended family, and in-laws. Marital quality. Three instruments were used to assess the marital quality construct: the Marital Adjustment Test (MAT; Locke & Wallace, 1959), the Multidimensional Satisfaction Scale (MDS; Roberts & Rothbard, 1999), and four subscales of the Personal Assessment of Intimacy in Relationships Scale (PAIR; Schaefer & Olson, 1981): Emotional, Social, Sexual, and Recreational Intimacy. The MAT, a widely used measure of overall relationship functioning, focuses on global happiness and differences in the relationship. Higher MAT scores (range 2158) represent greater relationship quality. In the current study, Cronbachs alpha for husbands and wives ranged from .78 to .91 across the three assessments. The MDS assesses satisfaction with 11 functional aspects of relationships: social pleasure, division of labor, problem solving, sexual intimacy, emotional security, emotional closeness, balance of power, love/acceptance, companionship, personal growth, and expression of affection. Respondents rated their satisfaction with each aspect on a 9-item scale ranging from 1 (not at all) to 9 (completely). The MDS was highly reliable for both husbands and wives at all three assessments, with alphas ranging from .91 to .96. The PAIR measures the expected versus realized degree of intimacy in romantic relationships in terms of 5 conceptually relevant domains. For the current study, four of the ve domains were assessed with the realized version of this measure. These 24 items assessed the following four factors of perceived intimacy: social, emotional, sexual, and recreational intimacy. Participants indicated how frequently they experienced each event. Response options ranged from 1 (strongly disagree) to 5 (strongly agree). Cronbachs alpha for husbands and wives ranged from .87 to .94 across the three assessments. Covariates. A number of demographic variables were included in the analyses. Spouse-specic covariates were age (in years), race/ethnicity (1 European American, 0 other racial/ ethnic group), parental status at marriage (1 yes, 0 no), and highest level of education (range: 1 less than eighth grade to 8 masters, doctoral, other postgraduate degree). We also included, as a covariate, the length of time the couple lived together (in months) before marriage.

completed all three assessments indicated that husbands and wives in couples who did not respond at Wave 3 were more likely to have had children prior to marriage: husbands, 2(1, N 394) 4.05, p .05; wives, 2(1, N 394) 6.96, p .01, and were more likely to be non-White: husbands, 2(1, N 394) 7.60, p .01; wives, 2(1, N 394) 4.65, p .05, than those who participated. In addition, wives in couples who did not participate at the third anniversary included their husbands in fewer of their family interactions, t(362) 2.38, p .05, than wives who participated. Husbands who did not participate at the third anniversary also reported higher levels on the Multidimensional Satisfaction Scale, t(392) 2.09, p .05, than those who only completed the rst two assessments. The participants and nonparticipants did not signicantly differ with regard to any other sociodemographic characteristics or any other social network or marital quality variables that were of interest in this study. Overall, results of these analyses suggest that there were several variables that were associated with sample attrition. Most notably, it appears that we were more successful at retaining European American couples at all three waves as compared with couples of other races/ethnicities. In addition, couples who lived together prior to marriage were more likely to drop out of the study at Wave 1, whereas couples who had children prior to marriage were more likely to drop out of the study at Waves 2 and 3. With regard to the substantive variables of interest, there was some indication at Wave 2 that we were more successful at retaining more happily married couples as well as couples who were more involved in each others networks. However, with one exception, this was not the case at Wave 3. Moreover, although all of these differences proved to be statistically signicant, most were relatively small and were detected as signicant given the power inherent in our large sample size. For example, whereas participants and Wave 2 dropouts differed with respect to Wave 1 marital satisfaction, the effect size of these differences was approximately .25 to .28. According to Cohen (1988) this is slightly higher than a small effect size (d .20) and corresponds to a correlation value of approximately .12 to .14. Therefore, we believe that the current ndings are unlikely to be biased by sample attrition.

Measures
For each couple member, data on network interdependence were obtained in the self-administered questionnaires. Demographic data were collected during the initial brief interview at the city hall. All variables were scored so that higher scores represent higher levels of the construct. Network interdependence. Indicators of the network interdependence latent construct were network overlap, the average percentage of time that husbands included wives in their network interactions, and the average percentage of time that wives included husbands in their network interactions. At each wave, both husbands and wives provided a list of individuals in their respective social networks in the past year. Four broad elicitation, prompts were used to increase the likelihood of respondents including all relevant members of their networks. Prompts concerned people providing emotional support, people with whom they socialized or regularly had fun with, people who helped with practical or nancial problems, and any others important to them. Participants provided the rst name and last initial for individuals elicited by these prompts. They also indicated their relationship status with each of the members that were reported (e.g., parent, sibling, friend). For each person listed (up to 23), respondents were asked to report on various attributes of each, including demographics (e.g., age, gender, marital status), psychological qualities (e.g.,

