21
BEHAVIORTHERAPY 16, 431-451 (1985) Problem-Solving Interactions of Depressed Women and Their Husbands ANTHONY BIGLAN HYMAN HoPS LINDA SHERMAN LARRY S. FRIEDMAN JUDY ARTHUR VIRGINIA OSTEEN Oregon Research Institute The problem-solving interactions of normal marital dyads were compared with those of two groups of couples in which the wife was clinically depressed. One group was experiencing marital distress; in the second group marital distress was minimal. Fifty-two couples were recruited and videotaped as they attempted to solve two salient relationship problems in two lO-min interactions. The inter- actions were coded by observers trained in the LIFE Coding System which specifies both verbal and nonverbal behaviors. Multivariate analyses compared the diag- nostic groups on rates of behaviors and on the conditional responding of one spouse to the other. Depressed women exhibited significantly higher rates of depressive behavior than their husbands or either of the normal spouses. De- pressed women also displayed less problem-solving behavior than their husbands, while both husbands and wives in couples with a depressed wife exhibited less self-disclosure than normal dyads. Sequential analyses suggested that depressive behavior is functional in reducing the spouses' aversive behavior. Additionally, it was found that couples with a depressed wife who were experiencing marital distress exhibited less facilitative behavior than all the other couples and had a This research was primarily supported by NIMH Grant #MH34517. It was also supported in part by BRSG S07-RR05612. Portions of this paper were presented in: L. P. Rehm (Chair), Depression in the Context of the Family. Symposium conducted at the Annual Convention of the American Psychological Association, Toronto, Ontario (1984, August). The authors wish to express their thanks to Phil Warner, Nancy Oostenink, Deborah Toobert, and Lloyd Maxfield for their help in the development of the LIFE Coding System; to the project observers, Ruth Gibian, Candacc Holcomb, Alan Silverblatt, Scott Smith, and Tam- my Tengs, for their contributions; and to Barbara Britz for manuscript preparation. Requests for reprints should be sent to Anthony Biglan, Oregon Research Institute, 195 West 12th, Eugene, OR 97401. 4 31 0005-7894/85/0431-0451 $ 1.00/0 Copyright 1985 by Association for Advancement of Behavior Therapy All rights of reproduction in any form reserved.

Problem-solving interactions of depressed women and their husbands

Embed Size (px)

Citation preview

Page 1: Problem-solving interactions of depressed women and their husbands

BEHAVIOR THERAPY 16, 431-451 (1985)

Problem-Solving Interactions of Depressed Women and Their Husbands

ANTHONY BIGLAN

HYMAN HoPS

LINDA SHERMAN

LARRY S. FRIEDMAN

JUDY ARTHUR

VIRGINIA OSTEEN

Oregon Research Institute

The problem-solving interactions of normal marital dyads were compared with those of two groups of couples in which the wife was clinically depressed. One group was experiencing marital distress; in the second group marital distress was minimal. Fifty-two couples were recruited and videotaped as they attempted to solve two salient relationship problems in two lO-min interactions. The inter- actions were coded by observers trained in the LIFE Coding System which specifies both verbal and nonverbal behaviors. Multivariate analyses compared the diag- nostic groups on rates of behaviors and on the conditional responding of one spouse to the other. Depressed women exhibited significantly higher rates of depressive behavior than their husbands or either of the normal spouses. De- pressed women also displayed less problem-solving behavior than their husbands, while both husbands and wives in couples with a depressed wife exhibited less self-disclosure than normal dyads. Sequential analyses suggested that depressive behavior is functional in reducing the spouses' aversive behavior. Additionally, it was found that couples with a depressed wife who were experiencing marital distress exhibited less facilitative behavior than all the other couples and had a

This research was primarily supported by NIMH Grant #MH34517. It was also supported in part by BRSG S07-RR05612. Portions of this paper were presented in: L. P. Rehm (Chair), Depression in the Context of the Family. Symposium conducted at the Annual Convention of the American Psychological Association, Toronto, Ontario (1984, August). The authors wish to express their thanks to Phil Warner, Nancy Oostenink, Deborah Toobert, and Lloyd Maxfield for their help in the development of the LIFE Coding System; to the project observers, Ruth Gibian, Candacc Holcomb, Alan Silverblatt, Scott Smith, and Tam- my Tengs, for their contributions; and to Barbara Britz for manuscript preparation. Requests for reprints should be sent to Anthony Biglan, Oregon Research Institute, 195 West 12th, Eugene, OR 97401.

4 31 0005-7894/85/0431-0451 $ 1.00/0 Copyright 1985 by Association for Advancement of Behavior Therapy

All rights of reproduction in any form reserved.

Page 2: Problem-solving interactions of depressed women and their husbands

4 3 2 BIGLAN ET AL.

greater reciprocity of facilitative behavior than did normals. The findings under- score the importance of marital interactions in depression.

This paper compares the problem-solving interactions of depressed women and their husbands with those of normal dyads. Investigation of the marital interactions of depressed women is appropriate for several reasons. There is a growing tendency to view depression as occurring in an interactional context which affects the occurrence and resolution of depression (Coyne, 1976a, 1976b; Coyne, Kahn, & Gotlib, 1984; Lew- insohn, Hoberman, Teri, & Hautzinger, in press). Evidence from interview studies indicates that depressed women have problems in their relation- ships with their husbands (Weissman & Paykel, 1974). The outcome of depressive episodes appears to be worse when marital discord is unre- solved (McLean, Ogston, & Grauer, 1973; Rounsaville, Weissman, Pru- soft, & Herceq-Baron, 1979). Examination of marital interactions could prove fruitful for identifying critical depressive behaviors and could clarify how depressed women's interactions with their spouses affect their depres- sion.

Existing studies of dyadic interactions do not completely clarify the role of marital distress among depressed women. A study reported by Hautzinger, Linden, & Hoffman (1982) and by Linden, Hautzinger, & Hoffman (1983) showed that maritally distressed couples with a depressed member differ in their interactions from nondepressed maritally distressed dyads. However, they did not examine whether couples with depression and little or no marital difficulty differ from normals. In the present study, the interactions of couples in which the wife was depressed and marital distress was present were compared with couples in which the wife was depressed and there was little or no marital distress.

Conditional responding was examined in this study as well as the rate of each behavior. Although studies of maritally distressed couples have demonstrated the importance of analyses of conditional responding in marital interactions (Gottman, 1979; Margolin and Wampold, 1981; Re- venstorf, Vogel, Wegener, Hahlweg, & Schindler, 1980), analyses of con- ditional responding between depressed women and their husbands have not been done. In this study, the conditional responding of each spouse to the other was analyzed in an attempt to isolate the contingencies that each partner provided for the other's aversive and prosocial behaviors.

