Depression, Marital Satisfaction and Communication in Couples: Investigating Gender Differences

www.elsevier.com/locate/bt

Available online at www.sciencedirect.com

Behavior Therapy 41 (2010) 306–316

Depression, Marital Satisfaction and Communication in Couples:Investigating Gender Differences

Barbara GabrielSteven R.H. BeachUniversity of Georgia

Guy BodenmannUniversity of Zuerich

The correlation between depression and dysfunctionalmarital interaction is well documented, but only a fewstudies have examined gender-related differences in maritalinteraction patterns of couples with a depressed partner. Inthis paper we examined differences in observed maritalcommunication in a sample of 62 Swiss couples presentingfor treatment of depression. There were 16 maritallydistressed couples with a depressed wife, 21 maritallynondistressed couples with a depressed wife, 18 maritallydistressed couples with a depressed husband, and 7maritally nondistressed couples with a depressed husband.Marital interaction behavior was found to depend ongender, depression, marital distress, as well as gender ofthe depressed partner. Our results suggest the need for agender-sensitive model of the link between marital interac-tion and depression.

INTERPERSONAL PROBLEMS AND SOCIAL deficits, espe-cially in the marital context, are strongly associatedwith the development, intensity, and course ofdepression (Backenstrass, 1998; Beach, Jones, &Franklin, 2008; Bodenmann, 2006; Joiner, Coyne,

This study was supported by the Swiss National ScienceFoundation SNF 610-062901 and 100013-109547/1.

Address correspondence to Barbara Gabriel, Institute forBehavioral Research, 514 Boyd, GSRC, University of Georgia,Athens, GA 30602; e-mail: [email protected]/10/306–316$1.00/0© 2010 Association for Behavioral and Cognitive Therapies. Published byElsevier Ltd. All rights reserved.

& Blalock, 1999). Several studies have identifiedlack of verbal and nonverbal positivity, asymmetryin interaction patterns, a higher level of passivity/withdraw, negative statements (e.g., complaints,negative self-statements) as well as negative reci-procity in couples with a depressed partner (e.g.,Backenstrass, 1998; Biglan et al., 1985; Gotlib &Whiffen, 1989; Hautzinger, Linden, & Hoffman,1982; Jacob & Leonard, 1992; Johnson & Jacob,1997; Sher, Baucom, & Larus, 1990). Couples witha depressed member are, on average, more negative,less positive and less congenial in their maritalinteractions compared to couples without a de-pressed partner (e.g., Johnson & Jacob, 1997;McCabe & Gotlib, 1993). In keeping with thispattern, depressed persons display in their maritalcommunication a higher frequency of interruption,expression of negative feelings, criticism anddefensiveness, as well as a lower level of nonverbalpositivity. The partners of the depressed persons, onthe other hand, show enhanced criticism andnegativity directed toward the depressed partnerand the marital relationship (e.g., Benazon &Coyne, 2000; Hautzinger et al., 1982).One might wonder, however, whether gender of

the depressed partner is associated with differencesin the patterns of interaction observed amongcouples with a depressed partner. Relatively fewstudies have compared couples with depressedwives to couples with depressed husbands (seeJohnson & Jacob, 1997, 2000, for an exception).As a consequence, gender-linked patterns in thiscontext are not currently well-understood. There

mailto:[email protected]

307depre s s ion , mar i tal sat i s fact ion , and commun icat ion

are several reasons to expect that gender-linkedpatterns may be important in depression. Somestudies have found a different time course in theconnection between marital satisfaction and de-pression as a function of gender, with evidence ofan effect of history of marital distress on futuredepression for wives and evidence of an effect ofhistory of depression on future marital distress forhusbands (e.g., Fincham, Beach, Harold, &Osborne, 1997; but see Kurdek, 1998, and Davila,Karney, Hall, & Bradbury, 2003, for evidence ofgender similarity in time course). There is alsoevidence of a gender difference in response tonegative feelings (see also Nolen-Hoeksema, 1990),and this gender difference may be intensified bydepression (Wilhelm, Roy, Mitchell, Brownhill, &Parker, 2002). In particular, men may be morelikely to engage in behaviors that minimize, distractfrom, or avoid interactions that potentially producenegative emotions, whereas women may be morelikely to confront, ruminate, and engage suchinteractions, resulting in greater arousal and greaterdisplay of negative emotion for depressed womenduring interaction tasks than for depressed men.In addition, there is a broader literature on

gender-linked differences in marital interactionthat suggest the importance of attending to gen-der-linked patterns. To the extent that women aremore relationship-oriented than men, for example,they may feel more responsible for the resolution ofrelationship difficulties (Baucom,Notarius, Burnett,& Haefner, 1990; Culp & Beach, 1998), whereasmen may focus more on independence (Culp &Beach, 1998; Gottman, 1994; Klinetob & Smith,1996). Additionally, wives exhibit higher frequen-cies of both positively and negatively valencedbehavior in marital interactions than husbands,including greater nonverbal positivity and increasedlevels of affect expression like self-disclosure,criticism, and complaints. Husbands, on the otherhand, display higher levels of nonaffective, task-oriented behavior and instrumental advice, but alsodisplay anger and blame avoidance at higher ratesand may be more conflict averse on average(Baucom et al., 1990; Christensen & Heavey,1990; Dindia & Allen, 1992; Gottman, 1994;Klinetob & Smith, 1996).Gender-linked patterns have also been captured

in research on demand/withdraw patterns and areparticularly salient when couples are allowed toselect their own discussion topic. In these situations,wives tend to display higher levels of demand andcriticizing behavior, and husbands aremore likely todisplay defensive, withdrawal, and stonewallingbehaviors (e.g., Christensen & Heavey, 1990;Eldridge & Christensen, 2002; Gottman, 1994;