Analyses
Model specication. To examine causal relations, we evaluated and compared a series of models (Figure 1) using structural equation modeling analyses. Separate models were estimated for

SOCIAL NETWORKS AND MARITAL QUALITY

387

Figure 1. Summary of models. Double-headed arrows represent covariations between the disturbance terms, which are not shown to enhance clarity. NI network interdependence; MQ marital quality.

the friend and family networks. Specically, seven models that differed with regard to the specied cross-sectional and longitudinal effects were estimated, resulting in the estimation of seven friend and seven family models. In a baseline stability model, the stability coefcients for husbands and wives marital quality and network interdependence over the rst 3 years of marriage was estimated. In the second model, we incorporated the crosssectional covariations between husbands marital quality and network interdependence and wives marital quality and network interdependence. In the third and fourth models, the prospective effects of network interdependence on husbands and wives marital quality were estimated. The fth and sixth models examined the prospective effects of network interdependence on husbands and wives marital quality. The nal model incorporated all of the prospective paths from the previous models that produced a signicant improvement in the t of the model when compared with the model that included only the cross-sectional associations between marital quality and network interdependence. Analyses are based on 347 couples in which both partners had complete and coded data at all three waves at the time of this analysis. However, because some respondents did not report any family members or friends in their social network at various waves, analyzing characteristics of the family and friend networks would have excluded these individuals. To minimize the loss of participants, we included couples in the friend analyses if at least one partner listed at least one friend at every wave. We included couples in the family analyses if at least one partner had listed at least one family member at every wave. For couples in which one of the members of the pair did not report any friends or family at any one of the waves, their network variables for that wave were estimated from their partners network variables with the expectationmaximization method in the missing values package of SPSS. Therefore, analyses involving the friend network are based on 325 couples, whereas analyses involving the family network are based on 317 couples. In all models, equality constraints were imposed on factor

loadings across time such that each marital quality and network interdependence manifest variable was assumed to load on its corresponding latent construct in the same way at all three waves. Equality constraints were also imposed on the marital quality factor loadings across members such that each of the indicators of marital quality was assumed to load on its latent construct in the same way for husbands and wives. The covariances among the disturbances of husbands and wives marital quality within each wave were estimated. Estimating these covariances accounted for within-time covariation that was not pertinent to our hypotheses. Measurement errors for the same-measured manifest variables were also allowed to correlate across time to account for systematic error variance due to the repeated measures design. Covariates were modeled as exogenous variables that were allowed to correlate among themselves and to predict each of the two substantive constructs (marital quality and network interdependence) at each wave. The four husband-specic covariates were included only in the prediction of husbands marital quality, whereas the four wife-specic covariates were included only in the prediction of the wives marital quality. The one couple-level covariate (length of cohabitation prior to marriage) was included in the prediction of both husbands and wives marital quality at all three waves. Model estimation and evaluation. Model parameters were estimated with the maximum likelihood procedure of the AMOS structural equation modeling program (Arbuckle, 1997). The chisquare statistic was used to evaluate the overall t of the models. A nonsignicant chi-square value indicates good t. However, because trivial differences between the predicted and observed matrices may result in a signicant chi-square when large samples are used, we used three other goodness-of-t indices that are less dependent on sample size: the comparative t index (CFI), the root mean square error of approximation (RMSEA), and the Tucker Lewis index (TLI). Values greater than .90 on the CFI and TLI indexes and values less than .05 on the RMSEA were considered

388

KEARNS AND LEONARD

to indicate an acceptable tting model. Nested models were compared with the chi-square difference test.

Results Changes in Marital Quality and Network Interdependence


To examine changes in marital quality and the amount of time that the spouse was included in network interactions over the early years of marriage, a repeated measures multivariate analysis of variance (MANOVA) using the general linear model (GLM) procedure in SPSS was conducted with couple member (husbandwife) and wave (Wave 1, Wave 2, Wave 3) as the two repeated measures factors and marital quality (MAT, MDS, PAIR) and percentage of time the spouse was present during network interactions (both the husband and wife variables) as the dependent variables. The couple was used as the level of analysis to account for the non-independence of husbands and wives responses (Kenny, 1995). Signicant omnibus multivariate effects were found for wave, F(10, 694) 13.94, p .01, and gender, F(5, 171) 11.10, p .01. Univariate analyses indicated a main effect of wave for all three marital quality variables: MAT, F(2, 350) 26.86, p .001; MDS, F(2, 350) 39.79, p .01; and PAIR, F(2, 350) 41.54, p .01. Pairwise comparisons (Sidak adjustment) indicated that marital quality, as assessed by all three indicators, decreased substantially across all three waves. There was also a main effect of gender for the PAIR, F(1, 175) 10.94, p .01, with husbands reporting lower levels than wives at all three

