The emotional life of the toddler is immersed in the affective climate of the family system. Toddlers growing up in families in which mothers struggle with major depressive disorder are particularly likely to experience a negative emotional environment, not only in relation to their mothers, but also in the context of the larger family system. Consequently, this exposure to a relational context high in negative emotion poses considerable risk for the competent emotional development of these young offspring of depressed mothers (Cicchetti & Toth, 1995, 1998). In this article, we focus on the construct of expressed emotion to characterize the affective environment in families with a depressed mother. We will examine the extent to which mothers and fathers exhibit negative-expressed emotion in regard to their toddlers, their spouse, and themselves, and processes that may serve to mediate relations between maternal depression and negative-expressed emotion in the family. By utilizing expressed emotion to characterize emotion expression among family members, a family systems perspective can be brought to the study of offspring development (Minuchin, 1985).

Expressed emotion was originally conceived in efforts to understand family processes that contribute to relapse in individuals with schizophrenia (Brown, Birley, & Wing, 1972; Brown & Rutter, 1966; Vaughn & Leff, 1976). For discharged patients with schizophrenia returning to their families, relapse has been consistently shown to occur more frequently among individuals living with relatives who exhibit high expressed emotion. The expressed emotion construct, composed of criticism, hostility, emotional overinvolvement, warmth, and positivity, was intended to capture variation in routine emotional features of family life, rather than highly atypical or unusual patterns of relating. Expressed emotion has traditionally been scored from the Camberwell Family Interview (CFI), in which a relative discusses the family member with a disorder, his/her illness, their history of relationship, and how they currently relate. The content and affective tone expressed during the interview are used to identify high expressed emotion, with the expectation that the features exhibited during the interview correspond to emotional communication enacted in family relations. Later research sought to reduce the time needed to obtain expressed emotion assessments, and the Five-Minute Speech Sample (FMSS; Magana, Goldstein, Karno, Miklowitz, Jenkins, & Falloon, 1986) was devised as an alternative assessment strategy.

In addition to schizophrenia, expressed emotion also has been investigated in relation to the course of disorder for other psychiatric (e.g., major depression, bipolar disorder, posttraumatic stress disorder, alcoholism) and medical conditions (e.g., asthma, epilepsy, diabetes; for a review see Wearden, Tarrier, Barrowclough, Zastowny, & Rahill, 2000). In a meta-analysis, Butzlaff and Hooley (1998) found larger effect sizes for the association of high expressed emotion and relapse among individuals with mood disorders, compared to schizophrenia. High expressed emotion spouses of depressed individuals have been found to be less positive and more negative, to make more critical remarks, to disagree with their depressed partner more often, and to be less accepting (Hooley, 1986). Furthermore, spouses rated high in expressed emotion also are more likely to attribute responsibility and controllability to their depressed partners for their depressive symptoms (Hooley & Licht, 1997). Thus, high expressed emotion from spouses of individuals with depression is likely to contribute to ongoing symptomatology and relapse, as well as to generate negativity more broadly in the family milieu, thereby affecting other family members.

Beyond the extensive research on expressed emotion in understanding the course of adult psychopathology, expressed emotion research has been extended to the study of child functioning and psychopathology. Parental-expressed emotion has been investigated in the families of clinic-referred and normative child populations, with developmental periods assessed ranging from preschool through adolescence. Evidence consistently had been attained that reveals that parental-expressed emotion is associated with a wide range of childhood behavior problems, psychiatric diagnoses, and medical conditions. In a community sample of 6- to 12-year-olds, Stubbe, Zahner, Goldstein, and Leckman (1993) found specificity in the associations of the dimensions of expressed emotion and different child diagnoses. Disruptive behavior disorders were more frequent in families with high criticism from parents, whereas anxiety disorders were associated with high parental emotional overinvolvement. Asarnow and her colleagues (Asarnow, Tompson, Hamilton, Goldstein, & Guthrie, 1994; Asarnow, Tompson, Woo, & Cantwell, 2001) have demonstrated that high expressed emotion is more frequent in families of children with depressive disorders, compared to children with schizophrenia spectrum disorders, attention-deficit/hyperactivity disorder (ADHD), or normal controls, and influences subsequent outcomes among children with depressive disorders (Asarnow, Goldstein, Tompson, & Guthrie, 1993). Maternal-expressed emotion also has been shown to correspond to a threefold increase in the incidence of offspring having an affective, substance abuse, or conduct disorder, and to confer additional risk for a child diagnosis beyond that associated with having a parent with an affective disorder (Schwartz, Dorer, Beardslee, Lavori, & Keller, 1991).