Rehmann&Holtzworth-Munroe, 2006). Because apart of the gender difference in demand-withdrawpatterns observed in previous research is due towives more commonly selecting the topic for couplediscussion when the source of the topic is notspecified by the therapist or researcher (e.g.,Christensen & Heavey, 1990; Klinetob & Smith,1996), it is important to assess who selects the topicof discussion in each interaction. This demand-withdraw pattern is strongly related to otherescalating coercive negative behaviors (e.g., con-tempt, belligerence, domineering) as well as maritaland psychological distress (e.g., Christensen &Shenk, 1991; Gottman, 1994; Gottman&Notarius,2000).Another line of research suggesting the impor-

tance of examining gender in combination withdepression and marital interaction was provided byMarchand and Hock (2000). In their investigationof correlates of self-reported attacking and avoid-ance behavior in a community sample, husbands'lower self-reported marital satisfaction was the onlypredictor of attacking and the strongest predictor ofavoidance behavior, explaining between 15 and 20percent of the variance. For wives, on the otherhand, higher self-reported depression was the onlypredictor of avoidance and the best predictor ofattacking behavior, explaining 26 to 57 percent ofthe variance in self-reported behavior in thesedomains. Conversely, recent work including behav-ioral observation did not support a consistentpattern of results linking gender, depression,marital satisfaction, and wife demand/husbandwithdraw behavior (e.g., Baucom et al., 2007).Accordingly, this is an area in need of furtherinvestigation.Despite much interest in the topic of gender roles

and depression and compelling reasons to examinethe impact of gender on depression-linked interac-tion behavior, little is currently known aboutgender differences in marital interaction in thecontext of depression. The few available studiessuggest that couples with depressed wives show lesspositivity and a trend to a higher negativity thancouples with a depressed husband (Jacob &Leonard, 1992; Johnson & Jacob, 1997, 2000).Depressed husbands, compared to control hus-bands, tend to display less problem solving, lowernegative reciprocity, more positive reactions, andhave less impact on their wives' subsequentbehavior. Wives of depressed husbands, on theother hand, are less likely to engage in positivereciprocity and display increased negativity (e.g.,Jacob & Leonard, 1992; Johnson & Jacob, 2000).Consistent with the broader literature on genderdifferences, affect-relevant communication in

308 gabr i e l e t al .

couples seems to be more strongly influenced bywives. Thus, as for all gender differences, it is likelythat commonalities in the behavior in couples ofdepressed wives and depressed husbands will begreater than will differences (Johnson & Jacob,1997). However, existing results indicate accentu-ation of gender-typical interaction patterns associ-ated with marital distress depending on the genderof the depressed person.It is also important to note that not all couples

with a depressed partner experience marital dis-cord. Based on an extensive review of publishedwork, Whisman (2001) reported that the averageDyadic Adjustment Scale (DAS) score of personsdiagnosed with depression (DAS=93) was in thedistressed range and approximately 1 3/4 standarddeviations below the comparison sample. Since thecutoff for marital discord on the DAS is 98, thisfinding also indicates that many couples with adepressed partner do not score in the discordantrange of the DAS. Coyne, Thompson, and Palmer(2002) reported similar findings in their study of 38depressed outpatients and 35 depressed inpatientswho they compared to a community sample.Replicating and extending the Whisman (2001)results, they found that approximately two thirds ofthe outpatients and approximately one half of theinpatients scored in the distressed range of the DAS.Thus, marital relationships are often (but notalways) distressed among depressed men andwomen.Because some interaction behaviors may be more

tightly connected to level of marital distress in thecontext of depression than others, it is necessary toexamine both maritally distressed and maritallynondistressed couples with a depressed partner andcontrast their interaction patterns. In studies thathave examined this issue, relatively few problematicinteraction behaviors have been found to be tieddirectly to level of depression (Schmaling &Jacobson, 1990). Interaction behavior that ismore tightly connected to level of depressionincludes more negative statements and less con-flict-resolution-oriented behavior. In contrast,greater marital distress is associated with a rangeof negative or aggressive behavior during maritalinteractions as well as less positive or facilitativebehavior (Hautzinger et al., 1982; Jackman-Cram,Dobson, & Martin, 2006; Nelson & Beach, 1990;Schmaling & Jacobson, 1990).In the present paper we investigated differences in

the observed marital interaction between fourgroups. Groups were distinguished on the basis ofmarital satisfaction and the gender of the patient(G1=maritally distressed couples with a depressedwife, G2=maritally nondistressed couples with a

depressed wife, G3=maritally distressed coupleswith a depressed husband, G4=maritally nondis-tressed couples with a depressed husband). Wecoded behavior using categories of the specificaffect coding system (SPAFF: Gottman, 1994) and“Kategoriensystem fuer partnerschaftliche Interak-tion” (KPI: Hahlweg et al., 1986). Based on theliterature reviewed above, the following resultswere expected: (a) significant main effects of gender(e.g., greater negativity and emotional expression inwives) and index person (patient versus partner; seeBenazon & Coyne, 2000; Hautzinger et al., 1982);(b) group differences for distressed versus nondis-tressed groups (e.g., less positivity, higher negativityin maritally distressed groups) and gender ofdepressed person (e.g., higher negativity in de-pressed women than depressed men) especially inthe sense of an exaggeration of gender-typicaldemand/withdraw patterns among distressed cou-ples with a depressed wife.

Methodprocedure

Couples were recruited for participation in atreatment study focusing on depression. Recruit-ment methods included newspaper advertisementsand referals from medical facilities in Switzerland.Criteria for participation in the study were (a)meeting DSM criteria (DSM-IV; American Psychi-atric Association, 1994) for unipolar depressionusing the Structured Clinical Interview for DSM(SKID; German version: Wittchen et al., 1990); (b)BDIN11 (German version: Hautzinger, Bailer,Worall & Keller, 1995); (c) enduring romanticrelationship (duration at least 1 year); (d) sufficientcognitive abilities (IQN80) to complete self-reportmeasures; and (e) agreement of the partner toparticipate on the study. Exclusion criteria included(a) bipolar disorder; (b) secondary depression withadditional comorbidities (e.g., psychosis, personal-ity disorders, alcoholism); as well as (c) acutesuicidal tendencies.Sixty-eight partners were successfully screened on

the phone by means of the BDI and SKID and thelist of inclusion and exclusion criteria. After thisprocedure, an appointment for a detailed diagnosisand videotapedmarital interactions was set. Two orthree weeks before the home assessments, ques-tionnaires were mailed to the home address andwere independently completed and returned at thepre-assessment. The marital interactions werevideotaped in a standardized setting at the couple'shome after the diagnostic interview. The durationof the conflict interaction was 15 minutes. Thecouples discussed a stress-relevant topic concerning


their relationship determined prior the discussion(see below).

