Journal of Family Psychology2001, Vol. 15, No. 4, 610-626

Copyright 2001 by the American Psychological Association, Inc.O893-32O0/01/S5.OO DOI: 10.1037//0893-320O.15.4.610

Understanding the Relationship Between Religiosityand Marriage: An Investigation of the Immediate

and Longitudinal Effects of Religiosity onNewlywed Couples

Kieran T. SullivanSanta Clara University

The association between religiosity and marital outcome has been repeatedlydemonstrated, but a complete understanding of this relationship is hindered bylimitations of theory and method. The purpose of the current study was to test 3explanatory models by assessing 2 samples of newlywed couples. Findings indi-cated that religiosity was associated with attitudes toward divorce, commitment,and help seeking cross-sectionally. Longitudinal effects, however, were mostconsistent with a moderating model, wherein religiosity had a positive impact onhusbands' and wives' marital satisfaction for couples with less neurotic husbandsand a negative impact for couples with more neurotic husbands. Overall, the impactof religiosity was weak over the first 4 years of marriage. Theoretical propositionsare offered to guide future research in delineating the types of marriages that maybe most affected by religiosity.

Researchers have been investigating the rela-tionship between religiosity and marriage formore than five decades. Much of this research ispredicated on the idea that couples who aremore religious are more likely than other cou-ples to have happy and stable marriages. Aninitial look at empirical findings seems to gen-erally support this idea. Compared to other cou-ples, couples who attend church more fre-quently have been shown to have higher maritalsatisfaction (Kunz & Albrecht, 1977; J. Wilson

This research was supported by National Instituteof Mental Health National Research Service AwardF31 MH10779 awarded to Kieran Sullivan, NationalInstitute of Mental Health Grant R29 MH48674awarded to Thomas Bradbury, and UCLA AcademicSenate Grant 4-564040-19900-07 also awarded toThomas Bradbury. Lauri Pasch, Thomas Bradbury,and Timothy Urdan are gratefully acknowledged fortheir assistance with various aspects of the prepara-tion of this article.

Correspondence concerning this article should beaddressed to Kieran T. Sullivan, Department of Psy-chology, Santa Clara University, 500 El CaminoReal, Santa Clara, California 95053-0333. Electronicmail may be sent to ksullivan@scu.edu.

& Musick, 1996), are less likely to perpetratefamily violence (Ellison, Bartkowski, & Ander-son, 1999), and are less likely to be divorced(Bahr & Chadwick, 1985; Glenn & Supancic,1984). Couples who score higher in more gen-eral measures of religiosity have also beenshown to be happier (Anthony, 1993;Bugaighis, Schumm, Jurich, & Bollman, 1985;Mahoney et al., 1999; M. R. Wilson & Fils-inger, 1986) and to have more stable marriages(Albrecht & Kunz, 1980; Nye, White, & Fride-res, 1973; White & Booth, 1991). These rela-tionships have emerged even after controllingfor important demographic variables, such asage at marriage (Call & Heaton, 1997; Schrum,1980). In addition, researchers have demon-strated that the relationship between these self-report measures of religiosity and marital satis-faction is not an artifact of social desirability orconventional responding (Filsinger & Wilson,1984; Schumm, Bollman, & Jurich, 1982).

However, some important issues inhibit acomplete understanding of how religiosity af-fects marriage. Convenience samples limit thegeneralizability of some studies: for example,the use of parents of students (e.g., Hunt &King, 1978), psychology classes and church

610

SPECIAL SECTION: RELIGIOSITY AND MARRIAGE 611

congregations known by the authors (Snow &Compton, 1996), and acquaintances of the au-thor (Kaslow & Robison, 1996). Some studieshave also used analytic techniques that limit theinterpretation of findings (e.g., Anthony, 1993;Kunz & Albrecht, 1977). The use of heteroge-neous married couples (i.e., couples married forvarying lengths of time, couples with and with-out children, first marriages and second mar-riages, etc.) makes it difficult to determine howreligiosity might differentially affect variousstages in marriage or different types of mar-riage. Finally, the vast majority of studies relysolely on cross-sectional data, making it impos-sible to determine the nature of the relationshipbetween religiosity and marital functioning. Inone of the few longitudinal studies on the effectof religiosity on marriage, Booth, Johnson,Branaman, and Sica (1995) reported a recipro-cal relationship between religiosity and maritalsatisfaction, such that changes in marital satis-faction predict changes in religiosity over time,calling into question the interpretation of previ-ous correlation studies. They concluded that "ingeneral, the link between religion and maritalquality is both reciprocal and weak" (p. 661).

Second, mixed and sometimes contradictoryfindings regarding the relationship between re-ligiosity and marriage have been reported occa-sionally. In contrast to previous findings thatreligiosity is related to marital satisfaction,Booth et al. (1995) found no relationship be-tween religiosity and future marital satisfaction.Schumm, Obiorah, and Silliman (1989) alsofound no relationship between church atten-dance and marital quality. Findings regardingmarital stability have been mixed. WhereasBooth et al. did find that increases in religiosityslightly decrease the probability of consideringdivorce, Thornes and Collard (1979) found nodifferences in the level of religiosity betweencouples who were still married and couples whohad divorced. This contradicts longitudinalfindings on church attendance that indicate thatattendance is strongly predictive of subsequentlikelihood of divorce (Clydesdale, 1997; Fer-gusson, Horwood, & Shannon, 1984). In a com-prehensive review of all studies published fromthe 1930s to the 1990s (including dissertations),Jenkins (1991) found conflicting evidence forthe propositions high religiosity promotes mar-ital satisfaction and increased church attendanceincreases marital satisfaction. Regarding stabil-

ity, he concluded that there is "moderate" sup-port for the proposition that "high religiositypromotes marital stability" (p. 270).

Perhaps the largest impediment to a morecomplete understanding of how religiosity af-fects marital functioning is that many studieshave been exploratory in nature or empiricallydriven rather than theory driven. It is interestingthat the two most frequently offered rationalesfor studying the effect of religiosity on maritalfunctioning have been (a) the lack of researchinvestigating this relationship and (b) previousempirical findings of a relationship between re-ligiosity and marital functioning. Some re-searchers have offered general theoretical ideasabout how religiosity affects marriage, pointingto the barriers that many religions impose re-garding divorce (Levinger, 1976) and, more re-cently, to the value that many religions share ofkeeping families intact (e.g., Booth et al., 1995;Call & Heaton, 1997). Some authors have pro-posed specific mechanisms that might mediatethe relationship between religiosity and maritaloutcome; however, few have actually testedthese models empirically (see Mahoney et al.,1999, for an important exception). Therefore, animportant next step is a focus on the process bywhich religiosity affects marital functioning.

The purpose of the present study was to clarifythe relationship between religiosity and maritalfunctioning by investigating three potential ex-planatory models. To accomplish this, two sepa-rate studies were conducted, one cross-sectionaland one longitudinal. The use of two studies per-mits replication of cross-sectional findings and acomparison of cross-sectional and longitudinal re-sults. In both studies, participating couples werehomogeneous (newlyweds, married for the firsttime, with no children) and were sampled usingmarriage licenses (Study 1) and newspaper adver-tisements (Study 2). The use of newlyweds mar-ried for the first time provides a clearer under-standing of initial associations between religiosityand marital satisfaction; following them longitu-dinally allows one to begin to understand theprocess by which religiosity affects marital satis-faction and stability.

Study 1

Rationale and Hypotheses

Cross-sectionally, religiosity may affect mar-ital functioning by (a) directly affecting cou-

612 SULLIVAN

pies' marital satisfaction (the direct model) or(b) moderating the relationship between maritalvulnerabilities and marital satisfaction (thecompensation model). Recent evidence for thedirect model is conflictual and sometimes weak;therefore, it was tentatively hypothesized thatreligiosity would not have a direct effect onmarital satisfaction or stability.

