Introduction

Considerable research has documented that prolonged exposure to interparental conflict (IPC) has negative consequences for children and adolescents, including increased risk for externalizing and internalizing problems (e.g., Davies and Windle 2001; Emery 1988). Research on the mechanisms by which conflict affects children’s well-being has tended to focus on two possibilities. The first suggests that IPC leads to maladaptive parenting, which in turn leads to child maladjustment (Cox et al. 2001; Fauber et al. 1990; Gerard and Buehler 1999). The second proposes that the impact of conflict is mediated in part by children’s cognitions, specifically, their attributions regarding the cause of conflict and their perceptions of their own ability to affect the conflict (Grych and Cardoza-Fernandez 2001; Grych and Fincham 1990). Cognitions have been found to be particularly predictive of depressive symptoms (Dadds et al. 1999). Most recently, research on the effects of IPC on children has taken a process-oriented approach, attempting to explore contextual, child, and over-time effects on children’s adjustment in the context of conflict (Cummings and Davies 2002).

Despite the extensive research that has examined the effects of IPC on children’s well-being, there are some important limitations. First, much of the evidence supporting suboptimal parenting as a mediator of the impact of IPC on children has come from cross-sectional studies. While useful, cross-sectional designs are not as rigorous as longitudinal designs in their ability to test mediational theories. Second, the research that has focused on the role of children’s cognitions has tended to examine cognitions specific to IPC (e.g., why the conflict is occurring, who is responsible). Again, this work has yielded important insight into the role of domain-specific attributions, but it does not address the question of whether children’s general cognitive style makes them more or less vulnerable to the impact of IPC. Third, while it has been suggested that both family factors and child level variables play a role in the link between IPC and child outcomes (Fincham et al. 1994), we found only one study (Grych 1998) that explored the possibility that parenting acts in concert with children’s cognitions in the context of conflict. Finally, few studies have explicitly focused on the effects of IPC on children living in low-income, urban environments. The limited scope of this research is unfortunate, since there are many meaningful differences between the lives of these children and those of children living in middle-class, suburban environments (McLoyd and Steinberg 1998). Before developing interventions based on research with majority children, it is incumbent on us to explore whether or not IPC affects low-income, urban children in the same way and via the same processes. In this article, we attempt to address some of these limitations by examining the effects of IPC on depressive symptoms in a sample of racially and ethnically diverse fifth- and sixth-grade children living in low-income, urban environments. Specifically, we examine the possibility that both parenting style and children’s general cognitive style contribute to the impact of IPC on child depression over time.

Interparental Conflict, Parenting, and Child Depression

In his early review of the research that investigated possible mechanisms by which IPC was linked with child problems, Emery (1982) suggested that high levels of conflict might lead to changes in parenting style that would in turn negatively affect children. Using the concept of spillover, a considerable body of research supports the theory that IPC often results in increased parent-child conflict (Almeida et al. 1999; Bradford et al. 2008), inadequate parent discipline (Buehler and Gerard 2002), and disruptions in the quality of parent-child affect (Sturge-Apple et al. 2006) that are associated with child maladjustment (see Erel and Burman 1995; Zimet and Jacob 2002 for reviews). The core tenet of the spillover hypothesis is that the emotional distress and distractions of IPC drain parental resources and make it less likely that parents will provide children with warmth, support, and structure, which in turn negatively impacts children’s emotional well-being. While evidence from a recent meta-analysis of 45 cross-sectional studies published from 1985 to 2005 exploring the direct connection between parenting and child and adolescent depressive symptoms determined that parenting accounted for only 5 to 8% of the variance in childhood depression (McLeod et al. 2007), parenting remains an important dimension to consider when exploring family systems level processes because of its often significant moderating and mediating effects. In their meta-analysis of 39 studies, Krishnakumar and Buehler (2000) found an overall moderate relationship between IPC and poor parenting.