Coercive Processes and Depression

The present study is based on the application of coercion theory to depressive social behavior. Coercion theory has been developed, chiefly by Patterson (1982), as an account of the problematic interactive patterns that are found among families of aggressive children (Patterson & Reid, 1970) and among distressed marital dyads (Patterson & Hops, 1972). Coercion involves the control of another person's behavior through aver- sire stimulation. Person A emits an aversive behavior to which Person

Page 3: Problem-solving interactions of depressed women and their husbands

SPOUSAL INTERACTIONS OF DEPRESSED WOMEN 433

B yields. Person A's aversive behavior is reinforced by Person B's yielding and Person B's yielding is negatively reinforced by the cessation of A's aversive behavior. Evidence obtained thus far suggests that coercive be- havior is typically occasioned by the other person's aversive behavior (Gottman, 1979) and is reinforced by its cessation (Patterson, 1982).

Coercion theory prompted us to develop three major sets of hypotheses about depressive social behavior. The first set has to do with the nature of aversive behavior among depressed people and those with whom they interact. The second set focuses on the functional effects of depressive and aggressive behavior in the social relationships of depressed people. The third set addresses skill deficits that may characterize depressed peo- ple and their partners.

Depressive behavior as one form of coercion. There appears to be a class of behaviors that depressed people emit that, while not directly aggressive, are aversive to others. Studies that have shown that depressed people have a negative impact on others provide indirect evidence for this hy- pothesis (Coyne, 1976a; Strack & Coyne, 1983; Youngren & Lewinsohn, 1980). More direct evidence was provided by Gotlib and Robinson (1982) who found that mildly depressed people emitted more negative statements about themselves and more complaints about things other than the person being addressed. Linden, Hautzinger, and Hoffman (1983) found that depressed people emit higher rates of several types of behavior that are probably aversive: statements of negative well-being, negative self-eval- uations, and negative expectations about the future. Given these consid- erations, the present study included four types of verbal and one type of nonverbal behavior as "depressive." The verbal behaviors consisted of: (a) self-derogations, (b) physical and psychological complaints, (c) com- plaints about anything other than the self or the person being addressed, (d) and ignoring the initiation of the other. Nonverbal "dysphoric" be- havior included slowed speech, downcast eyes, sighing, and crying. It was predicted that the depressed women would exhibit the class of behaviors we designated as "depressive" more frequently than their spouses or members of normal couples.

Previous studies of marital interactions have not clearly distinguished depressive and more directly aggressive behavior, especially at the non- verbal level. For example, in Gottman's (1979) work on dysfunctional marriages, negative nonverbal behavior apparently included both dys- phoric and irritated affective responses. Both Hinchcliffe, Hooper, and Roberts (1978) and Hautzinger et al. (1982) found that negative nonverbal behavior occurred more frequently among depressed persons and their spouses than among dyads with no depression. However, they did not code aggressive and depressive nonverbal behavior separately. In the present study, we attempted to distinguish depressive and more purely aggressive behavior. A set of verbal and nonverbal aggressive behaviors was defined. The verbal responses we designated as aggressive included (a) humiliate/provoke, (b) disapprove, (c) threat, (d) argue, and (e) com- mand unaccountable (e.g., "From now on, tell me if you are going to be

Page 4: Problem-solving interactions of depressed women and their husbands

434 BIGLAN ET AL.

home late"). The nonverbal aspects of behavior that were coded as ag- gressive were ones that suggested irritated or angry affect. They included speaking in a loud voice, rolling the eyes, angry gestures such as pointing a finger at another, physically destructive behavior such as slamming objects on the floor, and obscene gestures. Considering the evidence that coercive exchanges are particularly likely to occur among maritally dis- tressed dyads (Gottman, 1979; Margolin & Wampold, 1981; Bitchier, Weiss, & Vincent, 1975; Hahlweg, Revenstorf, & Schindler, 1984) we predicted that aggressive behaviors would occur more in maritally dis- tressed dyads in which the wife was depressed than in normal dyads. No hypotheses were made about the rates of aggressive behavior in couples with a depressed wife but no marital distress.

The functional effects of depressive and aggressive behavior. A second major set of hypotheses have to do with the functional effects of depressive and aggressive behavior. Coercion theory suggests that these behaviors are reinforced by the cessation of other people's aversive responding and perhaps by others' positive behavior. Direct evidence regarding the func- tional effects of depressive behavior is not available; as noted above, relevant analyses of conditional responding between depressed people and others have not been done. Indirect evidence comes from the study by Hokanson, Sacco, Blumberg, and Landrum (1980) who found that de- pressed people in powerful positions in a prisoner's dilemma game acted in an exploitative manner while communicating self-devaluation and helplessness. Presumably, such self-depreciation reduces the probability of counterattack.

In the present study we examined whether the depressive behavior of depressed women reduced the probability of their husbands' aggressive behavior to a greater extent than was true for normal couples. Such a coercive relationship is more likely to occur in couples that also report marital distress. We also examined the functional effects of aggressive behavior to see whether it reduced the probability of the partner's de- pressive and aggressive behavior more in couples with a depressed wife than in other couples.

The study also examined the positive consequences for depressive and aggressive behavior. Higher rates of depressive behavior among depressed women could be due to higher densities of positive reinforcement for these behaviors. Positive reinforcement of depressive behavior may be particularly likely among couples in which there is depression but little marital distress. For these couples, there may be fewer aversive inter- changes than in couples with both depression and marital distress and hence less opportunity for the negative reinforcement of depressive be- havior.

Skill deficits. The third major hypothesis derived from the work on coercive family processes is that depressed women and their spouses lack or fail to exhibit skills necessary for terminating aversive exchanges and for maintaining prosocial behavior. At least three types of behavior might be considered skillful when a couple attempts to discuss disagreements

Page 5: Problem-solving interactions of depressed women and their husbands

SPOUSAL INTERACTIONS OF DEPRESSED WOMEN 435

and arrive at solutions. First, "facilitative" behaviors such as agree and approve may be skillful in two ways: (a) they reduce the probability that the other person will be aversive, and (b) when emitted contingent on the other person's prosocial behavior they positively reinforce such behavior. In the present study, the response class facilitative included the following verbal categories: (a) approval, (b) agree/affirm, (c) empathize/reassure, (d) accept responsibility, (e) humorous, and (f) positive statements about anything other than the person addressed. Nonverbal behaviors that in- dicated happy or caring affect (such as smiles, caresses, and a warm voice tone) were also coded as facilitative.