waves. There was also a signicant main effect of wave for both the percentage of time the spouse was included in interactions with the family, F(2, 350) 13.36, p .01, and the percentage of time the spouse was included in interactions with friends, F(2, 350) 4.43, p .05. Pairwise comparisons (Sidak adjustment) indicated that the percentage of time that spouses were included with both family and friends increased substantially from Wave 1 to Wave 2. However, there was a slight decrease in the amount of time that the partner was included in family interactions from Wave 2 to Wave 3. This decrease, however, was not signicant. The Gender Wave interaction was not significant. Means for all variables across all three waves are presented in Table 1. Because the network overlap variables are dyadic measures, we conducted a separate repeated measures MANOVA to examine change in these variables. Wave (Wave 1, Wave 2, Wave 3) was entered as the repeated measures variable, and the two network overlap variables (friend overlap and family overlap) constituted the dependent variables. Signicant omnibus effects were found for time, F(4, 776) 11.67, p .01. Univariate analyses indicated signicant effects of time for both the proportion of overlapping friends, F(2, 388) 14.38, p .01, and the proportion of overlapping family members, F(2, 388) 12.95, p .01. Pairwise comparisons (Sidak adjustment) indicated that the percentage of both overlapping friends and family members increased substantially from Wave 1 (friend, 16%; family, 24%) to Wave 2 (friend, 23%; family, 35%). However, there were no signicant changes from

Table 1 Marital Quality and Network Interdependence Over the Early Marital Years
At marriage Variable Husbands MDS MAT PAIR Wives MDS MAT PAIR Friend network Overlap Spouse present Spouse present Family Network Overlap Spouse present Spouse present M (SD) Marital quality 7.27 120.44 3.96 7.35 123.30 4.07 (0.09)a (1.38)a (0.04)a (0.08)a (1.41)a (0.04)a 6.96 113.96 3.82 6.93 116.76 3.95 (0.11)b (1.84)b (0.04)b (0.12)b (1.73)b (0.04)b 6.62 111.21 3.71 6.54 111.99 3.79 (0.12)c (2.04)b (0.05)c (0.14)c (2.20)c (0.05)c First anniversary M (SD) Second anniversary M (SD)

Network interdependence 0.16 2.82 2.97 0.24 3.66 3.38 (0.01)a (0.10)a (0.10) (0.02)a (0.09)a (0.09)a 0.23 3.16 3.15 0.35 4.14 3.67 (0.02)b (0.10)b (0.10) (0.02)b (0.07)b (0.08)b 0.23 3.08 2.97 0.33 3.92 3.60 (0.02)b (0.10)b (0.11) (0.02)b (0.09)c (0.09)b

(husband) (wife) (husband) (wife)

Note. N 347. Within rows, values with different subscripts differed at p .05 on simple effects tests. Values with no subscripts did not differ from each other. MDS Multidimensional Satisfaction Scale; MAT Marital Adjustment Test; PAIR Personal Assessment of Intimacy in Relationships Scale.

SOCIAL NETWORKS AND MARITAL QUALITY

389

Table 2 Standardized Factor Loadings


At marriage Variable Husbands MDS MAT PAIR Wives MDS MAT PAIR Overlap Spouse present (husband) Spouse present (wife) Friend Family Marital quality .81 .82 .77 .82 .84 .77 .67 .63 .66 .80 .81 .75 .83 .84 .78 .48 .66 .48 .87 .86 .85 .89 .90 .85 .65 .68 .72 .88 .86 .84 .90 .91 .86 .42 .68 .47 .89 .87 .89 .93 .91 .89 .66 .70 .70 .89 .87 .88 .93 .91 .89 .51 .72 .54 First anniversary Friend Family Second anniversary Friend Family

Network interdependence

Note. The unstandardized factor loadings were constrained to be equal across time. All factor loadings are signicant at p .01. MDS Multidimensional Satisfaction Scale; MAT Marital Adjustment Test; PAIR Personal Assessment of Intimacy in Relationships Scale.

Wave 2 to Wave 3. Means across all three waves are presented in Table 1.

family: r .70, p .01) and second anniversaries (friend: r .60, p .01; family: r .58, p .01).