Parental-expressed emotion toward children provides an independent contribution to differentiating clinical and nonclinical child samples, beyond the influences associated with overall family functioning and psychological distress in mothers (Kershner, Cohen, & Coyne, 1996). Thus, parental-expressed emotion in families indicates a negative family process that is not reducible to other forms of family distress and dysfunction. Moreover, stability of parental-expressed emotion toward children over the course of 2 years has been demonstrated in a community sample of preschoolers followed to first grade (Peris & Baker, 2000), suggesting continuity in expressed emotion in families over time. Peris and Baker (2000) also demonstrated that preschool-expressed emotion predicted ADHD 4 years later when children were in third grade.

In addition to child psychopathology, parental-expressed emotion has been linked to other developmental domains, including attachment and behavioral inhibition. In a small German sample of 6-year-olds, Jacobsen, Hibbs, and Ziegenhain (2000) found that maternal-expressed emotion regarding the child was related to insecure attachment, particularly disorganized attachment, in 6-year-old children. This association was maintained even when earlier attachment organization and level of family stress were considered. Higher rates of maternal criticism also have been found among mothers of behaviorally inhibited children (Hirshfeld, Bierderman, Brody, Faraone, & Rosenbaum, 1997a), and maternal criticism was related to a higher number of child disorders and to higher externalizing symptoms. Particularly high rates of maternal criticism were directed at behaviorally inhibited children when mothers had anxiety disorders (Hirshfeld, Bierderman, Brody, Faraone, & Rosenbaum, 1997b).

In contrast to the findings of Stubbe et al. (1993), most research has found that high parental criticism is the crucial element of expressed emotion that is associated with child psychopathology, whereas emotional overinvolvement has not been shown to relate to functioning in child and adolescent samples (Daley, Sonuga–Barke, & Thompson, 2003; Kershner et al., 1996; McCarty & Weisz, 2002; Nelson, Hammen, Brennan, & Ullman, 2003; Peris & Baker, 2000; Wamboldt, O'Connor, Wamboldt, Gavin, & Klinnert, 2000). Emotional overinvolvement, implying intrusiveness and excessive emotional concern, is thought to impair autonomy strivings among adults with schizophrenia or other major psychopathology. However, the emotional overinvolvement construct, as identified in expressed emotion research, may not be tapping developmentally inappropriate behavior for parents of younger children and adolescents (McCarty & Weisz, 2002; Wamboldt et al., 2000).

Research has demonstrated the construct validity of parental criticism through observational studies of parent–child interaction. In a large sample of clinically referred children age 7 to 17, McCarty, Lau, Valeri, and Weisz (2004) found that parents with high expressed emotion criticism regarding their child evinced higher antagonism, negativity, disgust, and harshness, as well as less responsiveness during observed interactions with their child. Similarly, Daley et al. (2003) observed mothers with high child criticism to be less affectionate and more directive and to engage in less joint play when interacting with their preschool children. Mothers exhibiting high criticism also had more negative and fewer positive feelings about parenting their child. Additionally, it has been shown that high maternal criticism is related to maternal causal attributions about child problem behavior. Highly critical mothers are more likely to view their child's behavior problems as internal and controllable, and to blame the child for behavioral difficulty (Bolton, Calam, Barrowclough, Peters, Roberts, Wearden, & Morris, 2003). Recent research has demonstrated that maternal-expressed emotion contributes to the differential development of antisocial behavior in monozygotic twins (Caspi, Moffitt, Morgan, Rutter, Taylor, Arseneault, Tully, Jacobs, Kim–Cohen, & Polo–Tomas, 2004). This research indicates that maternal-expressed emotion is not primarily attributable to individual features of the child, and that maternal-expressed emotion influences the development of antisocial behavior through processes not attributable to genetic differences. Overall, the findings from these studies illustrate that the criticism captured in expressed emotion assessments corresponds to actual differences in parental attitudes and behavior directed to children, and that these parental attributions and interaction styles are inconsistent with the promotion of child competence.