A second form of skillful behavior is self-disclosure. Statements about one's own feelings and experiences that are not negative in content may deflect attacks from the other. A third type of behavior that may contribute to effective discussions is the proposal of solutions (cf. Gottman, 1979; Margolin & Wampold, 1981). Coyne (1976b) has suggested that depressed people tend to shift the interactive burden to the person with whom they are interacting. Thus, we would predict that depressed women propose fewer solutions than their husbands.

METHODS Subjects

The subjects were 52 married couples who were recruited as part of a stUdy that compared the home interactions of depressed women and their families with the interactions of a matched group of normal control fam- ilies (Hops, Biglan, Sherman, Arthur, Friedman, & Osteen, 1985). Twen- ty-seven couples were recruited in which the wife met diagnostic criteria for depression and was currently receiving treatment for depression. The diagnostic criteria included a score of 18 or more on the Beck Depression Inventory (BDI; Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) and a diagnosis of major depressive disorder according to the Research Di- agnostic Criteria (RDC; Spitzer, Endicott, & Robins, 1978) following the Schedule of Affective Disorders and Schizophrenia (SADS; Endicott & Spitzer, 1978) interview. These couples were further categorized according to the existence of marital distress. Thirteen of the couples with a mean Dyadic Adjustment Scale (DAS; Spanier, 1976) score of 100 or greater were categorized as nondistressed. The remaining 14, with a mean DAS score of 98 or less were classified as depressed and maritally distressed.

Twenty-five couples were recruited in which the wife exhibited no psy- chological difficulties. To be included in this group, the woman had to have a BDI score of 10 or less and had to be free of any diagnosis according to the SADS/RDC criteria. In addition, she and her husband had to have a mean DAS score greater than 100. Normal couples and depressed cou- ples were matched on each of three dimensions: (a) the number of children, 0a) the number of children under seven, and (c) socioeconomic status as measured by the Hollingshead Four-Factor Index of Social Status (1975).

Page 6: Problem-solving interactions of depressed women and their husbands

4 3 6 BIGLAN ET AL.

The Coding System The LIFE (Living in Familial Environments) Coding System was de-

veloped for the study of the interactions of depressed women and their families (Arthur, Hops, & Biglan, 1982). The coding system was adopted from the Family Interaction Coding System (FICS; Reid, 1978) and the Marital Interaction Coding System (MICS; Hops, Wills, Patterson, & Weiss, 1972). The major modifications involved the addition of content codes which were felt to be characteristic of depressed people and the addition of seven codes for affective behavior. The coding system con- tained 31 content and 7 affect categories. For the purposes of the present study, these categories were collapsed on an a priori basis into seven categories: (a) depressive, (b) aggressive, (c) facilitative, (d) solution pro- posal, (e) self-disclosure, (f) elicit response, and (g) other. Table 1 presents each of the individual behaviors that were included in each of these categories.

The videotaped interactions were coded by seven observers using hand- held micro-computers (Observational Systems, Inc., Seattle, WA). The coder keyed in a five-digit number indicating the person behaving, the content of their behavior, the affect associated with the behavior, and the target of the behavior. The behavior of either spouse was recorded in real time, as it occurred. Changes in content or affect were recorded indepen- dently. The interactions in this study were coded simultaneously with 21 problem-solving interactions from a study of women with chronic pain (Biglan & Thorsen, 1985). Tapes from the two studies were randomly interspersed.

Procedures Subjects were recruited through newspaper advertisements and contacts

with physicians and other therapists. Interested women who called the Family Studies Project office were screened for marital status, number and age of children, and general psychological history. At the initial intake session, each spouse was asked to complete the BDI and the DAS. The couple was then asked to engage in two 10-min problem-solving discus- sions which were videotaped for later coding. The topics for the discus- sions were chosen on the basis of the couple's responses to the Dyadic Adjustment Scale. Three or more items were selected on which the couple had the greatest disagreement. The couple chose two for discussion.

RESULTS Preliminary Analyses

Differences among groups were examined on six variables that might have a bearing on the interpretation of behavioral differences among groups. Univariate analyses of variance were used. Table 2 presents these results. On the BDI, wives in the normal group were significantly less depressed than wives in the two types of depressed dyads. (All post hoc comparisons are based on p < .05.) Also, husbands in the normal group were significantly less depressed on the BDI than either depressed or

Page 7: Problem-solving interactions of depressed women and their husbands

SPOUSAL INTERACTIONS OF DEPRESSED WOMEN 437

TABLE 1 CORRELATIONS BETWEEN CALIBRATION AND RELIABILITY CODER FOR EACH OF THE

COLLAPSED LIFE CODES

Code categories included in the collapsed category

Reliability of the collapsed

category

Collapsed Hus- categories Wives bands

Complaint Ignore Self-derogation Physical/psychological complaint Any content code with dysphoric affect Command unaccountable Humiliate/provoke Disapprove Threat Argue Any other code with irritated affect Upper Humorous Approval Affirm Empathize/reassure Accept responsibility Any other code with happy or caring affect Propose solution Self-disclosure Elicit response Command Dependency Comply Indulge Physical positive Physical negative Worry Future task Positive self-statement Teach Attend Talk Unable to hear

Depressive 0.66 0.46

Aggressive 0.94 0.96

Facilitative 0.92 0.87

Propose solution 0.77 0.73 Self-disclosure 0.49 0.65 Elicit response 0.81 0.88 Other 0.56 0.61

dep re s sed /mar i t a l l y dis t ressed subjects . W i v e s in the dep re s sed /mar i t a l l y dis t ressed group were s ignif icant ly m o r e d is t ressed o n the D A S t h a n all o ther wives (p < .05). I n add i t i on , dep res sed -on ly wives were m o r e dis- t ressed t h a n wives in the n o r m a l dyads . H u s b a n d s in the depressed /

Page 8: Problem-solving interactions of depressed women and their husbands

4 3 8 B I G L A l x l E T A L .

m

z .<

[.., <

r/3

o z 0 g~

8 r~

M ,<

Z m

.<

Z m

z Z

~u

Z O

0

o

z 0

.<

Z m

~ ' ' ~

Page 9: Problem-solving interactions of depressed women and their husbands

SPOUSAL INTERACTIONS OF DEPRESSED WOMEN 439

maritally distressed group had significantly lower DAS scores than hus- bands in other groups, but no differences were found between husbands in the depressed-only group and the normal husbands. Depressed and depressed/distressed dyads were matched to normals on a measure of socioeconomic status (Hollingshead, 1975). Hence, these groups did not differ on this measure.