Research on parenting suggests three continuous dimensions on which parenting can be conceptualized and, as a consequence, affected by IPC (Barber et al. 2005; Gray and Steinberg 1999; Schaefer 1965). The first can be understood as parent behaviors and attitudes that affect the emotional quality of the parent-child relationship. Terms such as rejection versus acceptance, involvement, and warmth and caring each describe this dimension. The second dimension is often termed psychological autonomy versus psychological control and refers to the concept whereby parents vary in encouraging their children to make their own decisions without applying pressure or rewards to control their child’s behavior. The third dimension refers to the extent to which parents set rules, expectations, and consequences on their child’s behavior and has been termed firm versus lax control, limit-setting, and structure. A parenting style that is high on each of these dimensions (i.e., acceptance or involvement, autonomy granting, and structure) is associated with positive outcomes for children (Grolnick et al. 1991), while parenting characterized by rejection, psychological control, and lax structure or discipline is associated with negative outcomes, including depression (Garber et al. 1997; O’Donnell and Grolnick 2004, Parenting and early adolescents’ symptomatology: a test of self-determination theory, "Unpublished manuscript"). This finding appears to hold across a diversity of cultures (Barber et al. 2005). Research examining these parenting dimensions in the context of IPC has found that IPC is associated negatively with parental warmth and positively with inconsistent discipline and use of psychological control (Brody et al. 1996; Burman et al. 1987). While there is considerable empirical support for the idea that suboptimal parenting mediates the link between IPC and child maladjustment, including depression (Fauber et al. 1990; Gerard and Buehler 1999; Stoneman et al. 1989), compromised parenting does not entirely explain the relation between IPC and child problems (see Fincham and Osborne 1993 for a discussion). As a result, some researchers have begun to examine the role of children’s cognitions.

Interparental Conflict, Children’s Cognitions, and Child Depression

Cognitive models of depression suggest that depressed individuals are more likely than nondepressed individuals to possess habitually pessimistic styles of explaining events (Nolen-Hoeksema et al. 1986; Peterson et al. 1985). According to the reformulated learned helplessness theory (Abramson et al. 1978), a pessimistic explanatory style is marked by a tendency to explain negative events as caused by factors that are internal, stable in time, and global in effect, and to see positive events as caused by external, unstable, and specific factors. A pessimistic explanatory style has been found to predict increases in depressive symptoms and vulnerability to episodes of clinical depression over time among both adults and children (Nolen-Hoeksema et al. 1992; Peterson and Seligman 1984). Recent research suggests that this relationship also holds among diverse populations of adolescents (Cardemil et al. 2002; Cardemil et al. 2008). Reinemann and Teeter Ellison (2004) found that negative cognitions mediated the relationship between negative life events and depressive symptoms in a sample of urban, ethnic minority middle-school children, the majority of whom were African American. Similarly, Kennard et al. (2006) found that a variety of cognitive variables, (i.e., self-efficacy, negative cognitive errors, and hopelessness) were associated with depressive symptoms in a diverse sample of adolescents.

A diathesis-stress model of depression (e.g., Abramson et al. 1988; Metalsky et al. 1993), whereby an underlying cognitive predisposition interacts with stressful life events to increase the likelihood of depressive symptoms, can provide a useful framework for understanding the relationship between explanatory style and IPC. In this model, children’s explanatory style could act as a moderator of the relation between the life stressor (IPC) and depression. Specifically, children who have a pessimistic explanatory style may be particularly vulnerable when faced with IPC and, as a result, may be more likely to experience depressive symptoms.

Interestingly, no previous research has explored the possible role of general cognitive style in understanding the link between IPC and symptoms of depression. Most research on the role of child cognitions as a predictor of the impact of IPC on children is based on Grych and Fincham’s (1990) cognitive-contextual framework and focuses on cognitions specific to the conflict. In this model, children’s affective and behavioral responses to IPC are determined in part by their causal attributions for the conflict, their attributions of responsibility and blame, and their beliefs in their ability to cope effectively with the conflict and its consequences. A number of studies have found support for this model, whereby in the context of IPC, feelings of threat and responsibility are strong predictors of internalizing problems, including depression (Gerard et al. 2005; Grych et al. 2000; Grych et al. 2003).