participants

From the 68 screened couples, 62 heterosexualcouples (37 depressed wives, 25 depressed hus-bands) seeking therapy for depression met theinclusion criteria and completed the questionnairesas well as the discussion task at the pre-assessment.In keeping with prior research, an average score of54 or more on the marital questionnaire (PFB;Hahlweg, 1996) was used to categorize couples asnondistressed. Those with an average score below54 were categorized as distressed (see Table 1). Thegroups contained 16 distressed couples with adepressed wife (G1), 21 nondistressed coupleswith a depressed wife (G2), 18 distressed coupleswith a depressed husband (G3), and 7 nondis-tressed couples with a depressed husband (G4).There were no significant differences betweengroups in the age of husbands and wives, durationof the close relationships, or years since firstdepression and BDI scores of the depressed personsas well as the partners (see Table 1). However, therewere significantly fewer children in G2 compared toG1, t(36)=2.81, pb .01, and G3, t(35)=3.58,pb .001. Chi-square comparisons revealed no sig-nificant group differences in the income, χ2(3)=2.66 (ns) or the education of wives, χ2(3)=3.25(ns), or husbands, χ2(3)=5.91 (ns). Significantlymore couples in G1 were married than couples inthe other groups, χ2(3)=11.25, pb .01. Husbandsin G1, on the other hand, showed a significantlyhigher percentage of employment than husbands inthe other groups, χ2(3)=10.02, pb .05, whereas

Table 1Sample Means and Standard Deviations for Demographic Variables by

Variable G1 (n=16) G2 (n=21)

M SD M SD

Age wife (year) 45.31 9.32 43.07 13.52Age husband (year) 48.44 8.66 44.81 14.17Number of children 1.91 1.13 0.95 0.93Years of relationship 18.98 9.95 14.94 12.75PFB spouse's 41.80 9.80 69.14 9.79BDI spouse's 6.38 5.76 5.15 6.38PFB patient's 39.36 13.48 67.61 11.39BDI patient's 25.89 7.46 24.55 7.29Years since first

depression10.70 7.01 11.09 13.44

PFB averagea 40.58 10.61 68.37 9.47

Note. G1=distressed couples with a depressed wife. G2=nondistresseG3=distressed couples with a depressed husband. G4=nondistressedBDI=Beck Depression Inventory. PFB=Partnership Questionnaire.a Average of wife's and husband's score.⁎ pb .05.

wives in G1 tended toward lower employment thanwives in the two depressed husband groups,χ2(3)=7.11, pb .1.

measures

Partnership Questionnaire (Partnerschaftsfragebogen,PFB)The PFB (Hahlweg, 1996) has 30 items and uses a4-point scale for each item. It measures maritalquality and satisfaction and consists of three scales:quarrelling (α=.93), tenderness (α=.95), and to-getherness/communication (α=.86). A total scorebelow 54 points designates a low level of satisfac-tion and discriminates reliably between distressedand nondistressed couples (Hahlweg, 1996). ThePFB is a widely used self-report scale with goodinternal consistency (α=.94) and has been exam-ined with regard to validity (Hahlweg, 1996). ThePFB correlates significantly (r=.63) with the Mar-ital Adjustment Test (Locke & Wallace, 1959;Schindler, Hahlweg & Revenstorf, 1998).

Beck Depression Inventory (BDI)The German version (Hautzinger et al., 1995) of theBDI (Beck & Steer, 1987) is a 21-item self reportmeasure of the intensity (4-point scale) of affective,cognitive, and somatic aspects of depression. Thereliability (α=.92) and validity are well-established,and gender bias is low (e.g., Beck, Steer, & Garbin,1988; Hautzinger et al., 1995).

Marital Interaction TaskThe conflict discussions were coded using categoriesof the SPAFF (Gottman, 1994) as well as of the KPI(Hahlweg et al., 1986). Two coders independently

Group

G3 (n=18) G4 (n=7) F(3,58)M SD M SD

47.06 9.83 36.00 8.43 1.8250.44 8.80 40.43 7.52 1.852.00 0.85 1.07 1.54 4.42 ⁎

20.13 10.9 10.32 6.93 1.7642.90 10.42 64.05 7.92 35.07 ⁎

7.11 3.94 5.43 2.57 0.5041.79 10.30 65.43 11.31 26.46 ⁎

26.50 8.22 21.36 6.81 0.879.16 8.94 17.07 9.89 0.97

42.34 7.83 64.74 8.50 40.68 ⁎

d couples with a depressed wife.couples with a depressed husband.


coded the interactions in a five-minute time intervalsin a double-blind-administration. Interobserver reli-abilities were assessed using the Kappa coefficient(Cohen, 1960). Kappas for verbal categories rangedfrom .80 to .97. Kappas for nonverbal categoriesranged from .66 to .85. The construct and predictivevalidity of this observational system have beendemonstrated in prior research (Gottman, 1994;Hahlweg et al., 1986).To facilitate micro-analytic coding of the dura-

tion and frequency of specific interaction codes, acomputer-supported system (Computer Aided Ob-servation System CAOS; Bourquard, Bodenmann& Perrez, 1992–2005) was utilized.Factor analysis (varimax rotation, principal

components) of scores for all 14 behavior codes,across the entire sample, was used to aggregate rarecategories into coherent sets of behavior.1 Thisprocedure resulted in the following seven behav-ioral categories: (1) eye contact, (2) interest/curiosity, (3) nonverbal positivity, (4) emotionalself-disclosure, (5) interruption, (6) criticism-dom-ineering (criticism, domineering), and (7) aggres-sion-defensiveness (contempt, belligerence,nonverbal negativity, stonewalling, defensiveness)(see Gottman, 1994, for details). Although it ismore typical to treat defensiveness as a separatecode (e.g., Gottman, 1994), it was insufficientlyfrequent in our data to do so. As a consequence, weutilized the summary code described above andfound that it fit with our own factor analysis of theobservational data. Other rare behavior codes,including “validation” or “affection,” could notbe combined with other categories on the basis offactor analysis and so were excluded from thecurrent analyses.Relative frequencies (rf) were calculated by

dividing the number of occurrences of a behaviorcategory by the total number of behaviors. Forrelative duration (rd), we used total duration of abehavior divided by the total duration of theinteraction (e.g., Bakeman & Gottman, 1997;Johnson & Jacob, 1997). As a third measure,dichotomous variables indicating the occurrence ofa behavior (0= this person does not show a certainbehavior; 1= this person shows a certain behavior)were also used.Since homogeneity of variance represents an

important precondition for analysis of varianceand related procedures (Bortz, 1993), we trans-formed the data to stabilize variance across groups.