The compensation model, in which religios-ity moderates the relationship between maritalvulnerabilities and marital satisfaction, mayhelp explain conflictual cross-sectional findings,assuming that couples' vulnerabilities variedacross samples in previous studies. This modelsuggests that religiosity may compensate forcouples' vulnerabilities and help them to remainrelatively satisfied despite these vulnerabilities.The idea that religiosity serves a compensatoryfunction in marriage has some support. Wallin(1957) found that among young wives who re-ported lower levels of sexual satisfaction, thosewho were highly religious had much higherlevels of marital satisfaction compared to thosewho were less religious. In fact, highly religiouswives with lower sexual satisfaction were justas happy as young wives who were sexuallysatisfied. This finding was replicated later formiddle-aged couples (Wallin & Clark, 1964). Inaddition, higher levels of religiosity were foundto keep wives who reported low rewards in theirrelationships satisfied despite the lack of re-wards (Hansen, 1987). In the current study, twovariables were used to identify vulnerable cou-ples: age and neuroticism. Age at marriage hasbeen one of the most consistent socio-demographic predictors of marital outcome (fora review of divorce predictors, see Karney &Bradbury, 1995), and neuroticism has beendemonstrated to predict both satisfaction andstability, within and between spouses who havebeen married for more than 50 years (Kelly &Conley, 1987).1

Another potential explanation for the conflic-tual findings regarding the direct effect of reli-giosity on marital satisfaction is that the rela-tionship is indirect, affecting other dimensionsof marital quality and functioning, which, overtime, may affect marital satisfaction and stabil-ity. In this model, the indirect effect of religi-osity on variables such as attitudes toward di-vorce and spouses' communication may or maynot have an impact on marital satisfaction at thetime these variables are measured. However,

these other dimensions of marital quality andfunctioning may predict marital satisfaction fur-ther down the line, providing an indirect path-way through which religiosity may affect mar-ital functioning. This model is consistent withthe findings that church attendance predictsmultiple dimensions of commitment to mar-riage (Larson and Goltz, 1989) and that theimpact of religion on one's life predictsspouses' communication skills, such thatspouses who are more affected by religion re-ported greater satisfaction with the patterns ofcommunication in their marriage (Snow &Compton, 1996). These findings point to twopotential domains in which religiosity mightaffect other measures of marital quality: attitu-dinal and behavioral. For the current research,the attitudinal domain was assessed by measur-ing couples' attitudes toward divorce, their levelof commitment to the relationship, and theirreported willingness to seek help in times ofmarital distress. The behavioral domain wasassessed by observing couples' communicationpatterns during an actual discussion in the lab-oratory. Based on initial findings, it was hypoth-esized that religiosity predicts couples' attitudes(i.e., their attitudes toward divorce, their com-mitment to the relationship, and their willing-ness to seek help in times of trouble). Becauseprevious findings in the behavioral domain werebased on spouses' self-reported satisfactionwith their communication patterns, rather thanactual communication behavior, it was unclearwhether religiosity predicts couples' behavior.

The purpose of Study 1 was to test the directand compensation models by analyzing the re-lationships between religiosity, marital quality,and risk variables for future marital problems.The relationship between religiosity and maritalattitudes and behavior were also tested, as apreliminary investigation of an indirect, longi-tudinal model of religiosity and marital func-tioning. A significant effect of religiosity onmarital satisfaction would support the directmodel. A significant interaction effect of religi-

1 Many areas of vulnerability might be compen-sated by religiosity, but age and neuroticism wereselected in this study because they are stable vari-ables that are present at the beginning of marriageand because of the consistent empirical evidence thatthey are among the strongest predictors of futuredissatisfaction and divorce.

SPECIAL SECTION: RELIGIOSITY AND MARRIAGE 613

osity and the risk variables (age and neuroti-cism) might indicate that religiosity reduces theimpact of risk variables on marital satisfactionand thus support the compensation model. Fi-nally, a significant effect of religiosity on otherdimensions of marital quality and functioning(i.e., divorce attitudes, commitment, help seek-ing, and communication behavior) would pro-vide preliminary support for an indirect, longi-tudinal model.

Method

Participants

One hundred seventy-two newly married coupleswere recruited through marriage licenses to partici-pate in a study of newlywed marriage. Marriagelicenses of recently married couples registered in LosAngeles County were screened to identify coupleswho were married for the first time, had been marriedless than 6 months, were between the ages of 18 and35, and had a minimum of 10 years of education.Couples who met the criteria were sent a letter de-scribing the project and requesting that they return apostcard if they were interested in participating. In-terested couples were interviewed by telephone toinsure that they met all inclusion criteria, includingthe additional criteria that they had no children, werenot currently expecting a child, could read and speakEnglish, and were living together. Eligible coupleswere invited to participate in the project, and the first172 who met the screening criteria and kept theirscheduled laboratory appointment were included inthe sample. Approximately 18% of the couples re-ceiving the initial letters returned the postcards (afigure that is comparable to the 18% reported byKurdek, 1991, in a similar study), and approximately56% of those who were interviewed by telephone metcriteria and were invited to participate. Husbandswere on average 27.6 (SD = 3.9) years old, had 15.6(SD = 2.2) years of education, and earned a grossannual income ranging from $21,000 to $30,000.Wives were on average 26.0 (SD = 3.4) years old,had 16.2 (SD = 2.0) years of education, and earneda gross annual income ranging from $11,000 to$20,000. Participants reported their ethnicity as Cau-casian (64%), Asian American (11%), Hispanic(16%), African American (5%), Middle Eastern(2%), and other (2%). Husbands identified them-selves as Protestant (41%), Catholic (31%), Jewish(5%), Mormon (2%), no religion (19%), and other(2%). Wives identified themselves as Protestant(47%), Catholic (26%), Jewish (5%), Mormon (3%),no religion (17%), and other (3%).

Procedure

Eligible couples were scheduled for a laboratorysession in which spouses independently completed aset of questionnaires including a consent form, de-mographic forms, measures of marital quality, and apersonality measure (see the Measures section). Cou-ples were also asked to engage in two 10-minproblem-solving discussions. In these discussions,spouses were asked to work toward a resolution of animportant marital problem. The topics for theproblem-solving discussions were selected indepen-dently by each spouse based on his or her responsesto the Inventory of Marital Problems (Geiss &O'Leary, 1981), a measure of the extent to whichspouses encounter difficulties with 19 commonsources of marital disagreement (e.g., communica-tion, in-laws, finances). The order of the discussionswas random, and the discussions were videotaped forlater coding. The session concluded with a debrief-ing, and participants were paid $75.

Measures

Religiosity. Religiosity was measured using a4-item scale that assessed spouses' religious behaviorand their self-identification as religious persons. Thisscale is a brief measure of religiosity, based on TheReligiosity Measure constructed by Rohrbaugh andJessor (1975), which attempts to capture importantdimensions of religiosity, including ritual, conse-quence, and experience (identified originally byGlock, 1959) as well as an overall rating of religios-ity. The following four questions were used: "Howoften do you attend religious services?" (measured ona 6-point scale ranging from never to more than oncea week; Ms = 2.9 and 3.1 for husbands and wives,respectively); "In general, how important are reli-gious or spiritual beliefs in your day-to-day life?"(measured on a 9-point scale ranging from not at allimportant to very important; Ms = 5.8 and 6.2);"When you do have problems or difficulties in yourwork, family, or personal life, how often do you seekspiritual comfort?" (measured on a 5-point scaleranging from never to almost always; Ms = 2.9 and3.3); and "In general, would you say you are areligious person?" (measured on a 9-point scale rang-ing from definitely no to definitely yes; Ms = 5.6 and5.9). The measure was reliable (coefficient a = .90and .89 for husbands and wives).

Dimensions of marital quality and functioning.Marital satisfaction was measured using the MaritalAdjustment Test (MAT; Locke & Wallace, 1959).The MAT is a widely used measure with high reli-ability demonstrated across many studies (splithalf = .90). Scores range from 2 to 158, with higherscores indicating greater marital satisfaction.

Divorce attitudes were measured using a question-naire based on Veroff (1988b). The questionnaire is a

614 SULLIVAN

9-item scale in which spouses are asked to rate theiragreement with items such as "Except in rare cases,couples should stay married no matter what" on a9-point scale ranging from strongly disagree tostrongly agree. The measure was reliable for hus-bands and wives-(a = .74).