Various investigators have found that parental-expressed emotion is related to parental psychopathology (Bolton et al., 2003; Brennan, Hammen, Katz, & Le Broque, 2002; Hibbs, Hamburger, Kruesi, & Lenane, 1993; McCarty & Weisz, 2002; Schwartz et al., 1991), and maternal depression is a frequently noted diagnosis. More importantly, in several studies, expressed emotion has been identified as a mediator of the relation between maternal psychopathology and child behavior problems and diagnoses (Bolton et al., 2003; McCarty & Weisz, 2002; Nelson et al., 2003). In one study (Nelson et al., 2003), multiple aspects of maternal depression, expressed emotion, and adolescent behavior problems were evaluated. Maternal depression was associated with internalizing and externalizing problems and with functional impairment in adolescents. Moreover, maternal criticism was found to partially mediate the relation between maternal depression and adolescent externalizing problems and functional impairment. Brennan et al. (2003) also established a mediating role for expressed emotion. In particular, paternal-expressed emotion was found to mediate the relations between maternal depression, paternal substance use, and adolescent diagnoses for depression. This study is particularly important because it incorporates a broader family perspective through attending to the role of the father's mental health and paternal-expressed emotion. These findings indicate that a broader perspective on other features of the family system are important considerations in understanding linkages among maternal depression, expressed emotion, and child functioning.

We know of no existing research that has extended the concept of expressed emotion to the study of the early family relational environment of toddlers or to the study of offspring of mothers with major depressive disorder during the child's second year of life. If toddlers of depressed mothers are at greater risk for experiencing criticism from their mothers, then these children are in jeopardy of being confronted with negative communications about the self at a time when the early self is being differentiated. Among older children, self-blaming comments and attributional style have been related to observed maternal criticism (Jaenicke, Hammen, Zupan, Hiroto, Gordon, Adrian, & Burge, 1987). Thus, negative-expressed emotion from parents toward a young child would constitute a detrimental family process likely to undermine competent emotional development in these children.

Expressed emotion research has largely focused on a single target of parents', spouses', or relatives' criticism and emotional overinvolvement (e.g., the patient with severe mental disorder, children with behavior problems). Less attention has been directed toward expressed emotion that may be exhibited more broadly in the family, across multiple relationships. Depressed mothers, for example, are likely to be highly self-critical (Blatt, Quinlan, Chevron, McDonald, & Zuroff, 1982; Coyne & Whiffen, 1995), as well as critical of their husbands, given high levels of marital dissatisfaction and conflict among couples with a depressed spouse (Coyne, Downey, & Boergers, 1992; Cummings, Davies, & Campbell, 2000; Weissman & Paykel, 1974). Similarly, the husbands of depressed women are more likely to exhibit negative-expressed emotion about their wives, potentially in a reciprocal fashion. In so doing, husbands may exacerbate symptoms in their wives and contribute to a higher risk of relapse. Thus, the toddler in these families may not only be subjected to parental criticism, but also is likely to experience negative-expressed emotion across different primary relationships in the family system. As a result, high levels of negativity, hostility, and criticism across relationships may characterize the family environment of toddlers with a depressed mother, generating a hazardous emotional context for socioemotional development.

Women with depressive disorders are more likely to be married to husbands with a history of psychopathology (Merikangas & Spiker, 1982; Peterson, Coyne, & Kessler, 1991). Moreover, depressed adults are more likely to have experienced conflict and negativity in their relationships with their parents during childhood (Blatt, 2004; Blatt & Homann, 1992), and Coyne et al. (1992) suggest that depressed persons and their spouses may be attracted to each other based on similarities in early adversity. Accordingly, a confluence of negative early childhood experiences, psychopathology with attendant high levels of negative affect, and martial conflict and dissatisfaction may characterize the marital relationship in families with a depressed mother (Cummings & Davies, 1994a, 1994b). Not only mothers with depressive disorders, but also their husbands are likely candidates for exhibiting high expressed emotion, particularly criticism. Moreover, these features of depressed families are likely to promote negative-expressed emotion across relationships, and may contribute to negative-expressed emotion in relation to the young child.