Reliabilities for Rates of Behavior To assess reliability, the sessions for 28 (of the 73) couples were in-

dependently coded by a second observer. Correlations were computed between the assigned observer and the reliability observer for the mean rate of each of the seven behaviors across the two topics for wives and husbands independently. As can be seen in Table 1, the reliabilities ranged from .46, for husbands' depressive behavior, to .96 for husbands' ag- gressive behavior. The mean reliability for husbands and wives was .74. Comparisons of the mean level of each behavior recorded by the two observers was made using t tests. Differences were found for the self- disclosures of wives, t = -2 .20 , p < .04, and husbands, t = -2 .30 , p < .03. The mean rates of the assigned calibrating observer were significantly higher. No hypotheses are presented to explain these differences.

Relationship Between Depressive and Aggressive Behavior In order to assess the discriminant validity of the depressive and ag-

gressive response categories, we examined the correlations between the rates of these types of behavior across couples. Table 3 presents corre- lations between the rates of depressive and aggressive behavior for the wives and husbands in each dyad. The rates of depressive behavior of the wives was not correlated with their rates of aggressive behavior. The same was true for the husbands. Nor was the rate of either partner's depressive behavior correlated with the rate of the other partner's ag- gressive behavior. However, the rates of aggressive behavior for the wives and husbands were highly correlated. The results were virtually identical when only data from the couples with a depressed wife were used.

Differences Among Groups in Rates of Behavior A 3 x 2 x 2 multivariate analysis of variance was performed on six of

the seven behavior categories. "Other" behavior was excluded to preclude the possibility of effects due to ipsativity. Diagnostic category (normals, depressed-only, and depressed/distressed) constituted the between-sub- jects factor. The two within-subjects factors were spouse (husband vs. wife) and topic (first vs. second).

A main effect was noted for groups, Wilks's lambda = .51, approximate F(12, 88) = 2.89, p = .002. Table 4 presents behaviors on which the three groups differed according to subsequent univariate analyses. Newman- Keuls post hoc tests (Kirk, 1968) were conducted to assess intergroup differences. The rate of self-disclosure was significantly different among the three groups; normal couples were significantly higher with virtually

Page 10: Problem-solving interactions of depressed women and their husbands

440 BIGLAN ET AL.

TABLE 3 CORRELATIONS BETWEEN RATES OF DEPRESSIVE AND AGGRESSIVE BEHAVIOR

Wives' Husbands' Husbands' aggressive depressive aggressive behavior behavior behavior

Wives' depressive behavior Wives' aggressive behavior Husbands' depressive behavior

.17 .20 .06 .18 .78***

.13

***p < .001.

identical scores for couples in the two clinical groups. The groups differed on the rate o f facilitative behavior; couples in which there was both depression and marital distress engaged in significantly less facilitative behavior than normal and depressed-only couples. Finally, differences on the rates o f aggressive behavior approached significance. Mean level o f aggressive behavior in normal dyads was lower than in the two clinical groups, though post hoc comparisons were not significant.

No main or interactive effects due to topic were found. However, there was a significant interaction between diagnostic group and the spouse factor, Wilks's lambda = .59, approximate F(12, 88) = 2.18, p = .019. Examinat ion o f the univariate Fs indicated that the interaction was due to depressive, F(2, 49) = 3.79, p = .029, and problem-solving behavior, F(2, 49) = 6.08, p = .004. Tests o f simple main effects (Kirk, 1968) showed that the mean rates o f depressive behavior o f the wives in the depressed- only group (M = 1.161) and in the depressed/distressed groups (M = 1.208) were significantly higher than those o f their husbands (M = .598 for depressed/distressed and M = .593 for depressed-only, respectively, p < .001) and the normal wives ( M = .690) and husbands (M = .648) (p < .05). Husbands in the depressed/distressed dyads displayed signifi-

TABLE 4 DIFFERENCES AMONG DIAGNOSTIC GROUPS ON MEAN RATES PER MIN OF THE BEHAVIORS

OF BOTH SPOUSES

D e -

Depressed pressed/ Univariate NO vs NO vs DO vs Normals only distressed F DO DD DD

Self-disclosure 1.48 1.01 1.00 4.70** * * Facilitative 1.98 1.65 1.05 10.39"** ** Aggressive 0.58 0.91 1.02 2.58~"

tP -< .10. * p -< . 0 5 .

**p -< .01. *** p -< .001.

Page 11: Problem-solving interactions of depressed women and their husbands

SPOUSAL INTERACTIONS OF DEPRESSED WOMEN 4 4 1

cantly more (p < .05) solution proposing behavior (M = .486) than did their wives (M = .272); a similar difference (M = .494 and M = .342 for husbands and wives, respectively) for the depressed-only dyads ap- proached significance (19 < . 10). However, among the normal couples, there was a nonsignificant tendency (p < . 10) for wives to propose more solutions than their husbands (M = .489 and M = .380, for wives and husbands, respectively).

Analysis of Conditional Responding We compared the three groups on differences in their conditional re-

sponding to each other, using the z-score statistic recommended by A1- lison and Liker (1982) and Gottman (1980). The z-scores provide a nor- malized measure of the difference between the conditional responding of one person to the other in comparison with the former's unconditional responding. For example, we were interested in the degree to which the husband's aggressive response to the wife depended upon whether she exhibited prior depressive behavior. The z-score for this relationship is

Z ~- p(H Aggressive/W Depressive) - p(H Aggressive)

(H Aggressive)(1 - p[H Aggressive])(1 - p[W Depressive])

where n k

(n - K)p(W Depressive)

= the number of behaviors in the sequence, = the number of events between the wife's antecedent behavior

and the husband's response, i.e., the lag.

The z-scores were calculated for each couple and then used in multivariate analyses of variance to examine between-group effects.