Marital commitment was measured using the ded-ication scale of the Commitment Inventory (CI; Stan-ley & Markman, 1992). This 12-item inventory in-cludes items such as "I want my marriage to staystrong no matter what happens" and "I want to havea strong identity as a couple with my partner." Scoresrange from 12 to 84, with higher scores indicatinggreater commitment. The CI had adequate reliability(as = .63 and .77 for husbands and wives).

Marital help-seeking was measured using a 14-item questionnaire based on Veroff s (1988a) MaritalHelp-Seeking Measure. This questionnaire askedspouses to imagine they encountered serious prob-lems in their marriage and to indicate the steps theywould take to resolve their difficulties by circling yesor no. Examples of items are "I would suggest we seea marriage counselor" and "I would talk to a priest,minister, or other religious person." Scores rangefrom 0 to 14 and reliability estimates were adequatefor husbands and wives (split-half = .61 and .73,respectively).

Behavior. Measures of spouses' negative behav-ior and positive behavior were obtained using behav-ioral coding of the videotaped problem-solving dis-cussions. The problem-solving discussions werecoded using the Specific Affect Coding System(SPAFF; see Gottman & Krokoff, 1989). Trainedgraduate and undergraduate coders were instructed toconsider nonverbal cues, verbal content, voice tone,volume, and speed when coding the speaker's affect.Each 5-s block was classified as either neutral, neg-ative (displays of anger, contempt, whining, sadness,or anxiety), or positive (displays of humor, affection,or interest) for each spouse. Summary codes werecreated to simplify analysis. The total amount ofnegativity displayed in each interaction was calcu-lated by summing all the 5-s intervals coded as anger,contempt, whining, sadness, and anxiety. The totalamount of positivity displayed in each interactionwas calculated by summing all the 5-s intervalscoded as humor, affection, or interest. Intercoderreliability was adequate; the percentage of observedagreement for all codes was .87 for husbands and .84for wives.

Risk variables. Spouses' ages at marriage weremeasured using the demographics questionnaire.Neuroticism was measured using the neuroticismsubscale of the Eysenck Personality Questionnaire(EPQ-N; Eysenck & Eysenck, 1978). Scores rangefrom 0 to 23, with higher scores indicating higherlevels of neuroticism. The EPQ-N was reliable forhusbands and wives (as = .86 and .79).

Analysis

The purpose of Study 1 was to evaluate two mod-els that might account for the cross-sectional relation-ship between religiosity and marital satisfaction andto conduct a preliminary test of a potential longitu-dinal model. First, bivariate correlations were com-puted among all the variables to identify zero-orderrelationships among religiosity, marital quality, be-havior, and marital risk variables. A series of hierar-chical regressions was performed to determinewhether the direct or compensation model was mostconsistent with the data. The direct model was eval-uated by testing whether husbands' and wives' reli-giosity accounted for a significant amount of thevariance of their own marital satisfaction after con-trolling for age at marriage.

The compensation model was evaluated by testingwhether the effect of risk factors (i.e., age at marriageand neuroticism) on husbands' and wives' maritalsatisfaction was moderated by their own level ofreligiosity. To test for moderation, the main effectspredictors (each risk variable and religiosity) wereentered in the first step of each analysis and theinteraction variable (Risk Variable X Religiosity)was entered in the second step. Moderator effects areindicated if the interaction effects are significant aftercontrolling for the main effects predictors (Baron &Kenny, 1986). To reduce multicollinearity among themain effects variables and the interaction terms, thepredictor variables were centered around their meansbefore the product terms were computed (see Aiken& West, 1991, for a description).

Finally, a preliminary test of the indirect modelwas conducted by analyzing whether husbands' andwives' religiosity predicted their attitudes (i.e., di-vorce attitudes, commitment, and help seeking) andtheir behavior (i.e., negativity and positivity duringproblem-solving discussions). Hierarchical regres-sion was used to test within-spouse relationships be-tween these variables and religiosity, after controllingfor marital satisfaction and age at marriage.

Results

Preliminary Analyses

Within-spouse correlations among all thevariables and means and standard deviations areshown in Table 1. Between-spouse correlationsamong all the variables are shown in Table 2.Husbands' and wives' religiosity was not sig-nificantly correlated with their own (rs = .09and - . 1 1 , respectively) or their spouses' (rs =- .06 and .04, respectively) marital satisfaction.This is inconsistent with the direct model ofreligiosity and marital functioning. However,neuroticism was negatively related to marital

SPECIAL SECTION: RELIGIOSITY AND MARRIAGE 615

Table 1Within-Spouse Correlations and Descriptive Statistics for All Variables

All variables

1. Marital satisfaction2. Religiosity _3. Divorce attitudes4. Commitment5. Help-seeking attitudes6. Negative conflict

behavior7. Positive conflict

behavior8. Age9. Neuroticism

HusbandsMSD

WivesMSD

1

- .11- .01

.38**- .03

- .19*

.25**

.01-.34**

126.217.6

130.016.2

2

.09—.47**.19*.17*

- .07

.05- .01- .02

17.28.3

18.48.2

3

- .04.56**—.15*.10

.07

- .05- .03- .05

48.014.4

45.515.1

4

.41**1 .15*

.05—.02

- . 2 1 *

.12- .18*- .15*

75.17.1

76.65.4

5- .06

.31**

.19*

.09—

.03

- .08.09

- .04

20.72.4

21.22.0

6

- . 2 1 *- .09

.10- .04-.05

—

-.30**- .15*

.13

29.424.8

40.028.7

7

.23**

.07-.01

.14-.07

- .18*

—-.04- .15

7.18.2

31.89.3

8

-.07.13

- .03- .19*

.21*

- .08

- .17*—

-.07

27.63.9

26.03.4

9

-.29**- .05- .05- .15*- .10

.13

- .10- .02

—

6.64.9

9.24.8

Note. Correlations for husbands are presented above the diagonal, and correlations for wives are presentedbelow the diagonal.*p<.05. **p<.0l.

satisfaction within (rs = - .29 and -.34) andbetween spouses (rs = - .23 and -.22), whichis consistent with part of the moderating model.Wives' age was not significantly related to mar-ital satisfaction within or between spouses, al-though husbands' age was significantly relatedto wives' satisfaction (r = -.16), with youngerhusbands having less satisfied wives.

Husbands' and wives' religiosity was signif-icantly correlated with their own (rs = .56 and.47, respectively) and their spouses' (rs = .46and .48, respectively) divorce attitudes. Religi-osity was also correlated with commitment

within and between spouses (rs = .15 and .19,respectively) for husbands and within spouse(r = .19) for wives. Finally, religiosity was alsosignificantly related to within-spouse (rs = .31and .17, respectively) and between-spouse (rs =.22 and .21, respectively) help-seeking behav-ior. Thus, it appears that as spouses' level ofreligiosity increases, their divorce attitudes be-come more conservative, their commitmentlevel increases, and the likelihood they wouldseek help in times of trouble increases. Com-mitment, in turn, was related to current maritalsatisfaction for husbands and wives within

Table 2Between-Spouse Correlations for All Variables

Wives' variable

1. Marital satisfaction2. Religiosity3. Divorce attitudes4. Commitment5. Help-seeking attitudes6. Negative conflict

behavior7. Positive conflict

behavior8. Age9. Neuroticism

1

.51**

.04- .05

.29**

.01

-.24**

.20**

.01- .23**

2

- .06.72**.46**.19*.22*

- .09

.05

.00- .08

3

- .04.48**.61**.14.14*

.06

.07- .07- .08

Husbands' variable

4

.27**

.10

.02

.20*- .04

-.07

.16*- .08- .17*

5

- .08.21*.15*.07.15*

- .02

- .02.17*.13*

6

- .15*- .06

.08- .13- .02

.69**

- .16*- .08

.12

7

.30**

.03- .05

.16*- .05

-.26**

.78**- .04- .11

8- .16*

.09

.05

.08*

.13*

- .14

-.11.60**

- .01

9

-.22**- .06- .03- .17*- .16*

.02

.01- .02

.11

* p < . 0 5 . ** />< .01 .