1 Results of the factor analysis are available from the first authorupon request.

This is particularly important for categories withrelatively few instances. We transformed the datafor relative frequency and relative duration byconverting the raw data into a 6-point Likert-typescale (0 to 5). Specifically, nonoccurrence wasassigned zero, a category was assigned “1” if itoccurred but did so at a frequency less than2⁎ (median/5), “2” if greater than “1” but lessthan 4⁎ (median/5), “3” if greater than “2” but lessthan 6⁎ (median/5), “4” if greater than “3” butless than 8⁎ (median/5) and “5” for all valuesgreater than “4.” As a result of this transformation,homogeneity of variance, as well as the stability ofthe results, was markedly improved.2

Discussion topic and gender of initiator. Prior tothe discussion, both partners independently rated alist of different topics indicating the intensity (4-point scale) of the strain produced by the topic inthe context of their marital relationship. This listcontained the following subject areas: finances,marital relationship (communication, sexuality,jealousy, violence), family (children, parents-in-law), leisure (friends, hobbies), consumption(drugs, alcohol, television), and religion. After thisprocedure, couples were instructed to discuss one ofthe most stressful discussion topics. In addition tointeraction coding, two coders rated the videotapedconflict discussion in terms of the gender of theperson initiating the issue and of the topic actuallydiscussed, assigning each discussion to one of the 6categories.

ResultsAnalysis of variance of relative duration (rd) andrelative frequency (rf) were conducted. First, a two-way mixed model multivariate/univariate analysisof variance (MANOVA/ANOVA) design, withgroup as between and gender as within (wife/husband) factors, was calculated. Because thedepressed person could be either the husband orthe wife, we also examined the within-subject factorof index person (i.e., patient vs. partner), in asecond set of two-way mixed MANOVA/ANOVAanalyses (between: four groups: within: patient/partner) (see also Christensen & Shenk, 1991;Kenny, Kashy, & Cook, 2006). Chi-square analy-ses for the occurrence/nonoccurrence of the behav-ior codes were conducted. In the case of directionalhypotheses, we utilized one-tailed significance tests.Repeated analyses using number of children as a

2More information about methodological aspects and results ofdata transformation are available from the first author uponrequest.

Table 2Means/Standard Deviations for Interaction Behavior by Group

Behavior codes G1 G2 G3 G4

rd rf rd rf rd rf rd rf

WivesEye contact 2.81/1.28 3.06/1.06 3.05/1.50 2.95/1.16 2.78/1.17 3.17/1.10 3.71/0.95 2.86/0.69Interest/curiosity 3.06/1.73 2.63/1.78 3.38/1.60 3.76/1.51 2.89/1.88 2.67/1.53 2.86/2.27 2.14/2.12Nonverbal positivity 2.50/1.51 2.19/1.17 3.71/1.79 3.14/1.85 2.22/1.67 2.28/1.64 3.86/1.95 3.43/1.81Emotional self-disclosure 3.56/1.71 3.38/1.46 3.43/1.36 3.57/1.25 3.39/1.46 3.06/1.51 3.14/1.35 3.14/1.46Interruption 3.19/1.47 2.76/1.61 3.39/1.54 3.00/1.91Aggression/defensiveness 2.19/2.01 2.50/2.10 0.48/0.98 0.52/1.21 1.83/1.76 2.00/1.75 1.71/2.36 1.86/2.41Criticism/domineering 3.06/2.14 3.00/2.10 1.19/1.72 1.38/2.06 1.94/2.10 1.89/2.17 1.57/2.37 1.57/2.37

HusbandsEye contact 2.88/1.36 2.69/1.20 2.86/1.15 3.33/1.20 2.61/1.15 2.83/1.20 3.14/1.57 2.86/1.68Interest/curiosity 2.31/1.49 2.75/1.61 3.10/1.70 3.29/1.52 3.11/1.64 3.33/1.72 2.86/2.19 3.14/2.34Nonverbal positivity 2.88/1.67 2.81/1.64 4.00/1.45 3.38/1.63 1.83/1.43 2.28/1.49 3.57/1.51 3.57/1.51Emotional self-disclosure 2.19/1.64 2.31/1.70 2.43/1.69 2.48/1.50 3.33/1.78 3.00/1.37 2.71/1.98 3.14/1.68Interruption 3.31/1.54 2.95/1.36 3.17/1.65 3.71/1.70Aggression/defensiveness 3.19/2.26 2.81/2.20 1.14/1.98 1.14/1.98 1.67/2.25 1.72/2.14 1.43/2.44 1.43/2.44Criticism/domineering 1.56/1.71 1.44/1.79 0.95/1.56 1.05/1.75 1.17/1.76 1.06/1.66 1.14/1.68 1.00/1.41

Notes. G1=distressed couples with a depressed wife. G2=nondistressed couples with a depressed wife. G3=distressed couples with adepressed husband. G4=nondistressed couples with a depressed husband.rd=relative duration. rf=relative frequencies.


covariate displayed the same pattern of significantresults, and so these results are not reported.Further, there were no significant effects of initiatoror topic and so these factors were not included insubsequent analyses.

gender and index person effects

For the relative duration (rd) data, there was asignificant multivariate effect (Hotteling's Trace) forgender, F(1, 58)=3.71, pb .01, as well as a signi-ficant multivariate effect for index person, F(6, 53)=2.10, pb .05. For relative frequency (rf) data, onlya significant multivariate effect for index person,F(6, 53)=1.86, pb .05, was found.

Table 3MANOVAs and ANOVAs: F-Values for Group, Gender, Index Person, a

Dependent variable group (3,58) gender (1

F F

rd rf rd

Multivariate effects 1.80 ⁎ 1.03 3.71 ⁎

Eye contact 0.76 0.33 1.53Interest/Curiosity 0.49 1.34 0.50Nonverbal positivity 6.00 ⁎ 2.51 ⁎ 0.00Emotional self-disclosure 0.66 0.19 5.15 ⁎

Interruption 0.47Aggression/defensiveness 3.63 ⁎ 3.24 ⁎ 2.02Criticism/domineering 1.99 1.33 6.77 ⁎

Note. rd=relative duration. rf= relative frequency.⁎ pb .05.