Whereas psychopathology in the relatives of patients with severe mental illness has been indicated as a contributor to expressed emotion (Wearden et al., 2000), there has been limited research on broader predictors of individual differences in the tendency to exhibit expressed emotion. When an individual is asked to describe a family member and their relationship during the CFI or FMSS, one must draw upon processes contributing to how one conceptualizes self and other. More broadly, representations of self and other have roots in the individual's experiences with caregivers during childhood (Bowlby, 1969, 1980; Bretherton, 1990; Sullivan, 1953). Given the likelihood of compromised childhood relationships with parents for both depressed mothers and their husbands, internal working models of self and other in relationships among depressed mothers and their spouses are prone to be negative. Consequently, when asked to describe the self and others in current relationships, aspects of negative early experiences may impinge upon how representations are formulated and expressed. This process could constitute an intergenerational contributor to current expressed emotion in the family, whereby earlier relationship difficulties influence the emotional features of current family relationships.

Negative affect as a trait also may contribute to the propensity of individuals to express criticism in relation to others, as well as to the self. Negative affect is a general dimension of subjective distress, including a variety of aversive mood states, such as anger, contempt, disgust, guilt, fear, and nervousness (Watson, Clark, & Tellegen, 1988). Individuals who exhibit high levels of trait negative affect would be expected to be more liable to expressing negativity, hostility, and criticism in relation to others. Moreover, trait-negative affect is expected to be considerably higher among depressed mothers, thereby increasing their vulnerability to exhibiting high expressed emotion. In turn, spouses may react to high negative-trait affect in their partners, and exhibit high expressed emotion in response. In families with a depressed mother, high levels of marital distress and dissatisfaction are likely to promote negative-expressed emotions by both spouses. Thus, negative childhood relationship experiences, trait-negative affect, and marital distress are examples of processes that are candidates for increasing expressed emotion in regard to another family member. Furthermore, these processes are expected to be heightened in families with a depressed parent, and may serve to mediate or account for higher levels of expressed emotion in these families.

Expressed emotion may constitute a generalized interpersonal style of an individual, and, as such, the tendency to exhibit expressed emotion may occur across multiple relationships. In contrast, expressed emotion may only occur in relation to a specific family member. Furthermore, there may be specificity in the most potent predictors of the tendency to exhibit expressed emotion in regard to particular relationship figures. For example, negative-expressed emotion in regard to the self may be more strongly predicted by early childhood relationship characteristics, whereas expressed emotion in regard to the spouse may be based more on marital dissatisfaction or a high negative affect in the spouse. Consequently, we will examine statistical models to identify the strongest predictors of expressed emotion and the specificity of these predictors in regard to different relationship figures.

From a family systems perspective, we also are interested in determining the extent to which the expressed emotion of one family member is influenced not only by internal processes, but also by contributions from other family members. For example, high negative-trait affect in one parent may contribute to greater negative-expressed emotion exhibited by the other parent in regard to the self or the child. Similarly, marital dissatisfaction experienced by one parent may translate into criticism of the child, or negative childhood relationship experiences in one parent may influence expressed emotion tendencies in the other parent. Thus, we anticipate that, in addition to within-individual predictors, there will be cross-person contributions to an individual's tendency to exhibit expressed emotion.

In this investigation we will examine the emotional climate of families of toddlers with a depressed mother. We will investigate expressed emotion exhibited by mothers and fathers in regard to their child, their spouse, and themselves and evaluate the following hypotheses:

1. Mothers with a history of major depressive disorder will exhibit higher levels of negative-expressed emotion, particularly in the form of criticism, in regard to their child, their spouse, and themselves than will mothers who do not have a history of major depressive disorder (MDD).
2. Fathers in families with a depressed mother will also be likely to be more critical of child, self, and spouse than will fathers in families with nondepressed wives.
3. Families with depressed versus nondepressed mothers will differ on variables potentially related to expressed emotion, including less positive maternal and paternal representations of their childhood relationships with parents, greater trait-negative affect for mothers and fathers, and greater marital dissatisfaction among spouses.
4. These relational/affective processes will be related to expressed emotion/criticality of child, self, and spouse.
5. These relational/affective processes will mediate the relation between families with a depressed mother and higher criticism across relationships.