Three sets of analyses were done on husbands' behavior given wives' behavior and three for wives' behavior given husbands' behavior. One set tested the hypothesis, derived from coercion theory, that among cou- ples with a depressed woman, depressive and aggressive behavior of one spouse would produce a decrement in the depressive or aggressive be- havior of the other spouse. Thus, in one MANOVA we compared the diagnostic groups on four z-scores: (a) husband aggressive given wife depressive, (b) husband depressive given wife depressive, (c) husband aggressive given wife aggressive, and (d) husband depressive given wife aggressive. And, in the parallel analysis we examined group differences on the z-scores for the wife's behavior given the husband's behavior. A second pair of MANOVAs--one for husband given wife and one for wife given husband--tested whether the groups differed in the positive con- sequences for depressive and aggressive behavior. The four dependent variables in these analyses corresponded to facilitative behavior and prob- lem-solving behavior given depressive or aggressive behavior of the spouse. A third pair of MANOVAs examined group differences in the response of each spouse to the facilitative behavior of the other. The

Page 12: Problem-solving interactions of depressed women and their husbands

442 BIGLAN ET AL.

TABLE 5 RELIABILITIES OF z-SCORES

Husband given wife Wife given husband

r t p r t p

Aggressive given depressive .74 - 1.27 Depressive given depressive .29 - . 2 0 Aggressive given aggressive .22 .57 Depressive given aggressive .67 - . 1 8 Facilitative given depressive .34 .47 Propose solution given depressive .40 - 1.12 Facilitative given aggressive .50 - 2 . 3 6 Propose solution given aggressive .60 - . 19 Depressive given facilitative .30 - . 4 8 Aggressive given facilitative .37 - 2 . 4 6 Facilitative given facilitative .55 - 1.90

n.s. .36 .72 n.s. n.s. .44 - . 7 7 n.s. n.s. .48 - 1.26 n.s. n.s. .69 .75 n.s. n.s. .38 .54 n.s. n.s. .23 .57 n.s. .03 .33 .40 n.s. n.s. .37 - . 6 7 n.s. n.s. .21 - . 6 2 n.s. .02 .64 - 1.57 n.s. n.s. .65 -1 .61 n.s.

dependent variables were the z-scores corresponding to depressive, ag- gressive, and facilitative, given facilitative behavior.

Reliabilities o f z-scores. The reliabilities of the z-scores were assessed following Gottman's (1980) recommendations. In this approach, z-scores were computed for the data provided by the reliability observers and these z-scores were compared with those obtained from the first observer. The scores were correlated with each other and t tests were performed to see if they differed in mean level. Table 5 presents the results of these analyses. On the whole, the reliabilities were modest. They ranged from .21 to .74, with a mean of .44. Differences in mean level were found on two of the z-scores.

Husband's conditional responding to wives. Table 6 presents results of the analyses of conditional responding for those z-scores that differed significantly among groups. There was a significant difference among groups in the husbands' conditional aversive responding to the wives" depressive and aggressive behavior, Wilks's lambda = .67, approximate F(8, 92) = 2.57, p = .014. As shown in Table 6, the groups differed in the degree to which the wives' depressive behavior reduced the probability of subse- quent aggressive behavior of the husband. Couples with marital distress and depression were significantly different from depressed-only and nor- mal couples. The wife's depressive behavior was more likely to reduce the probability of the husband's aggressive behavior among the depressed/ distressed couples than among the other couples. The groups also differed on the z-score for husbands' depressive behavior given wives' aggressive behavior, but only at p < . 10. Post hoc comparisons did not indicate that any of the groups differed significantly.

The MANOVA for the group differences on z-scores for husbands' facilitative and their proposal of solutions given the wives' depressive and aggressive behavior was also significant, Wilks's lambda = .71, approxi- mate F(8, 92) = 2.17, p -- .037. The only significant univariate F was for

Page 13: Problem-solving interactions of depressed women and their husbands

S P O U S A L I N T E R A C T I O N S O F D E P R E S S E D W O M E N 443

Z

Z

.1 < Z

Z 0

, e

O

D O

O

@

<

z

8 tl- tl. t~

~ O ~

I I I

' T ' T '

I I I ~ - I I

~.~ • ,~ m~-

a.~ a.~

e~

9

i¢~ o I'N

o. ~. o. ? , ?

I I I

I T I

!i' II II V V V

Page 14: Problem-solving interactions of depressed women and their husbands

4 4 4 BIGLAN ET AL.

husbands' proposal of solutions given wives" aggressive behavior. Post hoc analyses indicated that among depressed/distressed couples wives' aggressive behavior reduced the probability of the husbands" proposal of solutions to a significantly greater extent than was true among normal couples.

The MANOVA for the husbands' responding to their wives' facilitative behavior was significant, Wilks's lambda = .76, approximate F(6, 94) = 2.28,/7 = .043. The groups differed on two z-scores. One was the prob- ability of husbands' facilitative behavior given the facilitative behavior of the wife. Husbands in the depressed/distressed dyads were more likely to respond to their wives' facilitative behavior with facilitative behavior than were normal couples. The groups also differed on husbands" de- pressive behavior given wives' facilitative behavior. The depressed-only couples' z-scores indicated that the husbands in these groups were more likely to be depressive in response to their wives' facilitative behavior than was true for normal couples.

Wives" conditional responding to husbands. The MANOVA for wives' depressive and aggressive responding to husbands' depressive and ag- gressive behavior only approached significance, Wilks's lambda = .76, approximate F(8, 92) = 1.69, 17 = . 11. The probability of wives' depres- sive responding was reduced by the husbands' aggressive behavior to a significantly greater extent in the depressed/distressed couples than in normal or depressed-only couples.

The MANOVA for positive consequences to depressive and aggressive behavior was not significant.

The MANOVA for wives' responding to husbands' facilitative behavior was significant, Wilks's lambda = .75, approximate F(6, 94) = 2.47, p = .03. The groups differed on two z-scores. As with husbands' responding to wives the depressed/distressed couples had greater reciprocity of fa- cilitative behavior than did normals. In addition, depressed/distressed wives' depressive behavior was reduced in probability following the hus- bands' facilitative behavior to greater extent than was true for normals.

DISCUSSION The study has identified a number of ways in which couples with a

depressed wife differ from normal couples. Regardless of their level of marital distress, the wives in these couples exhibit more depressive be- havior than do normals. Both the depressed woman and her spouse self- disclose less than do normal partners. In normal couples, the wife tends to offer more solutions than her husband, but this pattern is reversed when the wife is depressed, especially when there is marital distress. In considering these findings, it should be remembered that we cannot en- tirely separate effects associated with depression from those associated with marital distress. The couples we have labeled as "'depressed-only" were not identical to normals in the levels of their marital distress; the husbands and wives in these dyads had significantly lower DAS scores than normal husbands and wives.