Follow-up ANOVAs revealed significant maineffects of gender for the relative duration ofemotional self-disclosure, F(1, 58)=5.15, pb .05,and for criticism/domineering, F(1, 58)=6.77,pb .01, with wives displaying higher levels thanhusbands (see Tables 2, 3). Significant main effectsof index person versus partner were found indicat-ing a lower relative frequency of interest/curiosity,F(1, 58)=3.93, pb .05, and higher values of aggres-sion/defensiveness [rd: F(1, 58)=4.29, pb .05; rf:Fn=(1, 58)=3.24, pb .05] among partners thanamong depressed persons. Depressed persons, onthe other hand, showed a lower duration ofnonverbal positivity, F(1, 58)=2.90, pb .05, and a

nd Group x Index Person

,58) index person(1,58)

group× indexperson (3,58)

F F

rf rd rf rd rf

1.77 2.10 ⁎ 1.86 ⁎ 1.62 ⁎ 1.040.16 0.76 0.17 0.70 0.961.67 1.20 3.93 ⁎ 0.30 0.821.20 2.90 ⁎ 0.62 0.02 0.543.37 ⁎ 2.25 3.04 ⁎ 1.82 1.330.85 0.04 0.680.17 4.29 ⁎ 3.24 ⁎ 0.65 0.177.45 ⁎ 0.22 0.17 3.78 ⁎ 3.76 ⁎


higher frequency of emotional self-disclosure, F(1,58)=3.04, pb .05, than their partners.

group effects

Significant multivariate group, F(18, 155)=1.80,pb .05 (Hotteling's Trace), and group-by-indexperson, F(18, 155)=1.62, pb .05, effects for relativeduration (rd) but not for relative frequency (rf) werefound. No significant multivariate effects wererevealed for group by gender [rd: F(18, 155)=1.06,ns; rf: F(21, 152)=1.00, ns] (see Tables 2, 3).ANOVAs with post hoc analyses to compare cells

of the design (Bonferroni, Dunnett's T3) were usedto explicate significant multivariate group differ-ences. The nondistressed couples (G2 and G4)displayed a significantly higher relative duration ofnonverbal positivity, F(3, 58)=6.00, pb .001, thandistressed couples with a depressed husband (G3),but were not significantly more positive thandistressed couples with a depressed wife. Further,a higher relative duration for aggression/defensive-ness for distressed (G1) than for nondistressed (G2)couples with a depressed wife was revealed, F(3,58)=3.63, pb .01, but no such differences wereobserved for couples with a depressed husband.The significant interaction effect of group withindex person for criticism/domineering, F(1, 58)=3.78, pb .01, was examined using univariateANOVAs to examine depressed persons and theirpartners separately. There were no group differ-ences in the relative duration of criticism/domineer-ing for nondepressed partners, F(3, 58)=0.95 (ns),but there were significantly higher values, F(3, 58)=4.15, pb .001, for distressed, depressed wives (G1)compared to the other three groups.Additional ANOVAs for gender effects within

each of the four groups were conducted. Only formaritally distressed couples with a depressed wife

Table 4Occurrence of Behavior by Group and Chi-square Value (df) for Group

Behavior codes G1 (n=16) G2 (n=21) G3 (

d p d p d

Eye contact 16 16 21 21 18Interest/curiosity 15 15 20 21 18Nonverbal positivity 15 15 20 20 16Emotional self-disclosure 15 14 21 18 17Interruption 16 16 19 20 18Aggression/defensiveness 11 12 5 7 8Criticism/domineering 13 9 8 7 8

Note. G1=distressed couples with a depressed wife. G2=nondistressedepressed husband. G4=nondistressed couples with a depressed husbd=depressed person. p=partner.⁎ pb .05.

(G1) were there significant multivariate effects [rd:F(6, 10)=17.71, pb .001; rf: F(7, 9)=4.47, pb .05]indicating a gender difference in interaction behav-ior between the depressed person and their partner.For distressed couples with a depressed wife (G1),significant effects were found indicating greateremotional self-disclosure [rd: F(1, 15) =7.59,pb .01; rf: F(1,15)=5.12, pb .05] and criticism/domineering [rd: F(1, 15)=11.25, pb .01; rf: F(1,15)=9.16, pb .01] for the depressed wives, whereashusbands displayed a higher relative duration ofaggression/defensiveness than their wives [G1: F(1,15)=3.87, pb .05].

occurrence of behavior

Occurrence of behavior was examined using chi-square tests for gender differences (Wilcoxon,McNemar), which indicated a significantly higheroccurrence of emotional self-disclosure in wives thanhusbands, χ2(1)=2.45, pb .01. The chi-square testfor index patient vs. partner, on the other hand,revealed significantly higher occurrence of aggres-sion/defensiveness for partners than patients,χ2(1)=2.89, pb .01.Chi-square tests (H-test Kruskal-Wallis) were

used to compare occurrence of behavior betweenthe four groups (see Table 4). There were signifi-cantly fewer persons displaying interest/curiosity inG4 (nondistressed, with depressed husband) com-pared to the other three groups, χ2(3)=7.09,pb .05, there was a higher occurrence of criticism/domineering, χ2(3)=8.12, pb .05, in G1 than ineither of the nondistressed groups, G2 (χ2(1)=7.82, pb .01) and G4 (χ2(1)=2.68, pb .05) as wellas of aggression/defensiveness in the two maritallydistressed groups, χ2(1)=16.63, pb .001.Analysis for patients and partners separately re-

vealed significantly higher occurrence of criticism/

Differences

n=18) G4 (n=7) group differences

overall depressed partner

p d p χ2(3) χ2(3) χ2(3)

18 7 7 1.95 0.00 0.0017 6 5 7.90 ⁎ 2.30 7.03 ⁎

16 7 7 2.44 1.21 1.2118 6 7 1.29 2.47 3.6518 7 7 5.95 3.97 1.9514 2 3 0.00 7.99 ⁎ 10.53 ⁎

10 3 3 8.12 ⁎ 7.62 ⁎ 2.66

d couples with a depressed wife. G3=distressed couples with aand.