Results

Expressed emotion in families with a depressed mother

Expressed emotion scores of criticism and emotional overinvolvement were determined for mothers and for fathers, for each of the relationships that were described in the separate FMSSs (self, spouse, and child). For each of these 3-point scores, t tests were conducted to compare the families with depressed and nondepressed mothers, as shown in Table 1. Consistent group differences were found for criticism, whereas no depressed group differences were observed for emotional overinvolvement. Depressed mothers expressed significantly higher criticism in regard to self, spouse, and child. Fathers in the depressed group also expressed significantly higher criticism for self and for spouse; no group differences were found for fathers' criticism of the child. Thus, higher levels of criticism across relationships mark the emotional climate in the families of the depressed group, and both mothers and fathers contribute to the expression of criticism.

Comparison of depressed and nondepressed groups on expressed emotion criticism and emotional overinvolvement

Correlations among the six mother and father criticism variables are presented in Table 2. For mothers, criticism expressed in one relationship corresponded to criticism expressed in other relationships. Similarly, fathers' criticism scores were significantly intercorrelated. However, in examining correlations between mothers and fathers, only one correlation was significant. Criticism expressed regarding the spouse was correlated for mothers and fathers (r = .20, p = .01), suggesting a degree of reciprocity among spouses in the tendency to be critical of each other.

Correlations among mother and father expressed emotion criticism variables

Depression group differences in marital satisfaction, trait-negative affect, and childhood relationships with parents

We wanted to examine influences from both mothers and fathers that could contribute to the higher levels of criticism observed in the depressed group families. Initially, group differences in maternal and paternal marital satisfaction, the tendency to experience and exhibit negative affect, and appraisals of the quality of parents' childhood relationships with their own parents were examined (see Table 3). In the families of the depressed group, significant differences were found in these domains for both mothers and fathers. Depressed mothers, compared to nondepressed mothers, reported significantly lower marital satisfaction and higher trait negative affect. Depressed mothers also indicated that their own mothers and fathers in childhood were less accepting and less encouraging of independence, relative to nondepressed mothers. The findings for fathers in the depressed group also indicated substantial group differences on these variables. Fathers in the depressed group reported significantly lower marital satisfaction, higher negative affect, and poorer relationship qualities with their own parents in childhood, than did fathers in the nondepressed group.

Comparison of depressed and nondepressed groups on relational/affective variables

Because some of the fathers in the depressed group had histories of psychopathology, we examined whether having an Axis I diagnosis currently or in the past 6 months contributed to the differences observed for the depressed group. Accordingly, we compared variables for the depressed group families depending on whether fathers had current/recent mental disorders or whether diagnoses were absent. No differences were found for any of the mother- or father-expressed emotion variables, and only one mother and one father relational/affective variable differed significantly between groups. Depressed mothers with husbands without current/recent diagnoses reported higher negative affect (M = 2.83, SD = 0.60) than did depressed mothers with husbands who had current/recent mental disorder (M = 2.84, SD = 0.61), t (98) = 2.24, p < .05. In contrast, fathers with current/recent diagnoses reported lower marital satisfaction (M = 3.48, SD = 0.72) than did fathers without diagnoses (M = 3.79, SD = 0.51), t (98) = 2.34, p > .05. Due to this limited effect of paternal psychopathology in the depressed group, we did not consider this feature further.

As expected, the relational/affective variables shared considerable variance. Among mothers, 13 of 14 correlations among these variables were significant. Similarly, among fathers, 13 of 14 correlations also were significant. In contrast, however, mother and father negative affect was not correlated, and the MFP variables regarding childhood relationships also were not significantly correlated. Mothers and fathers did, however, agree strongly on their marital satisfaction (r = .62, p < .001). Marital satisfaction was significantly positively correlated with the other spouse's childhood relationship quality with parents and negatively correlated with the spouse's negative affect. These relations were found for both mothers' and fathers' marital satisfaction.

We next examined how these relationship/affect variables correlated with the mother- and father-expressed emotion criticism variables (see Table 4), and important associations were identified. For mothers, negative affect and MFP childhood relationship variables were significantly correlated with self- and spouse criticism. Maternal marital satisfaction also was negatively correlated with criticism of her husband. In contrast, maternal criticism of the child was only correlated with maternal negative affect. For fathers as with mothers, marital satisfaction was negatively correlated with spousal criticism. Fathers' negative affect was not correlated with his criticism across the relationships examined, and self- and spouse criticism did not show the consistent relations with MFP childhood relationship variables as was observed among mothers. However, each of the fathers' MFP variables was negatively correlated with criticism of the child. Thus, fathers who reported less harmonious relations with their own parents tended to exhibit more criticism regarding their child.