Page 15: Problem-solving interactions of depressed women and their husbands

SPOUSAL INTERACTIONS OF DEPRESSED WOMEN 445

Depressive and Aggressive Behavior As predicted, depressed women exhibit a class of behavior that we had

labeled "depressive." Included in this class are self-derogations, com- plaints about physical and psychological functioning, complaints about things other than the person being addressed, and dysphoric nonverbal behavior such as frowns, sighs, slowed speech, and downcast eyes. Our labeling of the behaviors in this category as "depressive" is empirically supported by the fact that these behaviors were exhibited at an elevated rate only by the depressed women. That depressive behavior is discrim- inable from aggressive behavior was indicated by the fact that the rates of the two types of behavior are not correlated. That these behaviors are aversive to others is supported by research indicating that behavior of this sort produces negative reactions from others when it is exhibited by confederates who are role-playing (Howes and Hokanson, 1979; Hammen and Peters, 1978). The present study does not indicate the extent to which the individual behaviors that were included in this class do, in fact, con- verge. Nor is it clear which individual behaviors contributed most to the differences that were found. Finally, these results do not demonstrate that the class of behavior we have labelled as depressive is unique to women who are depressed.

There was a tendency for couples with depression (especially the de- pressed/distressed) to have higher rates of aggressive behavior. This was true for both the depressed woman and her husband. Though differences among diagnostic groups were not strong, they are consistent with the notion derived from coercion theory that the interactional context for depressed women involves directly aggressive behavior. Other studies that have compared depressed dyads with normals have found evidence of higher rates of various kinds of aggressive behavior in the depressed dyads (Hinchcliffe et al., 1978; Hautzinger et al., 1982). Arkowitz, Hol- liday, and Hurter (1982) found that the husbands of depressed women exhibited more negative nonverbal behavior but this was also true for husbands of psychiatric controls. In comparing maritally distressed and nondistressed couples, some studies have found differences in the rates of such behavior (Koren, Carlton, & Shaw, 1980; Haynes, Follingstad, & Sullivan, 1979; Birchler et al., 1975) while others have not (Margolin & Wampold, 1981).

Consequences of Depressive and Aggressive Behavior Analysis of the conditional responding to depressive behavior sup-

ported the coercion theory hypothesis that depressive behavior is func- tional in reducing the aversive behavior of others. However, this was true only for couples in which there was also marital distress. Among these couples, the wife's depressive behavior produced greater reductions in the aggressive behavior of her husband than was true for other couples. Additional evidence of aversive control techniques was found for the maritally distressed group. The husband's aggressive behavior reduced the likelihood of the wife's depressive behavior more than was true in

Page 16: Problem-solving interactions of depressed women and their husbands

4 4 6 BIGLAN ET AL.

depressed-only and normal couples. These findings underscore the im- portance of aversive control for understanding the behavior of depressed women who are also maritally distressed. The woman's depressive and aggressive behavior is one means of reducing the aversive behavior of her husband. He, in turn, can achieve brief respite from her depressive behavior by displaying aggressive behavior. Neither partner achieves long- term benefits; the rates of his aggressive behavior and her depressive behavior are elevated. But brief-though-immediate reductions in the probability of aversive events are a powerful reinforcer (Hineline, 1970). It may be that effective interventions with couples where there is both depression and marital distress require that the couples be taught alter- native means of coping with each other's aversive behavior (cf. Jacobson, 1984).

It should be noted that analyses of conditional responding do not pro- vide conclusive evidence that functional relationships exist. Experimental manipulations are needed in order to be sure that depressive and aggres- sive behavior of one partner is functional in reducing the other's aversive behavior and to be sure that such reductions do function to reinforce the first person's depressive or aggressive behavior.

Coercive control was not evident among depressed-only couples. Al- though women who were depressed and not maritally distressed exhibited as much depressive behavior as depressed women in the more martially distressed dyads, there was no evidence that their depressive behavior functioned to reduce their husbands' aversive behavior. Nor was there evidence that either partner's aggressive behavior was achieving reduc- tions of the other's aversive behavior. Moreover, there was no evidence that depressive and aggressive behaviors were attaining positive conse- quences. The question arises, therefore, whether the coercive processes that were identified among depressed/distressed couples are a function of marital distress rather than depression. That question cannot be answered by the design of the present study. The study that was reported by Haut- zinger et al. (1982) and by Linden et al. (1983) did demonstrate that couples with both depression and marital distress had more dysfunctional interactions than maritally distressed couples in which there was no depression. However, conditional responding was not assessed in that study. It seems likely that the aversive exchanges that are so typical of maritally distressed dyads (Patterson, Hops, & Weiss, 1975; Gottman, 1979; Margolin & Wampold, 1981) constitute one circumstance in which depressive behavior will be maintained. But the contingencies for de- pressive behavior in dyads that are not maritally dysfunctional are yet to be identified.

Skills Deficits Depressed/distressed couples appear to lack skills for coping with their

partners' aversive behavior. They emitted facilitative behavior at a lower rate than depressed-only or normal couples. When they were facilitative it was more likely to be contingent on the other person's facilitative

Page 17: Problem-solving interactions of depressed women and their husbands

SPOUSAL INTERACTIONS OF DEPRESSED WOMEN 447

behavior than was true for normal couples. This greater reciprocity of facilitative behavior is similar to Gottman's (1979) finding that positive behavior was reciprocated more among maritally distressed couples than among normals. Thus, each partner would respond positively only to the most positive behavior of the other. In addition, for depressed/distressed couples, when husbands were facilitative their wives' depressive behavior was less likely. Yet these couples exhibited less facilitative behavior than did normals. Thus, despite the fact that husbands' facilitative behavior could reduce their wives' depressive behavior, depressive behavior was occurring at an elevated rate, and behavior that could have reduced it did not occur. In this context, the lack of such positive behavior can be seen as a deficit in the skill of interrupting aversive interchanges.

The fact that facilitative behavior was associated with a reduced prob- ability of contingent aversive behavior, yet was not occurring frequently among depressed/distressed dyads is surprising. One might have expected that such contingencies would have negatively reinforced these prosocial behaviors. For these couples, it appears that the reduction of aversive behavior did not reinforce the other's facilitative behavior, but did rein- force aversive behavior. A review of the literature on aversive control suggests that reductions in the rate of the other's aversive behavior pro- duced by facilitative behavior may not be as reinforcing as the reductions that are produced by depressive and aggressive behaviors (Lewin & Biglan, 1985). This is because the latter forms of behavior are a direct attack on the other in addition to being effective in reducing the other's aversive attacks.