While focusing on cognitions specific to IPC has proven useful, it is plausible that researchers have much to learn by expanding their focus to general explanatory style. Should general explanatory style interact with IPC, we might be better able to predict which children are likely to be most affected by IPC. Researchers from a social cognitive perspective have begun to look at spontaneous and transient cognitions as related to but qualitatively different from the more enduring belief structures thought to underlie negative cognitions (Ingram et al. 1998). A child who has a pessimistic explanatory style may be more vulnerable to experiencing domain-specific negative attributions of threat and self-blame and increased symptoms of depression when faced with IPC.

Interparental Conflict and Depression Among Diverse Groups of Children

In addition to examining the interplay between cognitions and parenting, we have much to learn by conducting research on IPC with children from low-income and racial/ethnic minority backgrounds, particularly since the research that exists has produced equivocal findings. Some researchers have found that racial/ethnic minority children who report higher levels of IPC also report higher levels of depression and greater feelings of hopelessness (DuRant et al. 1995; Formoso et al. 2000). Similarly, some researchers have found that parenting style plays an important role in determining the effect of IPC on minority children (Dumka et al. 1997; Gonzales et al. 2000). One study that examined the cognitive-contextual framework with diverse groups of children suggested that cognitions mediated the link between conflict and depression among minority as well as majority children (Grych et al. 2000). Other researchers, however, have found that low-income, minority children may be less affected by high levels of IPC than their middle-class, white counterparts, perhaps because of the presence of additional stressors and/or of extended familial support (e.g., Buehler et al. 1998). Thus, it is not yet clear whether IPC affects minority children via the same processes that it impacts majority children or if there may be different factors that are important to consider.

Current Study

In this article, we present data from two time points of a larger longitudinal study investigating the development of depressive symptoms in low-income, urban fifth- and sixth-grade children. From a developmental perspective, this may be an especially critical time to explore the impact of IPC on children. Research suggests that IPC increases over the elementary school years, reaching a peak when children are in middle childhood and early adolescence (Cummings and Davies 2002). Early adolescence is also a challenging time for parents, as they need to adjust to children’s increasing desire for independence and need for support around emerging adolescent issues (Cummings et al. 2000).

In this study, we had three broad aims. Our first aim was to replicate in a low-income, urban sample the findings of previous cross-sectional studies that children’s reports of IPC, parenting, explanatory style, and depressive symptoms are correlated. Second, we aimed to extend the research on IPC and depression by conducting two separate sets of longitudinal analyses. We first hypothesized that the effect of IPC on change in children’s depressive symptoms over time would be partially mediated by increasingly maladaptive parenting behaviors. Specifically, we explored whether decreasing autonomy support, warmth, and structure mediated the hypothesized relation between increasing IPC and increasing depressive symptoms. Cummings and Davies (2002) have articulated that the goal of a process-oriented approach to understanding the impact of IPC on children is to identify, “the dynamic patterns of intra- and extra-organismic causal processes that underlie pathways of development in particular socioemotional contexts over time” (p. 51). It has been further suggested that process oriented research should explore whether changes in family process (in this case IPC and parenting) correspond over time to each other and to changes in children’s well-being (in this case depressive symptoms), while simultaneously exploring possible moderators (Cummings et al. 2000; Fincham et al. 1994). This brings us to our second hypothesis that children’s explanatory style would moderate the impact of change in IPC on change in depressive symptoms. We expected that children who had a pessimistic explanatory style would report more depressive symptoms when faced with increasing IPC than those who had a more optimistic explanatory style. Our third broad aim was to examine in an exploratory manner whether, consistent with the notion of additive risk, parenting and children’s explanatory style would act in concert to predict an increase in depressive symptoms in the context of increasing IPC. Would children who had a pessimistic explanatory style at the start of the study and experienced a negative change in parenting report a greater increase in depressive symptoms than those with an optimistic explanatory style?

Method

Participants

Participants were drawn from a sample of 159 students from two elementary schools located in urban, low-income neighborhoods of a mid-size Northeastern city. We included in this study only those children who identified themselves as living with two parental figures (including at least one biological parent) (n = 109) and who provided data at both Time 1 and Time 2. Of the original 159 students, 88 (36 males, 52 females) qualified for this study.