616 SULLIVAN

(rs = .41 and .38) and between (rs = .27 and.29) spouses. However, divorce attitudes andhelp seeking were not related to current maritalsatisfaction. In the behavioral domain, positiveand negative conflict behavior was related tocurrent maritarsatisfaction for husbands (rs =.23 and - . 2 1 , respectively) and wives (rs = .25and —.19). However, unlike in the attitudinaldomain, these behavioral variables were not re-lated to religiosity: Religiosity was not relatedto negative behavior within or between spousesfor husbands (rs = .10 and -.09) or for wives(rs = .07 and -.06), nor was it related topositive behavior within or between spouses forhusbands (rs = .07 and .07) or for wives (rs =.05 and .03).

Regression Results

Results of the hierarchical regression analy-ses used to evaluate the direct model and apotential indirect model are presented in Table3. Religiosity did not account for a significantamount of the variation in marital satisfactionfor husbands (/3 = .10, ns) or for wives (j3 =- . 1 1 , ns) after controlling for age. Religiosity

did, however, account for a significant amountof the variation in divorce attitudes, commit-ment, and marital help seeking after controllingfor marital satisfaction and age. Higher levels ofreligiosity were predictive of more conservativedivorce attitudes for husbands (/3 = — .59, p <.01) and for wives (/3 = - .47, p < .01), ac-counting for 33% and 22% of the variance,respectively. Higher levels of religiosity werealso predictive of higher levels of commitmentfor husbands (/3 = .14, p < .05) and for wives(/3 = .23, p < .01), accounting for an additional2% and 5% of the variance beyond that ac-counted for by marital satisfaction and age. Fi-nally, higher levels of religiosity were predic-tive of a greater willingness to seek help forhusbands (/3 = .30, p < .01) and for wives (/3 =.17, p < .05), accounting for an additional 9%and 3% of the variance beyond that accountedfor by marital satisfaction and age. Religiositydid not account for negative or positive behav-ior during conflict discussions for husbands(/3s = - .06 and .07) or for wives (/3s = .07 and.06).

To test the compensation model, a series of

Table 3Within-Spouse Predictions of Marital Quality and Behavior From Religiosity, Controlling forAge and Marital Satisfaction

Predictor

Step 1Marital satisfactionAge

Step 2Religiosity

F value"A/?2

Adjusted R2

Step 1Marital satisfactionAge

Step 2Religiosity

F value"\R2

Adjusted R2

Maritalsatisfaction"

- .07

.101.38.01.00

.01

- .111.03.01.00

Divorceattitudes Commitment

Husbands' variable

- .04- .04

.59**28.35**

.33**

.32**

Wives'

- . 08- .03

.47**15.82**

.22**

.21**

.39**- .16*

.14*14.74**

.02*

.19**

variable

.38**-.19**

.23**16.84**

.05**

.22**

Help-seekingattitudes

- .05.21**

.30**8.75**

.09**

.12**

- .03.09

.17*2.15

.03*

.02

Negativeconflict

behavior

- . 2 1 * *- .10

- .063.34*

.00

.04

.25**- .04

.074.28**

.01

.05

Positiveconflict

behavior

.21- .15*

.074.69**

.00

.06

- .18*-.20**

- .064.70**

.00

.06a Regressions analyses predicting marital satisfaction from religiosity controlled for age only. b dfs are 2 and169 for Marital satisfaction; dfi are 3 and 168 for all other columns.*/?<.O5. **/><.01.

SPECIAL SECTION: RELIGIOSITY AND MARRIAGE 617

hierarchical regressions were performed to de-termine whether religiosity moderated the effectof risk variables on marital satisfaction (seeTable 4). Age was not a significant predictor ofmarital satisfaction, but neuroticism did predictmarital satisfaction within and between spouses.Religiosity, however, did not emerge as a sig-nificant predictor of marital satisfaction in anyof the analyses. More important, the interactionbetween the risk variables and religiosity wasnot significant, indicating that religiosity wasnot moderating the effect of these risk variableson marital satisfaction.

Discussion

No support was found for the direct model ofreligiosity; neither husbands' nor wives' religi-osity significantly predicted their own or theirspouses' marital satisfaction. However, there issome evidence consistent with an indirect, lon-gitudinal model of religiosity; that is, husbandsand wives who were more religious had lesstolerance for the idea of divorce and a greaterlevel of commitment. They were also morelikely to be willing to seek help in times ofmarital distress. These beliefs and attitudesclearly do not act as mediators of the relation-ship between religiosity and current marital sat-isfaction (as one might expect at the beginningof a marriage) because no direct relationshipwas found between these two variables. How-ever, it is possible that these beliefs and atti-tudes about the importance of staying married,as well as the reported willingness to seek mar-ital help, may affect marital satisfaction andstability over time, providing an indirect path-way through which religiosity affects maritaloutcome longitudinally. This potential media-tional model was tested with the longitudinaldata collected in Study 2. No support was foundfor the compensation model; that is, religiositydoes not seem to moderate the impact of riskvariables such as age and neuroticism on maritalsatisfaction cross-sectionally. It was certainlypossible, however, that the compensation modelbetter described the longitudinal impact of thesevariables on marital satisfaction. It was alsopossible that religiosity.served as compensatorymechanism for marital satisfaction itself, keep-ing couples who experience declines in satisfac-tion from getting divorced. Both of these addi-tional hypotheses were tested in Study 2.

.a

IEC

<3

S

Is:

•2

I

1a

3

o"So

I

s

38 *o o

88 88

0 \ 0 O v -en Ov *O Tt

t̂ m mmrt © — o

f f

* ** * *mm — Tt— ts © m

f f ' r

) — — OSO O O

SB 88

ON 00 «—< r o

I I

B'u B 'o•o « g « getf « S us U 3

2 < Z > < Z

§

u V™ ft.

a*to *

n *

618 SULLIVAN

Study 2

The purpose of Study 2 was to determinewhether the findings in Study 1 would be rep-licated in another sample and to investigate thelongitudinal effect of religiosity on marital out-come. As with other cross-sectional studies,Study 1 did not allow for the evaluation ofseveral key questions, including whether religi-osity (or variables associated with religiosity)actually causes changes in marital satisfactionand stability; that is, whether being more reli-gious protects couples from experiencing de-clines in satisfaction or from divorcing over thecourse of their marriage. To answer these keyquestions, a second sample of newlywed cou-ples married for the first time was recruited. Asin Study 1, couples' religiosity, marital quality,and marital risk variables were assessed within6 months following their wedding (Time 1).Couples' marital satisfaction was then assessed1 year after Time 1 (Time 2). Finally, couples'current marital status (divorced or still married)was assessed 4 years after Time 1 (Time 3). Asin Study 1, three explanatory models weretested to determine the relationship betweenreligiosity and marital functioning; however,the specific hypotheses generated within each ofthe three models were extended for the longitu-dinal analyses as follows.

Findings consistent with the direct modelwould be that religiosity predicts marital satis-faction 1 year later and that religiosity predictsstability after about 4 years of marriage. Again,previous findings are conflictual and weak, so itwas tentatively hypothesized that religiositywould not have a direct effect on future satis-faction and stability.

For the indirect model, it was hypothesized(on the basis of the findings in Study 1) that theinitial relationship between religiosity and mar-ital satisfaction would be mediated by couples'attitudes but not by their behaviors. Further-more, the assumption that these attitudes (di-vorce attitudes and help-seeking attitudes) pre-dict marital satisfaction or marital stabilitylongitudinally would be empirically tested. Ifcouples' attitudes do predict marital outcome 4years later, and if there is evidence for a directeffect of religiosity on marriage over time, onecould then test whether attitudes act as a medi-ator between religiosity and marital outcome.

Finally, for the compensation model, it wasnot expected that religiosity moderates the ef-

fect of age at marriage or neuroticism on maritalsatisfaction at Time 1, based on the findingsfrom Study 1. However, it was tentatively hy-pothesized that religiosity would compensatefor the effect of these risk variables on maritalsatisfaction 1 year later. In addition, the ques-tion of whether couples who were more reli-gious were less likely to divorce when experi-encing declines in marital satisfaction comparedto couples who were less religious wasaddressed.