domineering for depressed patients, χ2(3)=7.63,pb .05, in G1 than in the other groups and a loweroccurrence of interest/curiosity for partners in G4,χ2(3)=11.25, pb .01, relative to partners in othergroups. There was also more aggression/defensive-ness for both partners in the maritally distressedgroups [patient:χ2(3)=7.99, pb .05; partner:χ2(3)=10.53, pb .010] than for those in the maritallynondistressed groups (see Table 4).

discussion topic and initiator

The groups did not differ in terms of gender of theinitiator of the discussion, χ2(3)=2.29 (ns), as wellas of the topic, χ2(3)=0.42 (ns), actually discussed.Topics discussed were 14.5 % finances, 37.7%marital relationship, 21.3% family strain, 19.7%leisure and 6.6% consumption (drugs, alcohol,television). Fifty percent of topics were initiated bywives.As expected, strain due to the topic of “marital

problems” was rated higher in the two maritallydistressed groups, χ2(3)=33.19, pb .001. Addi-tionally, the conflict topic of “family stress” for G1,χ2(3)=6.53.19, pb .05, and the topic of “leisureactivity” for G3, χ2(3)=9.33, pb .05, were rated asconveying higher strain than in the other groups.

DiscussionIn the present study we investigated genderdifferences and patient vs. partner differences inmarital interaction in the context of depression andmarital distress by comparing distressed coupleswith a depressed wife (G1), nondistressed coupleswith a depressed wife (G2), distressed couples witha depressed husband (G3) and nondistressedcouples with a depressed husband (G4). This setof comparison groups allowed us to examinepatterns influenced by the intersection of gender,depression, and marital distress. The most strikingresult was that, in addition to substantial andanticipated main effects of gender, index personeffects (patient versus partner), and effects ofmarital distress, we also found group differencesdue to the gender of the depressed person.Depression was associated with different maritalinteraction patterns depending on the gender ofdepressed patient, and gender modified the effect ofmarital distress in the context of depression.Main effects of gender for higher emotional self-

disclosure and criticism/domineering in wives rep-licated prior work (e.g., Baucom et al., 1990;Dindia & Allen, 1992; Gottman, 1994), suggestingrobust gender differences on these behavioraldimensions. Further, partners of depressed personsdisplayed more aggression/defensiveness and a

higher duration of nonverbal positivity as well asa lower frequency of emotional self-disclosure andinterest/curiosity than did depressed persons them-selves, replicating prior work (Benazon & Coyne,2000; Coyne, 1976; Hautzinger et al., 1982) andsuggesting asymmetry in the interaction of de-pressed persons and their partners. The results alsoconfirmed gender- as well as depression-linkedpatterns of marital interaction, in which emotionalself-disclosure seems to be associated with gender(being female) as well as level of depression (beingdepressed).In keeping with our expectation that gender

differences would emerge in the context of maritaldistress and depression, the maritally distressedgroups displayed the highest level of negativebehavior for depressed wives and the lowest levelof positive behaviors for depressed husbands. In linewith results from community samples (Christensen& Heavey, 1990; Gottman, 1994), we foundgender-typical interaction behavior as well as aneffect of level of distress when distressed, depressedwives were compared to nondistressed, depressedwives. In particular, distressed couples with adepressed wife compared to nondistressed coupleswith a depressed wife demonstrated increasedgender-typical demanding/withdraw patterns(Christensen & Heavey, 1990; Gottman, 1994).Group differences in demanding/withdraw behav-ior were not attributable to differences in the genderof the person initiating the discussion nor to thetopic of discussion.Whereas wives' depression was associated with

an exaggeration of gender differences in the contextof marital distress, husbands' depression wasassociated with diminished gender differences.Couples with a depressed husband showed a similarlevel of demanding and emotional expression forboth partners regardless of level of marital distress,replicating the findings of Johnson and Jacobs(1997, 2000). Additionally, distressed couples inthe depressed husband group showed the lowestnonverbal positivity. The lack of positive behaviorsin couples with a depressed husband may be asnoteworthy as the presence of negative behavior inthe couples with a depressed wife (see Beach et al.,2008). Accordingly, the present results highlight thepossibility that depression may influence expressionof gender-typical patterns of interaction differentlydepending on the gender of the depressed partner.An important cautionary note is raised by group

comparison of demographic data on division ofhousehold labor and employment. Across indices ofcivil status, household labor and employment, themarital structure in the distressed, wife depressedgroup (G1) seems to be the most traditional.


However, whereas more traditional couples typi-cally display lower wife-demand/husband-with-draw patterns (Rehmann & Holtzworth-Munroe,2006), in the current sample, the most traditionalcouples were also those demonstrating the greatestdemand/withdraw patterns. Another potential cau-tionary note is that both maritally distressed groups(G1 and G3) had significantly more children, andthis may have contributed to greater stress for them.These factors may partially explain group differ-ences, as the association between marital problemsand depression in parents, especially wives, hasbeen shown to be related to a higher number ofchildren, traditional household labor as well aslower partner engagement and support in parenting(e.g., Brown & Harris, 1978; Feinberg, 2002;Gabriel & Bodenmann, 2006).Surprisingly, we observed very little stonewalling

or avoidance in the problem-solving discussions.Given prior research on these behaviors we hadexpected to see more (e.g., Gotlib &Whiffen, 1989;Gottman, 1994). One possible explanation may bethat partners were particularly invested in theirrelationships because they had volunteered toparticipate in a treatment study (agreement ofboth partners to participate in the interventionstudy as well as the interaction task was required).Accordingly, they might have been more open toproblem discussion and lower in avoidance thancouples drawn from the community who were notseeking treatment. Alternatively, it may be thatthere are cultural differences between American andSwiss couples, leading to lower utilization ofescalation behavior sequences ending in stone-walling in this sample. Or, it may be that therewere other, more subtle manifestations of avoid-ance that we did not detect with our codes (e.g.,Backenstrass, 1998; Benazon & Coyne, 2000).Interestingly, it was relative duration and not

frequency of the behavior codes that revealed thestrongest differences between groups. The utiliza-tion of relative duration increased the power of ourinvestigation without changing the patterning ofresults and so should be considered for futurebehavior observation studies as a way of potentiallydecreasing inconsistency in results across studies(Baucom et al., 2007). Also, our success instandardizing variance seems to be a usefulmethodological approach for dealing with morespecific and infrequently used behavior codes. Thishas the potential to expand the specificity ofexamination of interactive behavior in couplesand other family units. Additionally, the value ofusing more specific interaction codes rather thansummary codes of “positive” and “negative” for amore nuanced examination should be considered.