Stability Model
The stability model was initially estimated including all covariate relations as detailed above. The parameters were then reestimated in a more parsimonious model in which the nonsignicant covariate relations with the substantive variables (marital quality and network interdependence) and the nonsignicant covariate intercorrelations (within and between partners) were excluded. Fit statistics, stability coefcients, and contemporaneous relations in the initial stability model were comparable in magnitude and direction to those in the more parsimonious model for both friends, 2 (488, N 325) 913.89, p .00, CFI .95, TLI .93, RMSEA .05, and family, 2(488, N 317) 921.34, p .00, CFI .94, TLI .93, RMSEA .06. In this nal trimmed stability model, all of the hypothesized factor loadings for the marital quality and network interdependence constructs were substantial and highly signicant for both the friend and family models (see Table 2), and the model t the data reasonably well for both friends and family (Table 3).1 The stability coefcients for each of the three substantive variables measured across time (husband marital quality, wife marital quality, network interdependence) were positive and highly signicant for both the friend and family models. It is worth noting that for marital quality and family network involvement, the stability was lower from Wave 1 to Wave 2 than it was from Wave 2 to Wave 3. At marriage, husbands and wives marital quality was positively correlated in both the friend (r .64, p .01) and family (r .62, p .01) models. Controlling for stability effects, residuals for husbands and wives marital quality remained signicantly associated at the rst (friend: r .69, p .01;

Cross-Sectional Relations Between Network Interdependence and Marital Quality


The addition of the cross-sectional covariations between husbands and wives marital quality and network interdependence provided a signicantly better t to the data than did the initial stability model that did not specify these cross-sectional paths. This was the case for both the friend 2 model, (6, N 325) 64.13, p .01, and the family 2 model, (12, N 317) 112.11, p .01. This indicated signicant cross-sectional associations between marital
Because equality constraints were imposed on the marital quality factor loadings across members such that each of the indicators of marital quality was assumed to load on its latent construct in the same way for husbands and wives, we used a chi-square difference test to compare this stability model with a model in which the marital quality factor loading was free to vary across members. An overall chi-square difference test was not signicant for either the friend model, 2(2, N 325) 1.94, ns, 2 or the family model, (2, N 317) 1.00, ns, supporting the assumption that the indicators of marital quality loaded on the latent construct in the same way for husbands and wives. Because equality constraints were also imposed for the corresponding factor loadings across time, we also tested a model in which the factor loadings were freely estimated at each wave. Although the chisquare difference test revealed that the overall chi-square differ2 ence was signicant for both the friend model, (12, N 325) 49.0, p .01, and the family model, 2(12, N 317) 53.64, p .01, examination of the freely estimated factor loadings revealed that the difference between the factor loadings was small. Therefore, we retained our original model because it ensured that the latent variables were invariant across time. This approach is more conservative and aids in the interpretation of the results.
1

390

KEARNS AND LEONARD

Table 3 Goodness-of-Fit Summary and Model Comparisons


Model 1. 2. 3. 4. 5. 6. 7. 1. 2. 3. 4. 5. 6. 7. Initial/stability Cross-sectional NI 3 Husband MQ NI 3 Wife MQ Husband MQ 3 NI Wife MQ 3 NI Final model Initial/stability Cross-sectional NI 3 Husband MQ NI 3 Wife MQ Husband MQ 3 NI Wife MQ 3 NI Final model
2

df 454 448 446 446 446 446 444 457 451 449 449 449 449 445

TLI .94 .95 .95 .95 .96 .95 .95 .93 .94 .94 .94 .94 .94 .95

CFI .95 .95 .95 .96 .95 .96 .96 .96 .95 .95 .95 .95 .95 .95

RMSEA .05 .05 .05 .05 .05 .05 .05 .06 .05 .05 .05 .05 .05 .05

df

Friend network 855.09 790.96 788.77 776.73 785.70 784.29 769.84 895.89 783.78 780.07 772.18 775.92 776.75 763.45 64.13 2.19 14.23 5.26 6.67 21.12 6 2 2 2 2 4 .01 ns .01 ns .05 .01

Family network 112.11 3.71 11.60 7.87 7.03 20.34 6 2 2 2 2 6 .01 ns .01 .05 .05 .01

Note. N for all friend models 325; N for all family models 317. TLI TuckerLewis index; CFI RMSEA root mean square error of approximation. NI network interdependence; MQ marital quality.

comparative t index;

quality and network interdependence. At each wave, husbands and wives marital quality was highly correlated with the interdependence of the friend and family network. Husbands and wives who reported highly interdependent friend and family networks reported higher levels of marital quality at every assessment. The only exception was that wives marital quality was not signicantly correlated with the interdependence of the family network at marriage. Because the addition of these covariations resulted in a signicantly better tting model, this model was used as the baseline model to which all subsequent models were compared.

Marital Quality Inuences on Network Interdependence


Next, we examined the prospective effects of husbands and wives marital quality on the interdependence of their friend and family networks. First we examined the friend and family models where we specied cross-lagged effects from husbands marital quality to network interdependence. The friend model did not t the data signicantly better than the cross-sectional model, providing no support for the longitudinal effect of husbands marital quality on the interdependence of the friend network. For the family model, however, the addition of the longitudinal paths provided a signicant improvement in the model t when compared with the cross-sectional model (Table 3). Husbands marital quality at marriage predicted the interdependence of the family network at the rst anniversary ( .16, p .05). However, husbands marital quality at the rst anniversary did not predict the interdependence of the family network at the second anniversary. When the longitudinal effects of wives marital quality on network interdependence were estimated, the chi-square difference test was signicant for both the friend and family models (Table 3). Wives marital quality at marriage predicted the interdependence of both the family, .12, p .10, and friend, .16, p .01, networks 1 year after marriage. Although wives marital quality at the rst anniversary predicted the interdependence of the family network at the second anniversary ( .14, p .10), it did not predict the interdependence of the friend network.