Method

Participants

Newspaper advertisements were used to invitecouples to participate in a longitudinal study on new-lywed marriage. Criteria for participation were iden-tical to the criteria used in Study 1. More than 350couples contacted the lab, and the first 60 coupleswho met the criteria and kept their scheduled ap-pointment were included in the sample.2 Four cou-ples (7%) withdrew from the study before Time 3,resulting in a final sample of 56 couples. Husbandswere on average 25.4 (SD = 3.4) years old and had15.6 (SD = 2.2) years of education. Wives were onaverage 24.0 (SD = 2.9) years old and had 16.2(SD = 2.1) years of education. Husbands and wiveshad a modal gross income between $11,000 to$20,000. Husbands reported their ethnicity as Cauca-sian (75%), Asian American (5%), Hispanic (11%),African American (5%), and other (4%). Wives re-ported their ethnicity as Caucasian (76%), Asian-American (9%), Hispanic (9%), African American(4%), and other (2%). Husbands identified as Protes-tant (25%), Catholic (18%), Jewish (18%), Mormon(7%), no religion (30%), and other (2%). Wivesidentified as Protestant (30%), Catholic (18%), Jew-ish (19%), Mormon (11%) no religion (20%), andother (2%).

2 A study was conducted to determine samplingdifferences between couples who responded to thepostcards and couples who did not (Kamey et al.,1995). Couples who responded had more educationand higher status jobs, and they were more likely tohave cohabited premaritally compared to coupleswho did not respond. In contrast, couples recruited byadvertisement (the sample used in Study 2) wereyounger, had lower incomes, and had fewer years ofeducation, compared to couples recruited by marriagelicenses (the sample used in Study 1). For completedetails on differences between responders and non-responders and between Sample 1 and Sample 2, seeKarney et al.

SPECIAL SECTION: RELIGIOSITY AND MARRIAGE 619

Procedure

Time 1. Time 1 procedures were very similar tothose used in Study 1. Couples participated in alaboratory session during which they completedquestionnaires and participated in a problem-solvingdiscussion. Couples were again assessed in four ar-eas: religiosity, marital quality (MAT, divorce atti-tudes, and help-seeking attitudes), behavior (theamount of negative and positive affect in a problem-solving discussion), and marital risk variables (ageand neuroticism). In this study, couples participatedin one 15-min problem-solving discussion in whichthey discussed a mutually agreed upon marital prob-lem chosen from the Inventory of Marital Problems.Couples were paid $50.

Follow-up. Couples' marital satisfaction wasmeasured again 1 year (Time 2) and 3 1/2 years(Time 3) following their laboratory session using theMAT as part of a larger packet of questionnaires. AtTime 2, one couple had already divorced; 55 couplesprovided marital satisfaction data. Couples' maritalstatus (divorced, separated, or still together) was alsoobtained at the final follow-up, approximately 5 yearsafter they were married. Of the 56 couples, 18 (32%)had separated or divorced and 38 couples (68%) werestill married. Couples were paid $25 for eachfollow-up.

Results

The purpose of Study 2 was to verify thecross-sectional findings of Study 1 and to de-termine whether religiosity had a causal effecton marital satisfaction and stability, either di-rectly, indirectly, or as a compensatory mecha-nism, by following newlywed couples over thefirst 5 years of their marriage. Hierarchical re-gression analyses were conducted to determinewhether the cross-sectional relationships foundin Study 1 were replicated in Study 2. Becausethe longitudinal predictors of marital satisfac-tion may be different than the longitudinal pre-dictors of marital stability, separate analyseswere used to evaluate the three models for lon-gitudinal outcomes. To test the direct model,hierarchical regression was used to test whetherhusbands' and wives' religiosity predicted theirown and their spouses' marital satisfaction overtime, after controlling for Time 1 satisfaction.To determine whether husbands' and wives'religiosity at Time 1 predicted whether theywere divorced or still married 3 1/2 years laterwhile controlling for Time 1 satisfaction, logis-tic regression was used.

The finding that religiosity does affect cou-

ples' attitudes but not actual behavior wheninteracting with each other was tested again inStudy 2. Moreover, longitudinal data allowedfor a more complete evaluation of the indirectmodel, providing the means to empirically ver-ify whether other areas of marital functioningthat are related to religiosity (i.e., couples' di-vorce attitudes) do actually predict future mar-ital satisfaction and stability. The presence ofsuch a relationship would provide even bettersupport for the indirect model and would indi-cate the possibility of a mediational model,wherein attitudes toward divorce, for example,mediated the relationship between religiosityand marital satisfaction 1 year later. Hierarchi-cal regression analyses were used to test theserelationships and to determine whether evi-dence supported a mediational model of religi-osity and marital satisfaction.

Finally, the compensation model was testedlongitudinally to determine (a) whether religi-osity moderated the impact of risk variables onfuture marital satisfaction and (b) whether reli-giosity moderated the impact of declines in mar-ital satisfaction on marital stability. Changescores were calculated between Time 1 andTime 2 satisfaction (change in satisfaction overthe 1st year of marriage) to determine whetherdeclines in satisfaction were compensated forby religiosity. Procedures for testing for mod-eration were used (as described above), usingthe change score and the measure of religiosityas predictors of divorce status 3 1/2 years later.

Cross-Sectional Analyses

Hierarchical regression analysis was used toevaluate whether the cross-sectional findings inStudy 1 were replicated in Study 2. As ex-pected, husbands' and wives' religiosity did notdirectly predict their own Time 1 marital satis-faction (/3s = —.04 and - .12, ns), nor didhusbands' religiosity predict their wives' Time1 marital satisfaction (/3 = .02, ns) after con-trolling for age. Interestingly, wives' religiositynegatively predicted husbands' Time 1 satisfac-tion (/3 = - .32, p < .05): Wives who weremore religious had less satisfied husbands. Con-sistent with Study 1, religiosity predicted thedivorce attitudes of husbands (/3 = —.64, p <.01) and of wives (/3 = —.57,p < .01), account-ing for 39% and 32% of the variance, respec-tively. Also consistent with Study 1, religiositypredicted the help-seeking attitudes of husbands

620

Religiosity

- .64" (H)

-.57** (WJ/

<

y

SULLIVAN

Divorce

Attitudes

Direct: .04 (H) -.08 (W)

-.03 (H)

\--19 (W)

T2 Marital

Satisfaction

Mediated: .04 (H) -.25 (W)

Religiosity

.32*. 3 4 "

Direct:Mediated:

Help-Seeking

.04 (H) -.08 (W)

.13 (H) -.05 (W)

-.24 (H)* ^ . 0 4 ( W )

T2 MaritalSatisfaction

Figure 1. Do attitudes mediate the relationship between religiosity and T2 marital satisfaction (aftercontrolling for Time 1 Marital Satisfaction)? H = husband, W = wife, T2 = Time 2. * p < .05. ** p < .01.

(/3 = .32, p < .01) and of wives (/3 = .34, p <.05), accounting for an additional 10% and 11%of the variance beyond that accounted for bymarital satisfaction and age. Consistent withStudy 1, religiosity did not predict negative orpositive behavior for husbands (/3 = .17) or forwives (/3 = .18).

Results were also consistent for analysesevaluating the compensation model. Althoughrisk factors were significantly related to maritalsatisfaction, the interaction of religiosity and therisk factors was not significant for any of theregression analyses. Thus, religiosity did notmoderate the impact of any of the risk variableson Time 1 marital satisfaction for husbands orfor wives.

Longitudinal Analyses

Procedures for testing mediational modelswere followed to determine whether divorceattitudes or help-seeking attitudes mediated therelationship between Time 1 religiosity andTime 2 marital satisfaction, controlling forTime 1 marital satisfaction. To test for media-tion (after first entering Time 1 marital satisfac-tion), the predictor variable (religiosity) and themediator (e.g., divorce attitudes) were enteredin a simultaneous regression to predict the out-come variable (Time 2 marital satisfaction). Ifthe effect of religiosity on Time 2 marital sat-isfaction decreases significantly or becomes

nonsignificant when controlling for divorce at-titudes' effect on marital satisfaction, for exam-ple, this suggests that mediation exists (Baron &Kenny, 1986).