At Time 1, 34 participants were in the fifth grade, and 54 participants were in the sixth grade. The average age at Time 1 was 10.74 years (SD = 0.69). Participants were racially and ethnically diverse, reflecting the student body of the schools where the percentage of Caucasian students was under 38%. Thirty-one participants identified themselves as Caucasian, 30 as Hispanic, 8 as African-American or Black, 9 as Asian, 7 as Bi-Racial, and 3 as “Other.” Eighty-seven percent of students in school one and 54% of students in school two were considered low-income compared to 27.7% of students in the state (Massachusetts Department of Education 2006).

Procedure

Study participants were initially recruited for a two-year longitudinal study examining depressive symptoms among low-income and minority children. The study was approved by the IRB of the sponsoring university and by the review board of the participating school system. The two waves of data analyzed in this article were separated by one year.

The recruitment procedure at the two schools was identical: We visited the classrooms of all of the fifth- and sixth-grade students, described the study, and provided consent forms for them to take home to their parents. As incentive to increase response rate, we provided a pizza party for those classes that returned the most forms (irrespective of whether or not parents provided consent to participate). In addition, all students who participated in the study were given $10 gift certificates to local shopping malls at each assessment point. We also invited all participants to a free stress-management workshop at the conclusion of the larger study.

At the time of each assessment, participating students completed a series of self-report questionnaires, including those used in the present study, in small groups of 5 to 15 students during school hours. Members of the research staff supervised children, assisting any who had difficulty with the instruments by reading indicated portions to them. Teachers were not present in the room during data collection.

Measures

The Children’s Perception of Interparental Conflict Scale (CPIC; Grych et al. 1992)

The CPIC is a 49-item scale designed to assess children’s exposure to IPC. For the purposes of this study, we used only the Conflict Properties subscale of the CPIC, which includes 18 items measuring the frequency, intensity, and resolution of conflict (e.g., “They may not think I know it, but my parents argue or disagree a lot.”). Participants respond to each statement using a 3-point Likert-type scale (i.e., “true,” “sometimes true,” “false”). Higher scores indicate higher levels of conflict and lower levels of resolution. In this study, internal consistency for the Conflict Properties subscale was good (α = 0.88 at Time 1 and Time 2).

The Child’s Report of Parental Behavior Inventory (CRPBI; Schludermann and Schludermann 1970)

The revised, short form of the CRPBI includes 30 items each on mothers’ and fathers’ parenting behaviors. The three dimensions of parenting measured are: (1) acceptance/rejection, which contains items assessing parental closeness to and acceptance of the child (e.g. “My mother is easy to talk to.”); (2) psychological autonomy/control, comprised of items assessing the extent to which parents use anxiety and/or guilt as a method of discipline (e.g. “My mother is less friendly with me if I do not see things her way.”); and (3) firm/lax behavioral control, which contains items that measure the extent to which parents directly monitor children’s behavior (e.g. “My mother lets me go any place I want without asking.”). Higher scores indicated more maladaptive parenting. Internal consistency for children’s reports of mothers’ and father’s parenting behavior ranged from α = 0.68 to α = 0.93. Reports of mother’s and father’s parenting were highly correlated, at the .05 significance level or better (r = .33 at Time 1 and r = .61 at Time 2 for acceptance; r = .80 at Time 1 and r = .69 at Time 2 for psychological control; and r = .56 at Time 1 and r = .35 at Time 2 for firm control), suggesting a high degree of overlap in perceptions of each parent. These scores were therefore combined for each of the three dimensions. The CRPBI is a widely used and well-established measure of parenting and previous research with a diverse sample provided evidence for cross-ethnic equivalence (Knight et al. 1994).

The Children’s Depression Inventory (CDI; Kovacs 1985)

The CDI is a 27-item symptom checklist that assesses depressive symptoms in children. Children report how often they experienced a variety of symptoms (e.g. sadness, guilt, loss of interest and difficulty concentrating) over the preceding two weeks. Higher scores reflect higher levels of depressive symptoms. Normative data collected in several large scale studies of public school children in both Canada and the United States suggests mean CDI scores between 9.28 and 9.72 (SD between 7 and 7.3) (Finch et al. 1985). In the current study, the internal consistency was good (α = .88 at Time 1, and α = .89 at Time 2).