Network Inuences on Marital Quality


Next, we examined the prospective effects of network interdependence on marital quality. First, we examined the friend and family models where we specied cross-lagged effects from the network interdependence construct to husbands marital quality. These models did not t the data signicantly better than the cross-sectional models (Table 3), indicating that there were not signicant longitudinal relations between the interdependence of the friend and family networks at one time and husbands marital quality at a subsequent time. When the longitudinal effects of network interdependence on wives marital quality were estimated, the chisquare difference test was signicant for both the friend and family models (Table 3). The interdependence of the friend .16, p .01) and family ( .16, p .01) networks ( at marriage predicted wives marital quality 1 year after marriage. However, the interdependence of the friend and family networks at the rst anniversary did not predict wives marital quality at the second anniversary.

Final Models
When all of the paths from the signicant friend models were incorporated into a nal model, the model t the data well, 2(444, N 325) 769.84, p .00, CFI .96,

SOCIAL NETWORKS AND MARITAL QUALITY

391

TLI .95, RMSEA .05. However, when all paths were simultaneously included in one model, only two signicant paths emerged. The interdependence of husbands and wives friend networks at marriage predicted wives marital quality 1 year after marriage ( .16, p .01). However, a reversal of this effect was seen from Waves 2 to 3. Wives marital quality at the 1 year anniversary predicted the interdependence of husbands and wives friend networks at the second year anniversary ( .17, p .01). There were a number of signicant covariate relations that emerged for the friend model (Table 4). Husbands and wives who lived together prior to marriage reported lower levels of marital quality at the time of marriage. In addition, European American wives reported better marital quality and greater network interdependence at the time of marriage. At the rst anniversary, European American husbands and wives reported greater network interdependence. Husbands who did not have children prior to marriage and who had more education also reported greater network interdependence at the rst anniversary, whereas wives who had children prior to marriage reported lower levels of marital quality at the rst anniversary. There were no signicant covariate effects at the second anniversary for the friend model. The nal family model, which incorporated all of the paths from the signicant family models, t the data well, 2 (444, N 317) 763.45, p .00, CFI .95, TLI .95, RMSEA .05. However, when all of all of the paths were simultaneously included in one model, only one signicant path emerged. The interdependence of husbands and wives family networks at marriage predicted wives marital quality 1 year after marriage ( .16, p .01). There were a number of signicant covariate relations that emerged for the family model (Table 4). Husbands and wives who lived together prior to marriage reported lower levels of marital quality and greater network interdependence at the time of marriage. In addition, European Amer-

ican wives reported better marital quality and greater network interdependence at the time of marriage, and older husbands reported greater network interdependence at the time of marriage. At the rst anniversary, wives who had children prior to marriage reported lower levels of marital quality, whereas husbands with more education reported greater network interdependence. At the second anniversary, wives who had children prior to marriage and husbands who had less education reported greater network interdependence. The standardized, covariate-adjusted path estimates for the nal friend and family models are shown in Figures 2 and 3, respectively. Note that for simplicity and ease of reading the gures, only the latent variables are depicted, and double-headed arrows represent correlations between the disturbance terms. In addition, the covariate relations are not shown in these gures, but the signicant covariate relations are detailed in Table 4.

Discussion
Previous research has examined how individual characteristics or couples interaction patterns inuence relationship growth or deterioration. However, this approach fosters the view that dyadic relationships are isolated from the demands of family and friends. The current ndings, however, highlight the importance of the social environment in understanding relational development. Specically, we have focused on the relationship between marital adjustment and the emergence of a joint, interdependent network of mutual friends and family members. It was hypothesized that, after marriage, pair members would report greater network interdependence. Moreover, we predicted that greater interdependence in husbands and wives networks would predict husbands and wives future marital quality. Consistent with these hypotheses, our results suggest that the social networks of men and women are clearly reshaped

Table 4 Signicant Covariate Relations With Latent Variables for Final Friend and Family Model
Latent factor Husbands marital quality Wave 1 Wives marital quality Wave 1 Wave 2 Network interdependence Wave 1 Wave 2 Friend model Time living together ( .22**) Wife race (.10*) Time living together ( .19**) Wife parental status ( .17**) Wife race (.46**) Husband race ( .22*) Husband education (.23**) Husband parental status ( .26**) Wife race (.22*) Family model Time living together ( .22**) Wife race (.12*) Time living together ( .16**) Wife parental status ( .13**) Husband age ( .32**) Wife race (.50**) Time living together (.25**) Husband education (.32**)

Wave 3

Husband education ( .17*) Wife parental status ( .26**)

Note. Time living together refers to duration of cohabitation (in months) prior to marriage. *p .05. **p .01.