Two mediational models were tested and aredepicted in Figure 1. Betas for husbands andwives are presented next to the appropriate path-ways, with betas for the direct effect of religiosityon Time 2 marital satisfaction and betas for themediated relationship between religiosity andTime 2 marital satisfaction presented below themodel. As explained above, religiosity was a sig-nificant predictor of both potential mediating vari-ables at Time 1 (i.e., divorce attitudes and help-seeking attitudes). However, divorce attitudes didnot predict Time 2 satisfaction, nor did help-seeking attitudes for husbands or for wives. Inaddition, religiosity did not predict Time 2 maritalsatisfaction in any of the models.

Logistic regression was used to test whetherreligiosity mediated the relationship betweenattitudes and marital stability after controllingfor Time 1 satisfaction. Religiosity did not sig-nificantly predict Time 3 divorce status for hus-bands or wives after entering Time 1 maritalsatisfaction (change in x2 = 2.33 and .19, its),nor did divorce attitudes (change in x2 = .03and .10, ns) or help-seeking attitudes (change inX2 — -00 and .33, ns). Overall these findings donot provide support for a mediational model ofreligiosity and marital outcome.

SPECIAL SECTION: RELIGIOSITY AND MARRIAGE 621

The compensation model was evaluated todetermine whether (a) the impact of risk vari-ables on longitudinal satisfaction was moder-ated by religiosity and (b) the impact of declinesin marital satisfaction on marital stability wasmoderated by religiosity. Results of hierarchicalregressions testing whether religiosity compen-sates for the effect of risk variables on longitu-dinal marital satisfaction are presented in Table5. Overall, little support was provided for thecompensation model, with most main effectsand predictors being nonsignificant. The inter-action of husbands' neuroticism and religiositydid significantly predict their own and theirwives' Time 2 marital satisfaction, after con-trolling for Time 1 marital satisfaction; how-ever, the nature of the interaction was inconsis-tent with the compensation model. Amonghusbands who were higher in neuroticism,higher religiosity led to lower levels of maritalsatisfaction for themselves and their spouses.Only among husbands who were low in neurot-icism did higher religiosity lead to higher levelsof marital satisfaction for spouses.

Two final sets of logistic regression analyseswere run to test whether the relationship be-tween changes in marital satisfaction and Time3 stability was moderated by religiosity. Hus-bands' and wives' changes in satisfaction fromTime 1 to Time 2 significantly predicted theirmarital status at Time 3 (x2 = 10.62 and 9.89,p < .01), but religiosity did not predict Time 3marital status for husbands or wives (x2 = 2.14and 0.06, ns), nor was the interaction betweendecline in satisfaction and religiosity significantfor husbands or wives within (x2 = 2.64 and1.40, ns) or between spouses C^2 = .33 and .07,ns). Thus, it does not appear that religiositymoderates the impact of changes in marital sat-isfaction on marital stability. To ensure thatreligiosity does not act differently as a moder-ator when considering only couples who de-clined in satisfaction, the analyses were runagain, excluding couples who maintained orincreased their marital satisfaction from Time 1to Time 2 (17 husbands and 21 wives were inthis category). The findings using the smallersamples (36 husbands and 32 wives) of spouseswho declined in marital satisfaction were verysimilar to the findings for the entire sample; theinteraction between decline in satisfaction andreligiosity was not significant for husbands or

oo

a

•2So

I

5

I1

aI1 !

•5 s

•2,0;

.2,0:

P I

O

csi oqt> ON

ro csOO

l ' I

CO —

r r

CN OO —

) m>O

qo

r r

E o

U-l COCO O

> > n

CM CO

o o

V~> CO

•̂ o

88

co r-— o

* #CO CO

E;§

R 3

00vo

^ £

7J § 5

<C5 #

622 SULLIVAN

wives withintween spouse

= 0.00 and 0.13, ns) or be-= 0.00 and 2.57, ns).3

Discussion

The association between religiosity and mar-riage at Time 1 seemed best described by theindirect model. There was little support for thedirect model, although wives' religiosity didpredict husbands' Time 1 satisfaction. It is in-teresting that husbands of more religious wiveswere actually less satisfied in their marriages, afinding that is inconsistent with much of theprevious literature. Consistent with Study 1,religiosity did predict other marital quality vari-ables in the attitudinal domain but not in thebehavioral domain. Finally, there was no indi-cation that religiosity moderated the impact ofrisk variables on Time 1 marital satisfaction.The consistency of results across the two studiesyields a fairly clear picture of how religiosityaffects marriage cross-sectionally in newly-weds. It appears that religiosity is important inthat it predicts spouses' attitudes toward divorce(with more religious spouses being less likely tosee divorce as an option) and their willingnessto seek help in times of trouble (with morereligious spouses being more willing to seekhelp). Being more religious does not seem tolead to higher satisfaction in the relationship; infact, there is some evidence that newlywed hus-bands who have more religious wives are actu-ally less satisfied with the relationship than hus-bands whose wives are not as religious.

Longitudinal analyses reveal similar findingsover time for the direct relationship betweenreligiosity and marital satisfaction. That is,higher levels of religiosity do not appear to leadto higher levels of marital satisfaction for eitherhusbands or wives. The attitudinal variables thatare predicted by religiosity at Time 1 (i.e., moreconservative divorce attitudes and higher will-ingness to seek help in times of marital trouble)do not lead to higher levels of satisfaction atTime 2, either. Therefore, it seems unlikely thatreligiosity has either a direct or an indirect im-pact on marital satisfaction over the first fewyears of newlywed marriage.

There is some evidence that religiosity mod-erates the relationship of risk variables on mar-riage, although not in a compensatory manner.Couples with high-risk husbands (i.e., husbandswith higher levels of neuroticism) who weremore religious reported lower levels of satisfac-

tion compared to couples with high-risk hus-bands who were less religious. Therefore, reli-giosity does moderate the impact of at least onetype of risk variable, but in a different fashionthan the compensation model would suggest.

Finally, these data provide no evidence thatreligiosity moderates the effect of changes insatisfaction on marital stability. Thus, highlyreligious couples who experience declines insatisfaction over the 1st year of marriage appearno less likely to be divorced within the first 4years of marriage than less religious coupleswho experience declines in satisfaction.

General Discussion

In trying to understand how religiosity mightaffect marriage, many researchers have reliedon cross-sectional data. This approach is lim-ited, not only because it is impossible to deter-mine whether religiosity is affecting maritalfunctioning or marital functioning is affectingreligiosity, but because it appears that the cross-sectional effect of religiosity on marriage isfundamentally different from the longitudinaleffect of religiosity and marriage. Religiosity isrelated to couples' attitudes. Specifically, cou-ples who are more religious are more likely tohave more conservative divorce attitudes andhigher levels of marital commitment, and theyare more likely to seek help in times of maritaltrouble. The finding that religiosity affects atti-tudes and not behavior is consistent with Boothet al.'s (1995) hypothesis that "some dimen-sions of marital quality may be affected byreligious involvement, while others may not.For example, attitudes toward marriage may beaffected by religion, while behavioral attributesof marriage remain unchanged" (p. 663).

Longitudinally, however, these attitudes do

3 One difference did emerge when testing onlyspouses who declined in marital satisfaction over the1st year of marriage. Among wives who declined inmarital satisfaction from Time 1 to Time 2 (n = 32),husbands' religiosity did predict the couples' maritalstatus at Time 3 (x2 = 4.56, p < .05). However, asmentioned in the text, the interaction between de-clines in marital satisfaction and religiosity was notsignificant, indicating that even among wives whosemarital satisfaction declined over time, religiosity didnot moderate the effect of declines on satisfaction.(The very small sample used for this analysis war-rants viewing all these findings with caution.)

