Patterns of Marital Conflict Predict Children's Internalizing and Externalizing Behaviors

Results are reported of a longitudinal study on how marital interaction affects children. Observational assessments of marital interaction during conflict resolution obtained when children were 5 years old predicted teachers' ratings of internalizing and externalizing behaviors when the children were 8 years old. Two distinct and uncorrelated marital interaction patterns were related to specific forms of child outcomes. The Mutually Hostile pattern, which correlated with later marital dissolution, also predicted externalizing behavior patterns in children 3 years later. The Husband Angry and Withdrawn pattern predicted child internalizing behaviors. Marital satisfaction and child temperament did not relate to child outcomes, nor did they interact with marital patterns to produce deficits in child adjustment. The findings suggest that the specific behaviors couples use when resolving marital disputes may contribute differentially to the presence of externalizing and internalizing behavior patterns in children. Portions of this article were presented at the biennial meeting of the Society for Research in Child Development, Seattle, Washington, in April 1991. This research was supported by National Institute of Mental Health Grant MH40484 to John M. Gottman. Correspondence may be addressed to Lynn Fainsilber Katz, Department of Psychology, University of Washington, Seattle, Washington, 98195. Received: January 30, 1992 Revised: May 1, 1993 Accepted: June 3, 1993 There is now convincing evidence to suggest that marital distress and conflict are associated with a wide range of deleterious child outcomes, including depression, withdrawal, poor social competence, health problems, poor academic performance, and a variety of conduct-related difficulties ( Cowan & Cowan, 1990 ; Easterbrooks, 1987 ; Emery & O'Leary, 1982 ; Forehand, Brody, Long, Slotkin, & Fauber, 1986 ; Gottman & Katz, 1989 ; Hetherington, Cox, & Cox, 1982 ; Peterson & Zill, 1986 ; Porter & O'Leary, 1980 ; Rutter, 1971 ; Whitehead, 1979 ). Although this literature is suggestive of a link between the parents' marriage and child outcomes, the direct relationship between marital quality and child outcomes has only rarely been examined in a prospective longitudinal study (e.g., Cowan & Cowan, 1990 ; Easterbrooks, 1987 ; Howes & Markman, 1989 ). One issue in this research is that marital quality has been almost uniformly conceptualized in terms of relationship satisfaction. Although this approach has generated important new findings, the simple, Developmental Psychology © 1993 by the American Psychological Association November 1993 Vol. 29, No. 6, 940-950 For personal use only--not for distribution. Page 1 of 19 11/27/2000 http://spider.apa.org/ftdocs/dev/1993/november/dev296940.html unidimensional focus on marital satisfaction has failed to identify the specific dimensions of marital quality that are correlated with or predictive of child functioning. As a result, although there is evidence that distressed marriages are correlated with negative child outcomes, it is unclear exactly what it is about those marriages that are most caustic to children's well-being. Identifying the dimensions of the marital relationship that are most detrimental to children's outcomes has both pragmatic and conceptual benefits. At a pragmatic level, it is important for building therapeutic programs for ailing marriages and minimizing negative consequences for children. Simply knowing that a couple is maritally distressed does not provide the therapist with specific information about the marital behaviors that need to be changed to have maximum benefit for the child. If characteristics of the marital relationship that are especially harmful to children can be identified, marital therapies for families with children can include a treatment component for improving those marital behaviors that are most destructive to children. At the conceptual level, there is a need to question the information that is obtained from the finding of a relationship between self-report of marital distress and child outcomes. It is well known that people who are stressed in other areas of their life also report being unhappily married. For example, marital satisfaction has been found to covary with a diverse range of negative life stressors and states, such as the quality of life, job stress, a variety of dysfunctional personality characteristics, and depression (e.g., see Barton & Dreger, 1986 ; Beach, Arias, & O'Leary, 1986 ; Burgess, Locke, & Thomes, 1971 ; Krokoff, 1984 ; Lewak, Wakefield, & Briggs, 1985 ; Smolen, Spiegel, & Martin, 1986 ; Yogev, 1986 ). Thus, self-report measures of marital satisfaction may, in part, reflect a general stress dimension rather than something specific about the marriage. An assessment of marital quality that is independent of couples' own ratings would add precision to understanding the relationship between functioning in the marital and child systems. Such an assessment is possible using the quantitative observation of marital interaction (e.g., Gottman, 1979 ; Weiss & Summers, 1983 ). Observing couples discussing issues important to their relationship is now an established approach for specifying dimensions of marital quality (e.g., see Markman & Notarius, 1987 ). Research on marriage has consistently demonstrated that the way couples resolve conflict is important in differentiating between happily and unhappily married couples (e.g., Gottman, 1979 ; Olson, Spengle, & Russell, 1979 ; Raush, Barry, Hertel, & Swain, 1974 ; Revenstorf, Vogel, Wegener, Hahlweg, & Schindler, 1980 ; Vincent, Weiss, & Birchler, 1975 ). Happily married couples have been found to display higher ratios of agreement to disagreement ( Gottman, 1979 ) and exhibit more positive nonverbal cues ( Birchler, 1977 ; Haynes, Follingstad, & Sullivan, 1979 ), more agreement and approval ( Vincent & Friedman, 1979 ), and less coercive and attacking behaviors ( Billings, 1979 ) than unhappily married couples. One of the most consistent discriminators between happily and unhappily married couples has been the degree of negative affect expressed during conflict resolution. Unhappily married couples have been found to show more negative affect and negative affect reciprocity than happily married couples (e.g., Gottman, 1979 ; Revenstorf et al., 1980 ). Affective differences that are independent of marital satisfaction have also been noted. For example, Margolin (1988) proposed that couples differ in the way in which emotions are expressed during conflict resolution, with some couples expressing their negativity very openly and directly and others keeping the conflict silent and hidden. The consequences of these different affective patterns of marital conflict resolution for children's socioemotional development have been largely unexplored. There are some preliminary data to support the hypothesis that the different ways in which adults resolve conflict have negative consequences for children. In a series of investigations, Cummings and Page 2 of 19 11/27/2000 http://spider.apa.org/ftdocs/dev/1993/november/dev296940.html colleagues consistently demonstrated that exposure to interadult anger is associated with distressed, angry, and physically aggressive reactions in children ( Cummings, 1987 ; Cummings, Iannotti, & ZahnWaxler, 1985 ; Cummings, Zahn-Waxler, & Radke-Yarrow, 1981 ). On the basis of clinicians' ratings of interviews with individual spouses, Rutter and colleagues ( Rutter et al., 1974 ) found a stronger relationship with child behavior problems in unhappy marriages characterized by interviewers as "quarrelsome" than those characterized as "apathetic." Thus, both these studies support the hypothesis that the particular way couples engage in conflict may be associated with negative or dysfunctional behavior patterns in children. In our research, we examined whether two different patterns of marital conflict were related to children's behavior problems. One negative marital interaction pattern we examined has been described as a "demand—withdraw" ( Christenson, 1987 , 1988 ) or "pursuer—distancer" pattern ( Fogarty, 1976 ), in which one spouse requests change through demands, criticism, and complaints and the other spouse retreats through withdrawal and avoidance. There is also evidence for a consistent gender difference in the way married men and women handle marital conflict, with husbands tending to withdraw and wives tending to engage in conflict ( Christensen, 1987 , 1988 ; Gottman & Levenson, 1988 ; Sullaway & Christensen, 1983 ). For example, in one of the earliest studies of marriage, Terman, Buttenweiser, Ferguson, Johnson, and Wilson (1938) reported that husbands' marital grievances were most likely to involve their wives' complaining, criticizing, and escalating emotion, whereas wives' grievances were most likely to involve their husbands' emotional withdrawal. The Demand—Withdrawn pattern appears to be aversive for both spouses. Christensen and colleagues have found that the frequency of the Demand—Withdrawn pattern is strongly associated with reports of marital dissatisfaction ( Christensen, 1987 , 1988 ; Sullaway & Christensen, 1983 ). Husbands' withdrawal from marital interaction has also been associated with greater physiological arousal on the part of both husbands and wives ( Gottman & Levenson, 