clinical implications

The current results suggest the importance ofgender-sensitive clinical models when interveningwith couples who have a depressed partner. Inparticular, the current data suggest that thepatterning of couple interaction will differ fordistressed couples with a depressed partner depend-ing on whether the depressed partner is male orfemale. This suggests that it will not be sufficient togeneralize interventions designed for depressedwives (e.g., Beach et al., 1990) when a depressedhusband is the identified patient. Rather, fordepressed husbands, it may be important in manycases to focus on greater affective expression ingeneral as a way to increase expression of positiveaffect. This appears to be a less common problemwhen the identified patient is the depressed wife.

caveats

Despite our efforts to utilize comprehensive andspecific coding systems, it must be noted that thecategories utilized were still rather broad and coarserelative to the nuances of dyadic interaction.Obviously, it will always be possible to wonder ifsome important nuances were overlooked in thecoding scheme utilized. It is also possible that sampleas well as culture-specific features may have influ-enced the results. Finally, because of the small samplesize of the nondistressed, depressed husband group(G4), it is also possible that some true differences(e.g., the between partner effect of frequency ofinterest/curiosity) may have been rendered nonsignif-icant in this group due to lower power relative toother groups rather than smaller effect size.

summary

Despite these concerns and caveats, however, in thepresent study we were able to identify differentinteraction patterns in couples with a depressedpartner associated with gender and marital distress.In particular, the present study found a lack ofpositivity in couples with a depressed husband,suggesting this may be a particularly importantdimension for couples with a depressed husband.We also found evidence of greater gender-basedasymmetry in the interaction of distressed, de-pressed patients when the depressed patient is thewife. These findings suggest the possible need fordifferent approaches to treatment depending on thegender of the depressed patient.

ReferencesAmerican Psychiatric Association. (1994). Diagnostic and

statistical manual of mental disorders (4th ed.). Author:Washington, DC.

Backenstrass, M. (1998). Depression und partnerschaftliche


Interaktion. [Depression andmarital interaction]. München;Berlin: Waxmann.

Bakeman, R., & Gottman, J. M. (1997).Observing interaction:An introduction to sequential analysis (2nd ed.). New York:Cambridge University Press.

Baucom, B., Eldridge, K., Jones, J., Sevier, M., Clements, M.,Markman, H., et al. (2007). Relative contribution of rela-tionship distress and depression to communication patterns incouples. Journal of Social and Clinical Psychology, 26,689–707.

Baucom, D. H., Notarius, C. I., Burnett, C. K., & Haefner, P.(1990). Gender differences and sex-role identity in marriage.In F. D. Fincham, & T. N. Bradbury (Eds.), The psychologyof marriage (pp. 150–171). New York: Guilford Press.

Beach, S. R. H., Jones, D. J., & Franklin, K. J. (2008). Marital,family, and interpersonal therapies for depression in adults.In I. H. Gotlib, & C. L Hammen (Eds.), Handbook ofdepression. New York: Guilford.

Beach, S. R. H., Sandeen, E. E., & O'Leary, K. D. (1990).Depression in marriage: A model for etiology and treatment.New York: Guilford.

Beck, A. T., & Steer, R. A. (1987). Beck Depression Inventory–Manual. The Psychological Corporation: San Antonio.

Beck, A. T., Steer, R. A., & Garbin, M. G. (1988). Psychometricproperties of the Beck Depression Inventory: Twenty-fiveyears of evaluation. Clinical Psychology Review, 8, 77–100.

Benazon, N. R., & Coyne, J. C. (2000). Living with a depressedspouse. Journal of Family Psychology, 14, 71–79.

Biglan, A., Hops, H., Sherman, L., Friedman, L. S., Arthur, J.,& Osteen, V. (1985). Problem solving interactions ofdepressed women and their husbands. Behavior Therapy,16, 431–451.

Bodenmann, G. (2006). Paartherapie bei depressiven Störungen[Couple therapy for depression]. In W. Lutz (Ed.), Problemein beziehungen und partnerschaft und ihre therapie. Einlehrbuch der paartherapie [Problems in social and maritalrelationships and their therapy]. München: Ernst ReinhardtVerlag.

Bortz, J. (1993). Statistik. [Statistics]. Berlin: Springer.Bourquard, E., Bodenmann, G., & Perrez, M. (1992-2005).

CAOS. Computer Aided Observation System. Fribourg:University of Fribourg.

Brown, G. W., & Harris, T. (1978). Social origins ofdepression: A reply. Psychological Medicine, 8, 577–588.

Christensen, A., & Heavey, C. L. (1990). Gender and socialstructure in demand-withdraw pattern of marital conflict.Journal of Personality and Social Psychology, 59, 73–81.

Christensen, A., & Shenk, J. L. (1991). Communication,conflict, and psychological distance in nondistressed, clinic,and divorcing couples. Journal of Consulting and ClinicalPsychology, 59, 458–463.

Cohen, J. (1960). A coefficient of agreement for nominal scales.Education and Psychological Measurement, 20, 37–46.

Coyne, J. C. (1976). Toward an interactional description ofdepression. Psychiatry, 39, 28–40.

Coyne, J. C., Thompson, R., & Palmer, S. C. (2002). Maritalquality, coping with conflicts, marital complaints, andaffection in couples with a depressed wife. Journal of FamilyPsychology, 16, 26–37.

Culp, L. N., & Beach, S. R. H. (1998). Marriage and depressivesymptoms. The role and base of self-esteem differ by gender.Psychology of Women Quarterly, 22, 647–663.

Davila, J., Karney, B. R., Hall, T. W., & Bradbury, T. N.(2003). Depressive symptoms and marital satisfaction:Within-subject associations and the moderating effects ofgender and neuroticism. Journal of Family Psychology, 17,557–570.

Dindia, K., & Allen, M. (1992). Sex differences in self-disclosure: A meta-analysis. Psychological Bulletin, 112,106–124.

Eldridge, K. A., & Christensen, A. (2002). Demand-withdrawcommunication during couple conflict: A review andanalysis. In P. Noller, & J. A. Feeney (Eds.), Understandingmarriage: Developments in the study of couple interaction(pp. 289–322). New York: Cambridge University Press.