392

KEARNS AND LEONARD

Figure 2. Covariate-adjusted substantive parameter estimates for friend models. Double-headed arrows represent covariations between the disturbance terms. Covariate relations are not shown for clarity but are available from the authors on request. NI network interdependence; MQ marital quality. *p .05; **p .01.

by marriage. After marriage, couples networks became more interdependent, as indicated by increased overlap between husbands and wives friend and family networks as well as increases in the percentage of time that the spouse was included in interactions with friends and family. The major period of change in network interdependence, however, seems to occur from premarriage to the rst year of marriage, whereas the interdependence of couples networks appears to become relatively stable at that point. Our results also indicate substantial cross-sectional associations between marital quality and network interdependence for both husbands and wives. For both husbands and wives, these relationships between marital quality and network interdependence were relatively weak prior to marriage. In fact, for wives, the association between marital quality and family network interdependence was not signif-

icant at the time of marriage. However, at the rst and second anniversaries, there were substantial cross-sectional associations in both the friend and family models. Husbands and wives who indicated having highly interdependent friend and family networks clearly reported better marital adjustment at each of the assessments. Moreover, longitudinal analyses indicated that network interdependence is important in understanding wives relational stability over the early years of marriage. Specically, results showed that the interdependence of the husbands and wives friend and family networks at marriage predicted wives marital quality at the rst anniversary. Neither the interdependence of the friend network nor that of the family network at the rst anniversary predicted wives marital quality at the second anniversary. However, there was evidence that direction of this effect was reversed over the

Figure 3. Covariate-adjusted substantive parameter estimates for family models. Double-headed arrows represent covariations between the disturbance terms. Covariate relations are not shown (for clarity) but are available from the authors on request. NI network interdependence; MQ marital quality. *p .05; **p .01.

SOCIAL NETWORKS AND MARITAL QUALITY

393

second year of marriage. That is, wives marital quality at the rst anniversary longitudinally predicted the interdependence of the friend network, even though there was no evidence of such an inuence in the preceding time period. Although wives marital quality at the rst anniversary did not signicantly predict the interdependence of the family network at the second anniversary, the direction of the effect was consistent with that of the friend network ( .13, p .20). Although most of the existing work that examines social network inuences on marital quality presents evidence of networks inuencing relationships rather than relationships inuencing networks, the current ndings suggest that there may be a reciprocal inuence between network interdependence and marital quality, at least for wives. It is possible that sharing networks plays a more important role in the early stages of couple development and that, over time, interdependence may become less causal in maintaining a couples stability (Julien et al., 1999). When two individuals marry, the couple must form a marital system that realigns the relationships with family and friends to include ones spouse. The failure to do so is likely to hinder the development of marital intimacy. It seems that for most couples, this realignment occurs quickly following the marital transition and becomes relatively stable by the rst anniversary. Moreover, this may be more critical for women. Cross and Madson (1997) have hypothesized that women tend to develop an interdependent self-construal, whereas men develop an independent self-construal. One implication of this is that women may be more likely to expect that, in a relationship, their partner should seek to incorporate her into the fabric of his familial and friendship network. To the extent that this occurs during the time leading up to marriage, it may foster the womans satisfaction. Whereas the longitudinal ndings for wives and the absence of longitudinal associations for husbands are of interest, it is important to keep in mind that there were strong cross-sectional associations between marital quality and network interdependence for both husbands and wives with respect to both friends and family. It is possible that some of these cross-sectional relationships may reect shorter-term longitudinal processes. Whereas the marital quality indices are representative of a current assessment (e.g., rate how happy you are in your relationship right now), the social network indices are representative of past-year involvement (e.g., list the people who provided you with support in the past year). Therefore, the observed cross-sectional associations may represent the relationship between network interdependence over the past year and current marital quality. Although the contemporaneous relations between network interdependence and marital quality are likely to account for part of this relationship, it is possible that this signicant effect is also partially driven by longitudinal processes. It is also possible that the signicant correlations among the residual errors for marital quality and network interdependence reect the fact that both marital quality and network interdependence are changing together and may be similarly inuenced by other variables. For example, these results