Children’s Attributional Style Questionnaire (CASQ-R; Thompson et al. 1998)

The CASQ-R uses a forced-choice format to assess a child’s explanatory style. Children are presented with both positive and negative hypothetical events followed by two reasons why the event might have occurred (e.g., “You get a bad grade,” with choices of “I am not a good student,” or “Teachers give hard tests.”). Respondents choose one of the two reasons. Each item taps one of the three dimensions of internality, stability, and globality. The CASQ-R was specifically developed in order to reduce subject burden, at a cost to internal and criterion-related reliability. While somewhat low (α = .50–.73), internal consistency reliability is consistent across studies (e.g., Conley et al. 2001; Robinson et al. 1995) and was 0.64 in this study. Test–retest reliability of the CASQ-R is considered acceptable (r = .71–.80; Seligman et al. 1984).

Results

Descriptive Statistics and Preliminary Analyses

Preliminary analyses included t-tests to determine if there were significant differences on key variables between drop-outs (n = 21) and completers (n = 88). Most students who dropped out of the study were sixth graders at Time 1 who attended a middle school other than the primary public school in the study city at Time 2 and could not be contacted to complete the study. Drop-outs did not differ from completers on Time 1 CDI, CASQ, CPIC, or CRPBI scores, nor were there significant school, grade, or gender differences among completers on any Time 1 variable of interest.

Means, standard deviations, and intercorrelations are shown in Table 1. Mean depression scores on the CDI suggest that this group of early adolescents reported fewer symptoms of depression on average in comparison to other, larger, normative samples. Paired t-tests of difference in CDI and CPIC scores from Time 1 to Time 2 were not significant. All significant correlations were in the expected direction. Specifically, higher levels of depressive symptoms were associated with a more negative explanatory style, higher levels of IPC, rejection/withdrawal, and use of psychological control. IPC was also significantly positively correlated with rejection/withdrawal and use of psychological control. Lax control was not significantly associated with any variable of interest.

Table 1 Means, standard deviations, and intercorrelations for children’s depressive symptoms and cognitive, parent conflict, and parenting variables
Full size table

Parenting as a Mediator of the Relation Between Interparental Conflict and Child Depressive Symptoms

To replicate the finding of previous researchers that parenting acts as a mediator of the relation between IPC and child outcomes using longitudinal data, we conducted a series of multivariate multiple regression analyses following Baron and Kenny’s (1986) guidelines. We used residual change scores from Time 1 to Time 2 for depressive symptoms, IPC, and parenting. Correlations between the residual change score for IPC and change scores for each of the three parenting dimensions indicated that increasing IPC was related to increasing reports of rejection/withdrawal (r = .35, p < .001), but not to increasing psychological control (r = −.05, p = n.s.), nor lax control (r = −.10, p = n.s.). Therefore, only rejection/withdrawal was tested as a mediator.

Results provided support for the mediation model. First, increasing IPC was significantly related to increasing depression scores [F (1, 85) = 4.66, p < .05]. Next, increasing conflict was significantly related to increasing rejection/withdrawal [F (1, 82) = 11.36, p < .001]. Third, rejection/withdrawal residual scores significantly predicted increasing depression scores [F (1, 83) = 15.68 p < .001]. Finally, the relationship between change in conflict and change in depression ceased to be significant when the variable rejection/withdrawal residual was included in the regression equation [F (1, 81) = .83, p = n.s.], whereas the relation between rejection/withdrawal residual scores and change in depression remained significant [F (1, 81) = 11.36, p < .001].

Explanatory Style as a Moderator of the Impact of Interparental Conflict on Depressive Symptoms

To explore whether children’s general cognitive style, represented by scores on the CASQ, would moderate the impact of IPC on depressive symptoms, we again conducted a series of multivariate multiple regression analyses. We first centered scores on the CASQ at Time 1 before testing the significance of the interaction term (Baron and Kenny 1986). Results provided support for the moderator model. After entering explanatory style at Time 1 in Step 1, we next entered residual change on parent conflict in Step 2, followed by the interaction term in Step 3 (see Table 2). The interaction term between explanatory style at Time 1 and change in IPC was significant [F (1, 80) = 4.84, p < .05]. Probing the interaction revealed that, for pessimistic children only, increasing IPC paradoxically predicted decreasing depression scores. Optimistic children’s depression scores were not significantly affected by change in IPC.