1992 ). Given the aversive nature of this pattern for both spouses, we reasoned that the Demand—Withdrawn pattern may also have negative consequences for their children. The present study also examined a negative marital interaction pattern that has been found to predict marital dissolution. Gottman (1993) recently reported that marriages headed toward dissolution are characterized by contempt, particularly by the wife. When spouses are contemptuous toward each other, they communicate a sense of superiority and moralistic disapproval through insults, mockery, or attributions of the partner's incompetence. Contemptuous statements are often accompanied by belligerent demands in which the spouse contests his or her partner's statements by trying to provoke a response or get a rise out of the partner. Because both partners usually engage in this negative communication style, we have labeled it a Mutually Hostile pattern of conflict resolution. Given the now substantial body of research identifying the numerous negative consequences of divorce for children ( Hetherington, Cox, & Cox, 1978 , 1982 ; Shaw & Emery, 1987 ; Wallerstein & Kelly, 1975 ), it seemed possible that children may be especially affected by a Mutually Hostile marital pattern because it bodes ill for the long-term prognosis of the stability of the marital relationship. Because observed marital behavior is not independent of marital satisfaction (e.g., Christensen, 1988 ; Gottman, 1979 ; Jacobson & Margolin, 1979 ; Weiss & Summers, 1983 ) and self-report of marital satisfaction has been the predominant measure of marital quality in this research area, the relative contributions of self-report and observational measures in predicting child outcomes were examined. First, we addressed the rival hypothesis that children's adjustment could be predicted from marital satisfaction alone. Second, any additive effects of negative interaction patterns and reported marital dissatisfaction were also tested to determine whether parents who show either the Mutually Hostile or Demand—Withdrawn pattern and also reported marital dissatisfaction had children who were at greater risk for adjustment problems than parents who showed only one of the two interaction patterns. Page 3 of 19 11/27/2000 http://spider.apa.org/ftdocs/dev/1993/november/dev296940.html It is also important to determine the degree to which characteristics of the children operate in conjunction with or independent of family processes to contribute to children's adjustment problems. Child effects such as gender and temperament have been identified as important factors in the link between marital turmoil and child outcomes (e.g., Emery, 1982 ). With respect to gender, some studies suggested that marital discord affects boys more than girls, whereas other studies suggested that all children are affected by marital discord but boys from maritally discordant homes show externalizing difficulties and girls show internalizing problems ( Block, Block, & Morrison, 1981 ; Emery, 1982 ). The issue of child temperament raises difficult questions regarding the direction of effects. Having a temperamentally difficult child may put a strain on the marital relationship. In terms of consequences for the child, being temperamentally difficult may be a high risk factor for children whose parents display a negative marital interaction pattern. Children who are seen as difficult by parents and whose parents show negative marital interaction patterns may demonstrate more adjustment problems than those whose parents only demonstrate negative interaction patterns. Both temperament and gender effects were explored in the present report. The effects of the Demand—Withdrawn and Mutually Hostile marital interaction patterns on children's adjustment were examined within a prospective longitudinal design. There have been no prospective longitudinal studies that have examined families to determine exactly what marital processes are associated with detrimental child outcomes. Indeed, only a handful of studies investigated the longitudinal impact of marital satisfaction on child outcomes (e.g., Cowan, Cowan, Heming, & Miller, 1991 ; Howes & Markman, 1989 ). In the present study, families were first seen before the children were in school, when they were, on average, 5 years old. Follow-up assessments occurred when the children were in school (on average, 8 years old). Teacher ratings of children's adjustment were obtained at follow-up to provide independent reports of child functioning. Children's externalizing (e.g., aggressive, hyperactive) and internalizing (e.g., depressed, withdrawn) difficulties were the two behavioral constellations targeted as child outcome variables.