The reduced level of facilitative behavior among depressed/distressed dyads is consistent with two studies of maritally distressed couples (Koren et al., 1980; Margolin & Wampold, 1981). However, since levels of depres- sion were not assessed, the findings may have been due to the presence of depression as well as marital distress. Further studies of marital distress need to examine the association between depression and interactive pat- terns.

The interactive burden in dyads with depression. Regardless of the ex- istence of marital distress, when the wife is depressed the focus of the interaction appears to be on her. The pattern is in keeping with Coyne's (1976b) prediction that the interactive burden in such relationships is shifted to the nondepressed person. There are two ways in which this was evident in the present study.

First, normal couples self-disclosed more than couples in which the wife was depressed. Much of the depressive behavior included behavior that would be classed as self-disclosure if it were not for its negative content (e.g., self-derogations and physical/psychological complaints). It appears then that in the dyads in which the wife is depressed, a good deal of the discussion is focused on her negative well-being while relatively little attention is focused on the husband. Hautzinger et al. (1982) obtained similar results.

Second, although much of the focus is on the depressed wife and her

Page 18: Problem-solving interactions of depressed women and their husbands

448 raGLAN ET AL.

complaints, she contributes relatively little to attempts to arrive at so- lutions to disagreements with her husband. Wives in the normal group proposed slightly (though nonsignificantly) more solutions than did their husbands. However in the depressed/distressed group, wives proposed significantly fewer solutions than their husbands. The same tendency was seen among depressed-only couples. This pattern is consistent with Coyne's prediction that the interactive burden falls on the nondepressed spouse. The result is in contrast to studies that have compared maritally distressed and nondistressed couples. Margolin and Wampold (1981) found that both the husbands and wives in such couples proposed fewer solutions than did the nondistressed husbands and wives. Gottman (1979) and Koren et al. (1980) found that distressed and nondistressed couples did not differ in their rates of proposing solutions. However, no measures of depression were obtained in those studies. The present finding suggests that the pattern of proposing solutions may depend on whether one of the partners is depressed.

Limitations of the Study At least four limitations of this study should be noted. First, the study

did not include a control group of dyads in which the wife was psycho- logically distressed though not depressed. Arkowitz et al. (1982) included such a group and found that some of their results were not unique to dyads containing a depressed wife. That may be true for some of the present findings.

Second, the present study did not include a group of dyads in which there was marital distress but no depression. We had difficulties recruiting such couples. This group would have allowed us to determine the extent to which the differences found between couples with marital distress and depression, and normal couples were due to marital distress rather than depression.

Third, the reliabilities of the collapsed codes and z-scores were highly variable. Relatively low reliabilities on several variables of interest lowers the probability of'detecting relationships in the data.

A fourth concern is with the precise procedure by which conditional and unconditional probabilities of behavior are calculated. In the present study, the conditional probability of one person's behavior given a be- havior of the other was computed based on all of the behavior of each person. I fa wife emitted three depressive behaviors in a row, the husband was considered to have had three opportunities to respond to her behavior. An alternative procedure would have been to only consider those re- sponses of the wife which were followed by some sort of response by the husband. We chose the former procedure because it seemed to more accurately reflect the contingencies that the wife was actually experiencing. This procedure tended to make the conditional probabilities smaller than the unconditional probabilities and the z-scores more negative.

A final concern is with the z-score statistic. Differences among groups in the z-scores may have been due to differences in the marginal distri-

Page 19: Problem-solving interactions of depressed women and their husbands

SPOUSAL INTERACTIONS OF DEPRESSED WOMEN 449

b u t i o n s o f b e h a v i o r s o r in t h e i r a u t o d e p e n d e n c e ( G a r d n e r & H a r t m a n n , 1984; G a r d n e r , H a r t m a n n , & Mi t che l l , 1982). S tud ie s a re n e e d e d to c o m - pa re the v a r i e t y o f m e t h o d s o f c o m p u t i n g a n d a n a l y z i n g c o n d i t i o n a l re- l a t i o n s h i p s t ha t a re in c u r r e n t use ( A l l i s o n & Liker , 1982; G o t t m a n , 1979; G o t t m a n , N o t a r i u s , M a r k m a n , Bank , Y o p p i , & R u b i n , 1976; M a r g o l i n & W a m p o l d , 1981; R e v e n s t o r f e t al . , 1980).

REFERENCES Allison, P. D., & Liker, J. K. (1982). Analyzing sequential categorical data on dyadic

interaction: A comment on Gottman. Psychological Bulletin, 91, 393-403. Arkowitz, H., Holliday, S., & Hutter, M. (1982, November). Depressed women and their

husbands: A study of marital interaction and adjustment. Paper presented at the annual meeting of the Association for Advancement of Behavior Therapy, Los Angeles. CA.

Arthur, J. A., Hops, H., & Biglan, A. (1982). LIFE (Living in familial environments) coding system. Unpublished manuscript, Oregon Research Institute, Eugene, OR.

Beck, A. T., Ward, C. H., Mendelson, M., Mock, J. E., & Erbangh, J. K. (1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561-571.

Biglan, A., & Thorsen, C. (1985). The interactive behavior of women with chronic pain. Unpublished manuscript, Oregon Research Institute, Eugene, OR.

Birehler, G. R., Weiss, R. L., & Vincent, J. P. (1975). Multimethod analysis of social reinforcement exchange between maritally distressed and nondistressed spouse and stranger dyads. Journal of Personality and Social Psychology, 31, 349-360.

Coyne, J. C. (1976a). Depression and the response of others. Journal of Abnormal Psy- chology, 85, 186-193.

Coyne, J. C. (1976b). Toward an interactional description of depression. Psychiatry, 39, 28-40.

Coyne, J. C., Kahn, J., & Gotlib, I. H. (1984). Depression. In T. Jacob (Ed.), Family interaction and psychopathology. New York: Plenum.

Endicott, J., & Spitzer, R. (1978). A diagnostic interview: The schedule for affective disorders and schizophrenia. Archives of General Psychiatry, 35, 837-853.

Gardner, W., & Hartman, D. P. (1984). On Markov dependence in the analysis of social interaction. Behavioral Assessment, 6, 229-236.

Gardner, W., Hartmann, D. P., & Mitchell, C. (1982). The effects of serial dependence on the use of X 2 for analyzing sequential data. Behavioral Assessment, 4, 75-82.

Gotlib, I. H., & Robinson, L.A. (1982). Responses to depressed individuals: Discrepancies between self-report and observer-related behavior. Journal of Abnormal Psychology, 91, 231-240.

Gottman, J. M. (1979). Marital interaction: Experimental investigations. New York: Ac- ademic Press.