Correlations of expressed emotion criticism with relational/affective variables

Of particular interest were associations of one parent's criticism variables with self-reported variables from the other parent. Paternal marital satisfaction was negatively correlated with maternal criticism of both her child and her spouse. Similarly, maternal marital satisfaction was negatively correlated with paternal self- and spouse criticism. Fathers' acceptance from both of his parents was correlated negatively with maternal spousal criticism, and fathers' negative affect was correlated positively with all three of the maternal criticism variables. In contrast, maternal negative affect showed no relationship to any of the fathers' criticism variables. However, mothers who reported that their fathers were more encouraging of independence had husbands who were less critical. Thus, relational/affective processes in one parent demonstrated associations with levels of criticism in the other parent, and differential relations were observed depending on the target of criticism.

Mediational models of the relations between maternal depression group and criticism

Given the differences observed between the depressed and nondepressed groups on expressed emotion criticism exhibited in multiple relationships, the differences between the depressed groups on a variety of maternal and paternal relational/affective variables, as well as various associations between the relational/affective variables and maternal and paternal criticism, we sought to determine whether these maternal and paternal relational/affective variables served to mediate relations between depression group membership and expressed emotion criticism. Moreover, as a result of considerable shared variance among the relational/affective variables, we sought to determine the most parsimonious models that could be identified to predict each of the maternal and paternal criticism variables. Following the guidelines delimited by Baron and Kenny (1986) for evaluating mediation, we have already established preliminary criteria for potential mediational relationships. In particular, we have demonstrated that the independent variable, maternal depression group, is related to the outcomes of expressed emotion criticism and to various potential mediators, that is, the maternal and paternal relational/affective variables. Relations between these latter variables and the expressed emotion criticism variables also have been demonstrated. Thus, in a series of hierarchical multiple regression analyses, we entered maternal depression group into the regression equation in the first step, and in the second step we added relational/affective variables that significantly correlated with the criticism outcome variable. Final models were trimmed so as to retain only relational/affective variables demonstrating an independent contribution to the regression equation. Table 5 presents the regression summary tables for the final models.

Regression models predicting expressed emotion and child behavior problems

Maternal criticism of self

Among maternal variables, maternal negative affect and aspects of the mother's relationships with her own parents were associated with maternal self-criticism. Paternal negative affect and the father's appraisal of acceptance from his own father also had significant associations. In Step 1 of the regression equation, depression group status was entered, and subsequently in Step 2, the potential mediating variables were added. Father's negative affect and the mother's appraisal of her father's encouragement of independence each made significant independent contributions to the regression equation. However, depression group status also retained a significant independent effect, and the strength of this relation was not substantially reduced compared to Step 1. Thus, the final regression model obtained is an additive model, with all three variables contributing independently to prediction. It is noteworthy that, in addition to maternal depression, negative affect of her husband and a variable involving negative relations with the mother's father in childhood each contributed to predicting the mother's tendency to be critical of herself.

Paternal criticism of self

In examining the possible mediators of the relation between depression group and paternal self-criticism, only two variables were found to be significantly associated with self-criticism (see Table 4); maternal marital satisfaction and paternal feelings of acceptance from his own father were negatively correlated with self-criticism. When these variables were entered in Step 2 of the regression equation along with depression group, only depression group and paternal acceptance retained independent contributions. As was the case in predicting maternal self-criticism, a mediational model was not supported for paternal self-criticism. Rather an additive model in which depression group and the father's feelings of acceptance from his own father contributed to the regression equation.

Maternal criticism of spouse

In addition to depression group, multiple maternal and paternal variables (marital satisfaction, negative affect, and MFP variables) were associated with mothers' level of criticism of their spouses. When these variables were entered into the regression equation in Step 2, two variables retained significant independent effects, maternal marital satisfaction and mothers' feelings of their own fathers encouraging independence. Additionally, the inclusion of these variables substantially reduced the effect of maternal depression group, with depression retaining only marginal significance. Thus, this model indicates that the relation of depression group status to greater maternal criticism of her spouse is mediated by the mother's marital dissatisfaction and feelings of her father not encouraging her independence in childhood.