Table 2 Hierarchical multiple regression analysis of explanatory style as a moderator of the relation between change in interparental conflict and change in depression
Full size table

Combined Effects of Interparental Conflict, Explanatory Style, and Parenting on Depressive Symptoms

Our final set of analyses explored whether, in the context of increasing conflict, children who reported a pessimistic explanatory style at Time 1 and who experienced decreasing parental warmth and acceptance would report a greater increase in depressive symptoms over time than children who were more optimistic at Time 1 and experienced either no change or an improvement in parenting. We followed Muller et al.’s (2005) outline for the analysis of moderated mediation with explanatory style as a moderator and parenting as a mediator. Results of regression models are presented in Table 3. Results did not support the moderated mediation hypothesis (i.e., the interaction of change in conflict and explanatory style at Time 1 did not significantly predict change in parental warmth/acceptance).

Table 3 Hierarchical multiple regression analysis relating change in conflict, explanatory style and change in parenting to change in depressive symptoms
Full size table

Discussion

The results of this longitudinal study provide support for the role of both parenting and children’s cognitions in understanding the effects of IPC on children’s experience of depressive symptoms. This study adds to the mounting evidence that individual and contextual factors interact to predict the development of depression in children and adolescents (see Zimet and Jacob 2002 for a review). The results help to clarify why IPC increases the risk for depression among only some children by adding to the literature specifying mediators and moderators of the impact of conflict.

The first aim of this study was to replicate in a low-income, urban sample the finding of previous cross-sectional studies that children’s reports of IPC, maladaptive parenting, negative cognitions, and depressive symptoms are correlated. The findings of this study are consistent with several previous studies (DuRant et al. 1995; Formoso et al. 2000; Grych et al. 2000) in that, among low-income children in urban environments, IPC, maladaptive parenting, and a pessimistic explanatory style related to depressive symptoms in ways similar to that found with majority children. Reports of conflict, maladaptive parenting style, and pessimistic explanatory style were associated with higher rates of depressive symptoms among children at both time points. The small sample size of the current study prevented us from exploring whether or not these relations held among specific racial/ethnic groups as compared to others. There is some evidence that the developmental trajectories of depression differ among children of different groups (Kistner et al. 2007) and some of these questions will be addressed in an upcoming paper based on the larger study of which this study was a part. It will be important for future research on the impact of IPC on diverse groups of children to further explore this issue. It has been suggested, for example, that parenting is impacted by cultural context and there is some evidence that low-income Latino parents, for example, may be stricter in response to perceived and real neighborhood threat (Bulcroft et al. 1996; Hill et al. 2003). It is possible that strength of relations between IPC, cognitions and depression might vary by group. It is also possible that moderating variables (e.g., extended familial support) might have specific group effects. The current study used a rough measure of SES and future studies would do well to look more carefully at contextual variables that might moderate the relations found here.

Second, the current study expanded on previous research by using longitudinal data to examine the finding in the literature that a decline in parenting often accompanies increasing IPC and that it is a change in the parent-child relationship that is deleterious to children’s well-being rather than the occurrence of conflict per se. The model was replicated in this sample and the hypothesis was supported. That is, over the course of one year, increasing parent rejection mediated the relation between increasing IPC and an increase in depressive symptoms. The fact that only change in acceptance/rejection was a significant mediator is consistent with a similar study by Sturge-Apple et al. (2006). It may be that increasing parental withdrawal is an ambiguous situation that leaves room for children to make cognitive errors (e.g., self-blame) that contribute to increasing depressive symptoms. Our findings are also consistent with the well-supported notion of spillover effects of IPC on child well-being via parenting. While we did not examine the question of child effects in the current study, previous research exploring the over-time relation between IPC and parent-child conflict found that the direction of effects was from IPC to parent-child conflict rather than the other way around (Almeida et al. 1999). However, in the case of IPC and parenting behaviors, the role that child effects (e.g., temperament, behavior) plays remains a largely unexplored question; one that needs to be addressed in future research.