could reect the impact of other events, such as getting a new job, moving, or having a child. To the extent that these ndings suggest that network interdependence has an inuence on marital quality, network interdependence could serve to promote relational stability. Having a joint social life may create opportunities to develop and maintain a couple identity, whereas separate social ties may provide support for individuation from the dyad and promote personal rather than couple identities (Julien et al., 1999; Milardo, 1982). It is also possible that communication with the partners friends and family promotes relational stability by reducing uncertainty in the relationship (Parks & Adelman, 1983). Partners usually expect to be involved with each others friends and family. Consequently, the lack of involvement may create uncertainty regarding the partners feelings, whereas increased involvement with the partners network may promote security and reinforce ones value in the relationship. Finally, if one is embedded in the partners friend and family networks, then terminating the marital relationship could endanger a whole series of relationships and activities. Therefore, networks of mutual friends and family may be thought of as an irretrievable investment that is too costly to risk losing through marital disruption (Johnson, 1982; Levinger, 1979; Parks & Adelman, 1983). Whereas the current ndings are important, there are several limitations of the study that should be considered. Arguments can be made that changes in network interdependence are partially a function of the pairs familiarity with one another. Dating or living with a partner for an extended period of time may result in changes in network characteristics prior to marriage. It should be noted, however, that we did include the length of time that the couple lived together before marriage as a covariate in our analyses and that its effect was modest. However, it is possible that whether or not a couple lived together before marriage may act as a moderating factor. Future studies should consider this as well as other possible life changes associated with marriage, such as whether the couple moved, graduated, or had a child. Second, the focus on couples in the early years of marriage denes a specic, developmental transition. We view this as advantageous because it allows us to examine factors predictive of change in marital quality and factors associated with the onset of marital distress. It would be useful, however, for future research to examine how couples networks continue to change beyond the honeymoon phase and how these changes inuence the couples marital quality as they face additional stressors and challenges in their lives. One nal limitation to this study must be noted. Although only 7% of the newlyweds declined to participate, only 72% of the couples who agreed to participate in the study returned both husband and wife questionnaires, for an effective rate of 66%. A signicant number of couples did not return their questionnaires for the initial assessment, and there were some differences between couples who completed the initial assessment and those who did not. Although this is an important limitation, it should be noted that our recruitment and retention rates compare favorably with

394

KEARNS AND LEONARD Johnson, M. P. (1982). Social and cognitive features of the dissolution of commitment to relationships. In S. Duck (Ed.), Personal relationships 4: Dissolving personal relationships (pp. 5173). London: Academic Press. Johnson, M. P., & Leslie, L. (1982). Couple involvement and network structure: A test of the dyadic withdrawal hypothesis. Social Psychology Quarterly, 45, 34 43. Jones, W. L. (1980). Couple network patterns of newcomers in an Australian city. Social Networks, 2, 357370. Julien, D., Chartrand, E., & Begin, J. (1999). Social networks, structural interdependence, and conjugal adjustment in heterosexual, gay, and lesbian couples. Journal of Marriage and the Family, 61, 516 530. Julien, D., & Markman, H. (1991). Social support and social networks as determinants of individual and marital outcomes. Journal of Social and Personal Relationships, 8, 549 568. Kelley, H. H., & Thibaut, J. W. (1978). Interpersonal relations: A theory of interdependence. New York: Wiley. Kenny, D. A. (1995). The effect of nonindependence on signicance testing in dyadic research. Personal Relationships, 2, 6775. Kim, H. J., & Stiff, J. B. (1991). Social networks and the development of close relationships. Human Communication Research, 18, 70 91. Kurdek, L. A. (1998). Developmental changes in marital satisfaction: A 6-year prospective longitudinal study of newlywed couples. In T. N. Bradbury (Ed.), The developmental course of marital dysfunction (pp. 180 204). New York: Cambridge University Press. Lee, G. R. (1979). Effects of social networks on the family. In W. R. Burr, R. Hill, F. I. Nye, & I. L. Reiss (Eds.), Contemporary theories about the family (Vol. 1, pp. 2756). New York: Free Press. Leonard, K. E., & Roberts, L. J. (1998). Marital aggression, quality, and stability in the rst year of marriage: Findings from the Buffalo Newlywed Study. In T. N. Bradbury (Ed.), The developmental course of marital dysfunction (pp. 44 73). Cambridge, England: Cambridge University Press. Levinger, G. (1979). A social exchange view of the dissolution of pair relationships. In R. L. Burgess & T. L. Huston (Eds.), Social exchange in developing relationships (pp. 169 193). New York: Academic Press. Locke, H. J., & Wallace, K. M. (1959). Short marital-adjustment prediction tests: Their reliability and validity. Marriage and Family Living, 21, 251255. MacDermid, S. M., Huston, T. L., & McHale, S. M. (1990). Changes in marriage associated with the transition to parenthood: Individual differences as a function of sex-role attitudes and changes in division of household labor. Journal of Marriage and the Family, 52, 475 486. Milardo, R. M. (1982). Friendship networks in developing relationships: Converging and diverging social environments. Social Psychology Quarterly, 45, 162172. Milardo, R. M. (1986). Personal choice and social constraint in close relationships: Applications of network analysis. In V. J. Derlega & B. A. Winstead (Eds.), Friendship and social interaction (pp. 145166). New York: Springer-Verlag. Milardo, R. M., Johnson, M. P., & Huston, T. L. (1983). Developing close relationships: Changing patterns of interaction between pair members and social networks. Journal of Personality and Social Psychology, 44, 964 976. Parks, M. R., & Adelman, M. B. (1983). Communication networks and the development of romantic relationships: An expansion of