Paternal criticism of spouse

Interestingly, similar predictors of maternal spousal criticism also were retained in the model of fathers' spousal criticism. In Step 2, when entered along with depression group, two variables—fathers' marital satisfaction and mothers' feelings about her father encouraging independence—retained significant independent effects. Moreover, the effect of maternal depression group was substantially reduced and no longer significant, indicating that fathers' marital satisfaction and aspects of the mother's relationship with her father mediated the relation between depression group status and father's spousal criticism. This model is noteworthy, given the contribution of both paternal and maternal processes in accounting for higher levels of spousal criticism in families with depressed mothers. Moreover, as in the case of maternal spousal criticism, current and family of origin relational influences were found to be contributors to fathers expressing criticism of their wives.

Maternal criticism of the child

Both maternal and paternal variables contributed to the prediction of mothers expressing criticism of their child. In particular, in Step 2, both maternal and paternal trait negative affect were retained in the regression equation when added along with depression group. Additionally, these two variables completely mediated the relation between depression group and maternal child criticism. Not only does maternal negative affect contribute to tendencies to be more critical of the child, but also the presence of higher levels of paternal negative-trait affect represents another process operating to increase maternal criticism of the child.

Paternal criticism of the child

In contrast to the findings for mothers, neither maternal nor paternal negative affect was related to fathers expressing criticism of their child. Moreover, no maternal variables were correlated with paternal criticism of the child, and paternal criticism levels did not differ between the depressed and the nondepressed groups. One domain of paternal variable was found to relate to higher child criticism from fathers. The MFP variables of fathers' feelings that their mothers and fathers encouraged independence, and that their fathers were accepting were negatively correlated with paternal criticism. Given shared variance among these variables, when examined collectively in a regression equation, only mother's encouragement of independence retained a significant effect. Thus, fathers' tendency to express criticism of their child appeared to be uniquely related to childhood family of origin relationships.

Child behavior problems and expressed emotion criticism

Both mothers and fathers provided assessments of their child's behavioral functioning through independent completion of the CBCL. Mothers' and fathers' reported levels of total behavior problem t scores were significantly correlated (r = .33, p < .001). Mothers in the depressed group (M = 50.63, SD = 8.09) reported significantly higher levels of child behavior problems than did mothers in the nondepressed group (M = 47.33, SD = 7.90), t (151) = 2.37, p < .02. The two groups did not differ, t (152) = 1.03, ns, in the level of father-reported child behavior problems.

The expressed emotion criticism of the child by both mothers and fathers was correlated with maternal reports of child behavior problems (rs = .19 and .20, ps < .05, for mothers and fathers, respectively) and with paternal reports of child behavior problems (rs = .26 and .30, ps < .001, respectively). Thus, higher criticism expressed by parents corresponds to higher levels of child behavior problems reported by both parents.

Given the higher levels of child behavior problems reported by mothers in the depressed group, we wanted to determine if maternal and paternal affect/relational variables and/or expressed emotion criticism would serve to mediate this relation. Among the maternal and paternal relational/affect variables, the only variable that correlated with mothers' reports of child behavior problems was maternal negative affect. Following the previous procedures for evaluating mediation, we first examined whether maternal criticism of the child qualified as a mediator. When entered in the second step of a regression equation along with depression group, both group status and maternal criticism retained independent effects in predicting child behavior problems. Thus, maternal criticism did not serve as a mediator of the relation between depression group status and higher maternal report of child behavior problems.

Subsequently, we also included maternal negative affect in the regression model at Step 2. In this model, the independent effect of maternal criticism was no longer significant. Rather maternal negative affect made an independent contribution and served to mediate the relation between maternal depression group and child behavior problems. Finally, we also considered contributions from paternal variables. The only paternal variable associated with maternal report of child behavior problems was fathers' expressed criticism of the child. When this variable was added in Step 2, along with depression group and maternal negative affect, the mediating effect of maternal negative affect was retained, and fathers' criticism contributed further variance to prediction of maternal reports of child behavior problems. However, because fathers' criticism was not associated with depression group status, paternal criticism did not serve as mediator.

Although fathers' reports of child behavior problems were not related to depression group, we evaluated whether maternal and paternal negative affect and criticism collectively related to fathers' reports of behavior problems. Following depression group status entered in Step 1, in Step 2, maternal and paternal negative affect and maternal and paternal child criticism were entered. Each of these variables retained a significant independent effect in contributing to fathers' reports of child behavior problems. Thus, both maternal and paternal contributions were associated with fathers' reports of child behavior problems.