SPECIAL SECTION: RELIGIOSITY AND MARRIAGE 623

not appear to affect marital satisfaction or sta-bility. Instead, religiosity appears to affect mar-ital satisfaction by moderating the effect of atleast one risk variable, neuroticism. It is inter-esting that religiosity has a positive relationshipwith marital satisfaction for husbands who areless neurotic. For more neurotic husbands, reli-giosity actually has a negative relationship withsatisfaction. Thus, it appears that religiositydoes promote marital satisfaction over time, butonly for relatively healthier husbands, that is,those who are less reactive and negative ingeneral.

Although religiosity does appear to havesome impact, both cross-sectionally and longi-tudinally, on marriage, it is important to notethat the relationship appears to be weak andinconsistent. Cross-sectionally, the only directassociation between religiosity and marital sat-isfaction is a negative one, with husbands mar-ried to wives who are high in religiosity beingless satisfied (Study 2). Although religiosity isrelated to marital attitudes, these attitudes donot themselves appear to have a positive effecton marital satisfaction or stability longitudi-nally, at least over the first 4 years of marriage.Therefore, it appears that the more conservativedivorce attitudes associated with higher levelsof religiosity do not necessarily make couplesless likely to get divorced, nor do more positiveattitudes toward help seeking. Longitudinally,there is no indication that religiosity has a directeffect on whether couples stay together. Theonly indication that religiosity has a positiveeffect on marital satisfaction over time is thatreligiosity predicts higher levels of satisfactionamong husbands low in neuroticism. This is aninteresting finding, one that might begin to ac-count for the null and sometimes contradictoryresults in these and other studies.

Another possible explanation for the appar-ently weak relationship between marital satis-faction and stability is that religiosity may be-come more important to marital satisfaction orstability later in marriage. For example, consis-tent with most religions' support of family life,couples with children may be more affected bytheir level of religiosity. The lack of evidencethat religiosity makes couples less likely to di-vorce or separate in the face of marital distressmay also be unique to newlywed marriage, inthat there is a much more restricted range ofmarital satisfaction scores. As couples' mar-

riages mature and they encounter challengessuch as the birth of children, religiosity maybecome more important in predicting maritalstability.

At least three important methodological lim-itations in the studies presented here should betaken into account. The first is the relativelysmall sample size in Study 2. Fifty-six couplesprovided data throughout the data collection,and for one set of analyses (whether religiositymoderated declines in marital satisfaction) only36 husbands and 32 wives were appropriate foranalyses. It is certainly possible that some mar-ginally significant findings might have emergedas significant with a larger sample (e.g., divorceattitudes might mediate the relationship be-tween religiosity and marital satisfaction). Sec-ond, because these were large studies designedto assess many variables related to marital out-come and processes, brief assessment instru-ments were often used. Most relevant to thisarticle, the measure of religiosity was brief, withonly one question to assess each of three dimen-sions of religiosity and one global question. It iscertainly possible that a longer, more thoroughmeasure, or a measure of religiosity that is moreproximal to marriage (e.g., involvement in jointreligious activities and perceptions regardingthe sanctification of marriage; see Mahoney etal., 1999), might have been a more powerfulpredictor of marital functioning. Finally, only afew potential mediating and moderating vari-ables were tested. It certainly seems reasonablethat other variables not considered here mayserve to mediate the relationship between reli-giosity and marital outcome or that their impactmay be moderated by couples' religiosity.

The current studies were designed to testthree explanatory models of religiosity and mar-ital outcome. Many psychologists have beencalling for a more theory-based approach toempirical research, and specifically "within thepsychology of religion, the cry for good theoryhas reached the level of cacophony" (Hood,Spilka, Hunsberger, & Gorsuch, 1996, p. 446).What theoretical propositions can these studiesoffer to guide ongoing research in the area ofreligiosity and marital outcome? Overall, it ap-pears that religiosity may affect marriage, butonly under certain conditions. The identificationof these conditions is an important next step tomore fully understanding the impact of religi-osity on marriage over time. Two potential do-

624 SULLIVAN

mains for identifying these conditions are cou-ple type and type of religiosity. Both domainsmay be further broken down into a series oftheoretical propositions. In the domain of cou-ple type, religiosity may differentially affectcouples based "on the intrapersonal characteris-tics of the individual spouses (e.g., other per-sonality variables in addition to neuroticism,family history) or on the interpersonal charac-teristics (e.g., how spouses support one another,whether spouses engage in violence) of the cou-ple. In the domain of type of religiosity, manymultidimensional models of religiosity havebeen proposed that may account for the differ-ential impact of religiosity on marriages. Someexamples include intrinsic versus extrinsic ori-entations (Allport, 1966), religion as a meansversus religion as an end (Batson & Ventis,1982), and guilt-oriented versus love-orientedreligiosity (McConahay & Hough, 1973; for anexcellent summary of multidimensional models,see Hood et al., 1996).

The current longitudinal findings give somesupport for the proposition that religiosity's ef-fect on marital satisfaction is dependent on thetype of couple, particularly on the intrapersonalcharacteristics of the husbands. In this case, itseems that religiosity operates to enhance themarital satisfaction of couples with less nega-tive and less reactive husbands. Among hus-bands who are more negative and more reactive,religiosity seems to actually reduce the maritalsatisfaction of husbands and wives. This find-ing, which certainly requires replication, opensthe door to potentially important hypothesesregarding the interaction between religiosityand personality variables in predicting maritalfunctioning. For example, it is possible thatneurotic people do not think about or use reli-gion in constructive ways for their marriages. Inaddition, despite the lack of evidence from thesestudies that religiosity affects communicationbehavior during conflict, it is certainly possiblethat religiosity affects marital satisfactionthrough other important interpersonal domains.

It also remains to be seen whether religiositymight differentially affect marital satisfactionbased on the type of religiosity that character-izes the spouses. One reasonable supposition isthat the effect of religion is different for spouseswho are intrinsically as opposed to extrinsicallyoriented. People who are intrinsically orientedare personally committed to their faith, devout,

and more open and tolerant of different ideasand positions. People who are extrinsically mo-tivated are more likely to "follow the rules,"have more superficial beliefs, and be less toler-ant of different viewpoints (Hunt & King,1971). One possibility may be that religiositymay affect marital satisfaction for couples whoare more intrinsically oriented and may affectmarital stability for couples who are more ex-trinsically oriented.

In summary, I add my voice to the cacophonyof calls for theory-directed research on religios-ity and marriage. The use of the theoreticalpropositions offered here or elsewhere allows amore systematic and ultimately fuller under-standing of how religiosity operates in maritalrelationships to affect couples' satisfaction andstability. This understanding would be useful toclergy and psychologists who work with cou-ples to improve and enhance the quality of theirmarriages.

Implications for Application andPublic Policy

The most obvious application of the currentwork is for those who work with couples pre-paring for marriage, but these findings may alsoapply to clinicians working with married cou-ples in distress. Couples therapists and clergywho work with couples should be careful not toassume that religious devotion shields couplesfrom declines in satisfaction or divorce. Instead,those who work with couples should considercarefully, with the couple, the role that religi-osity plays in the relationship. Careful process-ing of the role of religiosity in the relationship,along with an understanding of each partner andother important aspects of the relationship, mayenhance the long-term effectiveness of maritalinterventions by clinicians and clergy.

References

Aiken, L. S., & West, S. G. (1991). Multiple regres-sion: Testing and interpreting interactions. New-bury Park, CA: Sage Publications.

Albrecht, S. L., & Kunz, P. R. (1980). The decisionto divorce: A social exchange perspective. Journalof Divorce, 3, 319-337.

Allport, G. W. (1966). The religious context of prej-udice. Journal for the Scientific Study of Religion,5, 447-457.

Anthony, M. J. (1993). The relationship between

SPECIAL SECTION: RELIGIOSITY AND MARRIAGE 625

marital satisfaction and religious maturity. Reli-gious Education, 88, 97-108.

Bahr, H. M., & Chadwick, B. A. (1985). Religion andfamily in Middletown, USA. Journal of Marriageand the Family, 47, 407-414.