Gottman, J. M. (1980). Analyzing for sequential connection and assessing interobserver reliability for the sequential analysis of observational data. Behavioral Assessment, 2, 361-368.

Gottman, J. M., Notarius, C., Markman, H., Bank, S., Yoppi, B., & Rubin, M. E. (1976). Behavior exchange theory and marital decision-making. Journal of Personality and Social Psychology, 34, 14-23.

Hahlweg, K., Revenstorf, D., & Schindler, L. (1984). Effects of behavioral marital therapy on couples" communication and problem-solving skills. Journal of Consulting and Clin- ical Psychology, 52, 553-566.

Hammen, C. L., & Peters, S. D. (1978). Interpersonal consequences of depression: Re-

Page 20: Problem-solving interactions of depressed women and their husbands

450 BIGLAN ET AL.

sponses to men and women enacting a depressed role. Journal of Abnormal Psychology, 87, 322-332.

Hautzinger, M., Linden, M., & Hoffman, N. (1982). Distressed couples with and without a depressed partner: An analysis of their verbal interaction. Journal of Behavior Therapy and Experimental Psychiatry, 13, 307-314.

Haynes, S. N., FoUingstad, D. R., & Sullivan, J. C. (1979). Assessment of marital satis- faction and interaction. Journal of Consulting and Clinical Psychology, 47, 789-791.

Hinehcliffe, M., Hooper, D., & Roberts, F.J. (1978). The melancholy marriage. New York: Wiley.

Hineline, P. N. (1970). Negative reinforcement without shock reduction. Journal of the Experimental Analysis of Behavior, 14, 259-268.

Hokanson, J. E., Sacco, W. P., Blumberg, S. R., & I.andrum, G. C. (1980). Interpersonal behavior of depressive individuals in a mixed-motive game. Journal of Abnormal Psychology, 89, 320-332.

Hollingshead, A. B. (1975). Four factor index of social status. Unpublished manuscript. (Available from A. B. Hollingshead, P.O. Box 1965, Yale Station, New Haven, CT 06520.)

Hops, H., Biglan, A., Sherman, L., Arthur, J., Friedman, L. S., & Osteen, V. (1985). Home observations of family interactions of depressed women. Submitted for publication.

Hops, H., Wills, T., Patterson, G. R., & Weiss, R. L. (1972). Marital interaction coding system (MICS). University of Oregon and Oregon Research Institute (NAPS Document #02077). Eugene, OR.

Howes, M. J., & Hokanson, J.E. (1979). Conversational and social responses to depressive interpersonal behavior. Journal of Abnormal Psychology, 88, 625-634.

Jacobson, N. S. (1984). Marital therapy and the cognitive behavioral treatment of depres- sion. the Behavior Therapist, 7, 143-147.

Kirk, R. E. (1968). Experimental design: Procedures for the behavioral sciences. Belmont, CA: Brooks/Cole Publishing Co.

Koren, P., Carlton, K., & Shaw, D. (1980). Marital conflict: Relations among behaviors, outcomes, and distress. Journal of Consulting and Clinical Psychology, 48, 460--468.

Lewin, L., & Biglan, A. (1985). The implications of experimental studies of aversive control for the study offamily interactions. Unpublished manuscript, Oregon Research Institute, Eugene, OR.

Lewinsohn, P. M., Hoberman, H., Teri, L., & Hautzinger, M. (in press). An integrative theory of depression. In S. Peiss & R. Bootzin (Eds.), Theoretical issues in behavior therapy. New York: Academic Press.

Linden, M., Hautzinger, M., & Hoffman, N. (1983). Discriminant analysis of depressive interactions. Behavior Modification, 7, 403--422.

Margolin, G. & Wampold, B. E. (1981). Sequential analysis of conflict and accord in distressed and nondistressed marital partners. Journal of Consulting and Clinical Psy- chology, 49, 554-567.

McLean, P. D., Ogston, K., & Grauer, L. (1973). A behavioural approach to the treatment of depression. Journal of Behaviour Therapy and Experimental Psychiatry, 4, 323-330.

Patterson, G. R. (1982). A social learning approach to family intervention: Vol. 3. Coercive family process. Eugene, OR: Castalia Publishing.

Patterson, G. R., & Hops, H. (1972). Coercion, a game for two: Intervention techniques for marital conflict. In R. E. Ulrich & P. Mountjoy (Eds.), The experimental analysis of social behavior. New York: Appleton-Century-Crofts.

Patterson, G. R., Hops, H., & Weiss, R. L. (1975, May). Interpersonal skills training for couples in early stages of conflict. Journal of Marriage and the Family, 37, 295-303.

Patterson, G. R., & Reid, J. B. (1970). Reciprocity and coercion: Two facets of social

Page 21: Problem-solving interactions of depressed women and their husbands

SPOUSAL INTERACTIONS OF DEPRESSED WOMEN 451

systems. In C. Neuringer and J. Michael (Eds.), Behavior modification in clinical psy- chology. New York: Appleton Century Crofts.

Reid, J. B. (Ed.). (1978). A social learning approach to family intervention: Vol. 2. Obser- vation in home settings. Eugene, OR: Castalia Publishing.

Revenstorf, D., Vogel, B., Wegener, C., Hahlweg, K., & Schindler, L. (1980). Escalation phenomena in interaction sequences: An empirical comparison of distressed and non- distressed couples. Behavioral Analysis and Modification, 4, 97-115.

Rounsaville, B. J., Weissman, M. M., Prusoff, B. A., & Hereeq-Baron, R.L. (1979). Marital disputes and treatment outcome in depressed women. Comprehensive Psychiatry, 21, 480-490.

Spanier, G. B. (1976). Measuring dyadic adjustment: New scales for assessing the quality of marriage and similar dyads. Journal of Marriage and Family Therapy, 38, 15-28.

Spitzer, R., Endicott, J., & Robins, E. (1978). Research Diagnostic Criteria: Rationale and reliability. Archives of General Psychiatry, 35, 773-782.

Strack, S., & Coyne, J. C. (1983). Social confirmation of dysphoria: Shared and private reactions. Journal of Personality and Social Psychology, 44, 798-806.

Weissman, M. M., & Paykel, E. S. (1974). The depressed woman: A study of social rela- tionships. Chicago: University of Chicago Press.

Youngren, M. A., & Lewinsohn, P. M. (1980). The functional relation between depression and problematic interpersonal behavior. Journal of Abnormal Psychology, 89, 333-341.

RECEIVED: August 27, 1984 FINAL ACCEPTANCE: May 10, 1985