Fincham, F. D., Beach, S. R.H., Harold, G. T.,&Osborne, L.N.(1997). Marital satisfaction and depression: Different causalrelationships for men and women? Psychological Science, 8,351–357.

Feinberg, M. E. (2002). Co-parenting and the transition toparenthood: A framework for prevention. Clinical Childand Family Psychology Review, 5, 173–195.

Gabriel, B., & Bodenmann, G. (2006). Befinden, Part-nerschafts- und Elternzufriedenheit bei Eltern im Zusam-menhang mit kindlichem Verhalten [Well being, marital andparenting satisfaction in parents in association with childbehavior]. Psychologie in Erziehung und Unterricht, 53,213–225.

Gotlib, I. H., & Whiffen, V. E. (1989). Depression and maritalfunctioning: An examination of specificity and genderdifferences. Journal of Abnormal Psychology, 98, 23–30.

Gottman, J. M. (1994). What predicts divorce? Hillsdale, NJ:Erlbaum.

Gottman, J. M., & Notarius, J. M. (2000). Decade review:Observing marital interaction. Journal of Marriage and theFamily, 62, 927–947.

Hahlweg, K. (1996). Fragebogen zur Partnerschaftsdiagnostik(PFB). [Partnership questionaire]. Hogrefe: Göttingen.

Hahlweg, K., Reisner, L., Kohli, G., Vollmer, M., Schindler, L.,& Revenstorf, D. (1986). Development and validity ofa new system to analyze interpersonal communication:Kategoriensystem fuer partnerschaftliche Interaktion. In K.Hahlweg, & N. S. Jacobson (Eds.), Marital interaction:Analysis and modification. New York: Guilford Press.

Hautzinger, M., Bailer, M., Worall, H., & Keller, F. (1995).Beck-Depressions-Inventar (BDI). Testhandbuch. [BeckDepression Inventory (BDI). Manual]. Huber: Bern.

Hautzinger, M., Linden, M., &Hoffman, N. (1982). Distressedcouples with and without a depressed partner: An analysis oftheir verbal interaction. Journal of Behavior Therapy andExperimental Psychiatry, 13, 307–314.

Jackman-Cram, S., Dobson, K. S., & Martin, R. (2006).Marital problem-solving behavior in depression and maritaldistress. Journal of Abnormal Psychology, 113, 380–384.

Jacob, T., & Leonard, K. (1992). Sequential analysis of maritalinteractions involving alcoholic, depressed, and nondistressedmen. Journal of Abnormal Psychology, 101, 647–656.

Johnson, S. L., & Jacob, T. (1997). Marital interactions ofdepressed men and women. Journal of Consulting andClinical Psychology, 65, 15–25.

Johnson, S. L., & Jacob, T. (2000). Sequential interactions inthe marital communication of depressed men and women.Journal of Consulting and Clinical Psychology, 68, 4–12.

Joiner, T., Coyne, J. C., & Blalock, J. (1999). On theinterpersonal nature of depression: Overview and synthesis.In T. Joiner, & J. C. Coyne (Eds.), The interactional natureof depression (pp. 3–19). Washington, DC: AmericanPsychological Association.

Kenny, D. A., Kashy, D. A., & Cook,W. L. (2006).Dyadic dataanalysis. New York: Guilford.

Klinetob, N. A., & Smith, D. A. (1996). Demand-withdrawcommunication in marital interaction: Tests of interspousalcontingency and gender role hypotheses. Journal ofMarriage and the Family, 58, 945–958.


Kurdek, L. A. (1998). The nature and predictors of thetrajectory of change in marital quality over the first 4years of marriage for first-married husbands and wives.Journal of Family Psychology, 12, 494–510.

Locke, H. J., & Wallace, E. M. (1959). Short maritaladjustment and prediction tests: Their reliability andvalidity. Marriage and Family Living, 21, 251–255.

Marchand, J. F., & Hock, E. (2000). Avoidance and attackingconflict-resolution strategies among married couples: Rela-tions to depressive symptoms and marital satisfaction.Family Relations, 49, 201–206.

McCabe, S. B., & Gotlib, I. H. (1993). Interactions of coupleswith and without a depressed spouse: Self-report andobservation of problem-solving situations. Journal of Socialand Personal Relationships, 10, 589–599.

Nelson, G. M., & Beach, S. R. H. (1990). Sequential interactionin depression: Effects of depressive behavior on spousalaggression. Behavior Therapy, 21, 167–182.

Nolen-Hoeksema, S. (1990). Sex differences in depression. PaloAlto, CA: Stanford University Press.

Rehmann, U. S., & Holtzworth-Munroe, A. (2006). A cross-cultural analysis of the demand-withdraw marital interac-tion: Observing couples from a developing country. Journalof Consulting and Clinical Psychology, 74, 755–766.

Schindler, L., Hahlweg, K., & Revenstorf, D. (1998).Partnerschaftsprobleme: Diagnose und Therapie. Hand-

buch für den Therapeuten. [Marital problems: diagnosisand therapy]. Berlin: Springer.

Schmaling, K. B., & Jacobson, N. S. (1990). Marital interactionand depression. Journal of Abnormal Psychology, 99,229–236.

Sher, T. G., Baucom, D. H., & Larus, J. M. (1990).Communication patterns and response to treatment amongdepressed and nondepressed maritally distressed couples.Journal of Family Psychology, 4, 63–79.

Whisman,M. A. (2001). Depression andmarital distress: Findingsfrom clinical and community studies. In S. R. H. Beach (Ed.),Marital and family process in depression (pp. 3–24). Washing-ton, DC: American Psychological Association.

Wilhelm, K., Roy, K., Mitchell, P., Brownhill, S., & Parker, G.(2002). Gender differences in depression risk and copingfactors in a clinical sample. Acta Psychiatrica Scandinavica,106, 45–53.

Wittchen, H. U., Zaudig, M., Schramm, E., Spengler, P.,Mombour, W., Klug, J., et al. (1990). StrukturiertesKlinisches Interview für DSM-III-R (SKID). [Structuredclinical interview for DSM-III-R]. Weinheim: Beltz TestGesellschaft.

RECEIVED: November 3, 2008ACCEPTED: September 12, 2009Available online 29 January 2010

Documents

Depression, Marital Satisfaction and Communication in Couples: Investigating Gender Differences