those of other studies of newly married couples (e.g., Huston & Houts, 1998; Veroff, Douvan, Orbuch, & Acitelli, 1998). For example, Veroff et al. (1998) reported a very similar rate of 65%. Similarly, Rusbult, Bissonnette, Arriaga, and Cox (1998) successfully recruited 72% of couples who had applied for a marriage license and had a completion rate for the rst assessment of 75%, for an effective rate of 53%. Huston and Houts (1998) reported an effective rate of 42%. Other researchers who have conducted research on community samples of newlyweds have had even lower rates of involvement (e.g., Kurdek, 1998; 18% recruitment success, 38% Time 1 completion, for an effective rate of 7%). Therefore, although this level of participant loss is an important limitation, it should be noted that our effective rate of involvement is equal to or better than that of other studies of young married couples. Despite these potential limitations, the present study highlights the importance of the social environment in predicting marital stability in newlywed couples. The current study not only replicated the ndings of previous studies on social networks but also expanded the base from which claims can be made regarding the linkage between marital quality and the structure of social networks. Although previous research has provided evidence of variations in network characteristics as a function of time and state of relationship development, these studies are based largely on cross-sectional data, which places restrictions on the ndings, particularly as they relate to developmental questions. Results of the current study speak directly to the question of how the social networks of romantic couples change with the passage of time and with the transition to marriage.

References
Arbuckle, J. (1997). AMOS users guide Version 4.0. Chicago: Small Waters. Bott, E. (1971). Family and the social network (2nd ed.). New York: Free Press. Bumpass, L. L. (1990). Whats happening to the family? Interactions between demographic and institutional change. Demography, 27, 483 498. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum. Cotton, S., Cunningham, J. D., & Antill, J. K. (1993). Network structure, network support and the marital satisfaction of husbands and wives. Australian Journal of Psychology, 45, 176 181. Cross, S. E., & Madson, L. (1997). Models of the self: Selfconstruals and gender. Psychological Bulletin, 122, 537. Hansen, F. J., Fallon, A. E., & Novotny, S. L. (1991). The relationship between social network structure and marital satisfaction in distressed and nondistressed couples: A pilot study. Family Therapy, 18, 101114. Hoyt, D. R., & Babchuk, N. (1983). Adult kinship networks: The selective formation of intimate ties with kin. Social Forces, 62, 84 101. Huston, T. L., & Houts, R. M. (1998). The psychological infrastructure of courtship and marriage: The role of personality and compatibility in romantic relationships. In T. Bradbury (Ed.), The developmental course of marital dysfunction (pp. 114 151). New York: Cambridge University Press.

SOCIAL NETWORKS AND MARITAL QUALITY uncertainty reduction theory. Human Communication Research, 10, 5579. Parks, M. R., Stan, C. M., & Eggert, L. L. (1983). Romantic involvement and social network involvement. Social Psychology Quarterly, 46, 116 131. Roberts, L. J., & Rothbard, J. C. (1999). Development of the Multidimensional Marital Satisfaction Scale. Unpublished manuscript, Research Institute on Addictions. Rusbult, C. E., Bissonnette, V. L., Arriaga, X. B., & Cox, C. L. (1998). Accommodation processes during the early years of marriage. In T. Bradbury (Ed.), The developmental course of marital dysfunction (pp. 74 113). New York: Cambridge University Press. Schaefer, M. T., & Olson, D. H. (1981). Assessing intimacy: The PAIR inventory. Journal of Marital and Family Therapy, 7, 47 60.

395

Surra, C. A. (1988). The inuence of the interactive network on developing relationships. In R. M. Milardo (Ed.), Families and social networks (pp. 48 81). Newbury Park, CA: Sage. Veroff, J., Douvan, E., Orbuch, T. L., & Acitelli, L. K. (1998). Happiness in stable marriages: The early years. In T. N. Bradbury (Ed.), The developmental course of marital dysfunction (pp. 152179). New York: Cambridge University Press. Wellman, B., & Wortley, S. (1989). Brothers keepers: Situating kinship relations in broader networks of social support. Sociological Perspectives, 31, 273306.

Received January 9, 2002 Revision received July 27, 2003 Accepted October 14, 2003

Anda mungkin juga menyukai