Baron, R. M._ & Kenny, D. A. (1986). Themoderator-mediator variable distinction in socialpsychological research: Conceptual, strategic, andstatistical considerations. Journal of Personalityand Social Psychology, 51, 1173-1182.

Batson, C. D., & Ventis, W. L. (1982). The religiousexperience: A social-psychological perspective.New York: Oxford University Press.

Booth, A., Johnson, D. R., Branaman, A., & Sica, A.(1995). Belief and behavior: Does religion matter intoday's marriage? Journal of Marriage and theFamily, 57, 661-671.

Bugaighis, M. A., Schuram, W. R., Jurich, A. P., &Bollman, S. R. (1985). Factors associated withthought of marital separation. Journal of Divorce,9, 49-59.

Call, V. R., & Heaton, T. B. (1997). Religious influ-ence on marital stability. Journal for the ScientificStudy of Religion, 36, 382-392.

Clydesdale, T. T. (1997). Family behaviors amongearly U.S. baby boomers: Exploring the effects ofreligion and income change, 1965-1982. SocialForces, 76, 605-635.

Ellison, C. G., Bartkowski, J. P., & Anderson, K. L.(1999). Are there religious variations in domesticviolence? Journal of Family Issues, 20, 87-113.

Eysenck, S. B., & Eysenck, H. J. (1978). Impulsive-ness and venturesomeness: Their position in a di-mensional system of personality description. Psy-chological Reports, 43, 1247-1255.

Fergusson, D. M., Horwood, L. J., & Shannon, F. T.(1984). A proportional hazards model of familybreakdown. Journal of Marriage and the Family,46, 539-549.

Filsinger, E. E., & Wilson, M. R. (1984). Religiosity,socioeconomic rewards, and family development:Predictors of marital adjustment. Journal of Mar-riage and the Family, 46, 663-670.

Geiss, S. K., & O'Leary, K. D. (1981). Therapistratings of frequency and severity of marital prob-lems. Journal of Marital and Family Therapy, 7,515-520.

Glenn, N. D., & Supancic, M. (1984). The social anddemographic correlates of divorce and separation inthe United States: An update and reconsideration.Journal of Marriage and the Family, 46, 563-575.

Glock, C. (1959). The religious revival in America?In J. Zahn (Ed.), Religion and the face of America.Berkeley: University of California Press.

Gottman, J. M., & Krokoff, L. J. (1989). Maritalinteraction and satisfaction: A longitudinal view.Journal of Consulting and Clinical Psychology, 57,47-52.

Hansen, G. L. (1987). The effect of religiosity on

factors predicting marital adjustment. Social Psy-chology Quarterly, 50, 264-269.

Hood, R. W., Spilka, B., Hunsberger, B., & Gorsuch,R. (1996). The psychology of religion: An empiricalapproach. New York: Guilford Press.

Hunt, R. A., & King, M. B. (1971). The intrinsic-extrinsic concept: A review and evaluation. Journalfor the Scientific Study of Religion, 10, 339-356.

Hunt, R. A., & King, M. B. (1978). Religiosity andmarriage. Journal for the Scientific Study of Reli-gion, 17, 399-406.

Jenkins, K. W. (1991). Religion and families. In S. J.Bahr (Ed.), Family research: A sixty-year review,1930-1990 (pp. 235-288). Lexington, MA: Lex-ington Books.

Karney, B. R., & Bradbury, T. N. (1995). The lon-gitudinal course of marital quality and stability: Areview of theory, method, and research. Psycholog-ical Bulletin, 118, 3-34.

Karney, B. R., Davila, J., Cohan, C. L., Sullivan,K. T., Johnson, M., & Bradbury, T. N. (1995). Anempirical investigation of sampling strategies inmarital research. Journal of Marriage and the Fam-ily, 57, 909-920.

Kaslow, F., & Robison, J. A. (1996). Long-termsatisfying marriages: Perceptions of contributingfactors. The American Journal of Family Therapy,24, 153-169.

Kelly, E. L., & Conley, J. J. (1987). Personality andcompatibility: A prospective analysis of marital sta-bility and marital satisfaction. Journal of Personal-ity and Social Psychology, 52, 27-40.

Kunz, P. R., & Albrecht, S. L. (1977). Religion,marital happiness, and divorce. International Jour-nal of Sociology of the Family, 7, 227-232.

Kurdek, L. A. (1991). Marital stability and changesin marital quality in newlywed couples: A test ofthe contextual model. Journal of Social and Per-sonal Relationships, 8, 27-48.

Larson, L. E., & Goltz, J. W. (1989). Religiousparticipation and marital commitment. Review ofReligious Research, 30, 387-400.

Levinger, G. (1976). A social psychological perspec-tive on marital dissolution. Journal of Social Issues,32, 21-47.

Locke, H. J., & Wallace, K. M. (1959). Short maritaladjustment prediction test: Their reliability and va-lidity. Marriage and Family Living, 21, 251-255.

Mahoney, A., Pargament, K. I., Jewell, T., Swank,A. B., Scott, E., Emery, E., & Rye, M. (1999).Marriage and the spiritual realm: The role of prox-imal and distal religious constructs in marital func-tioning. Journal of Family Psychology, 13, 321—338.

McConahay, J. B., & Hough, J. C , Jr. (1973). Loveand guilt-oriented dimensions of Christian belief.Journal for the Scientific Study of Religion, 12,53-64.

Nye, F. I., White, L.t & Frideres, J. S. (1973). A

626 SULLIVAN

preliminary theory of marital stability: Two mod-els. International Journal of Sociology of the Fam-ily, 3, 102-122.

Rohrbaugh, J., & Jessor, R. (1975). Religiosity inyouth: A personal control against deviant behavior.Journal of Personality, 43, 136-155.

Schrum, W. (1980). Religion and marital instability:Change in the 1970s? Review of Religious Re-search, 21, 135-147.

Schumm, W. R., Bollman, S. R., & Jurich, A. P.(1982). The "marital conventionalization" argu-ment: Implications for the study of religiosity andmarital satisfaction. Journal of Psychology andTheology, 10, 236-241.

Schumm, W. R., Obiorah, F. C , & Silliman, B.(1989). Marital quality as a function of conservativereligious identification in a sample of Protestant andCatholic wives from the midwest. PsychologicalReports, 64, 124-126.

Snow, T. S., & Compton, W. C. (1996). Maritalsatisfaction and communication in fundamentalistProtestant marriages. Psychological Reports, 78,979-985.

Stanley, S., & Markman, H. (1992). Commitment:The role of dedication and constraint in personalrelationships. Journal of Marriage and the Family,54, 595-608.

Thornes, B., & Collard, J. (1979). Who divorces?London: Routledge & Kegan Paul.

Veroff, J. (1988a). Marital Help-Seeking Measure.Unpublished manuscript, University of Michigan.

Veroff, J. (1988b). Measure of Divorce Attitudes.Unpublished manuscript, University of Michigan.

Wallin, P. (1957). Religiosity, sexual gratification,and marital satisfaction. American Sociological Re-view, 22, 300-305.

Wallin, P., & Clark, A. L. (1964). Religiosity, sexualgratification, and marital satisfaction in the middleyears of marriage. Social Forces, 42, 303-309.

White, L. K., & Booth, A. (1991). Divorce over thelife course: The role of marital happiness. Journalof Family Issues, 12, 5-21.

Wilson, J., & Musick, M. (1996). Religion and mar-ital dependence. Journal for the Scientific Study ofReligion, 35, 30-40.

Wilson, M. R., & Filsinger, E. E. (1986). Religiosityand marital adjustment: Multidimensional interre-lationships. Journal of Marriage and the Family,48, 147-151.

Received October 20, 1999Revision received September 22, 2000

Accepted October 2, 2000 •

Wanted: Your Old Issues!

As APA continues its efforts to digitize journal issues for the PsycARTICLESdatabase, we are finding that older issues are increasingly unavailable in our inventory.We are turning to our long-time subscribers for assistance. If you would like to donateany back issues toward this effort (preceding 1982), please get in touch with us atjournals@apa.org and specify the journal titles, volumes, and issue numbers that youwould like us to take off your hands. (Your donation is of course tax deductible.)