Third, we hypothesized that children who reported having a pessimistic general cognitive style at the start of the study would experience a greater increase in depressive symptoms in response to increasing IPC than children who reported a more positive explanatory style. In other words, we conceptualized IPC as a stressor that would interact with underlying cognitive predispositions toward depression to predict an increase in symptoms over time. While our findings provide support for the hypothesized moderating role of children’s cognitions in the relation between conflict and depression, they are not consistent with a diathesis-stress model. Rather, we found that pessimistic children who experienced increasing conflict paradoxically reported decreasing symptoms of depression.

It is unclear to what to attribute this unexpected finding. It may be the result of a ceiling effect, since children who reported having a generally negative cognitive style scored significantly higher on depression at both time points than did those with a more positive cognitive style. It may be that children who are already experiencing elevated symptoms of depression do not respond to IPC and associated changes in parenting in expected ways because there are additional factors relevant to these children that were not explored here (e.g., participation in treatment, changes around pubertal status or other life stressors). This explanation may be especially relevant to children living in a low-income urban context given that there are qualitatively different stressors impacting them (McLoyd and Steinberg 1998). Previous research has similarly found that low-income, minority children may be less affected by high levels of IPC than their middle-class, white counterparts (Buehler et al. 1998), and our findings reinforce the need for more research with understudied populations. However, among optimistic and less depressed children, increasing conflict was unrelated to change in depression, suggesting that a positive general cognitive style can serve as a buffer against depression in the context of IPC, a finding with clear clinical implications.

Finally, this study was innovative in exploring whether IPC, parenting, and general cognitive style would interact to predict change in depressive symptoms over one year. The data did not support an interaction between cognitions and change in parenting. This lack of a significant interactive effect is consistent with the results of previous research (Grych 1998), which found a similarly weak correlation between affective qualities of children’s relationship with their fathers and their attributions about IPC. While more research is needed to further explore the interactive effects of parenting and children’s cognitions on child well-being, it may be that, at least in the case of IPC, the effects of parenting and cognitions are more accurately understood as additive rather than interactive. The small sample size of the current study combined with the relatively low levels of depression reported here may also have limited power to detect interaction effects. Furthermore, change in depression and IPC from Time 1 to Time 2 was not significant. We might expect to find significant interaction effects over a longer period of time that would allow for greater change in family dynamics and child well-being.

The results of this study should be considered preliminary for several important reasons. First, all of the data are child self-reported and so provide only one perspective on the child’s environment and psychological well-being. It is possible that, because of a negative response bias, children who were more depressed in the current study saw their parents as less accepting and more controlling and perceived higher levels of IPC than their less depressed peers. The results of this study should ideally be replicated using multiple informants (parent and child) and/or observational measures. Second, while the measures used in the study demonstrated adequate internal consistency, they have not been extensively validated in low-income urban children. This limitation affects most research conducted with understudied populations and further reflects the need to replicate the findings of this study with similar populations, as well as the need to evaluate further established measures with minority children.

Despite these limitations, this study has a number of notable strengths and we are encouraged by several of the findings. In particular, to our knowledge, it is one of only two empirical studies of the effects of IPC and parenting that uses longitudinal data and one of only three that include measures of both parenting and cognitions in the same design. It is also the only study that focuses on general cognitive style rather than on cognitions specific to conflict. Finally, this study is part of a growing body of work exploring predictors of depression among low-income, urban children and is important because of its applicability to both research and clinical work with understudied and underserved populations.

The findings of this study suggest that parenting style and children’s cognitions are important and are appropriate targets of clinical intervention for children from high conflict homes. The finding that children with a more optimistic general cognitive style were buffered against the impact of increasing IPC suggests that programs that focus on cognitions would be beneficial for children experiencing high family conflict. This finding, in combination with the fact that children who had a negative general cognitive style experienced higher rates of depressive symptoms than more optimistic children across time and conditions, also suggests that these children may benefit most from treatment that focuses both on negative cognitions and contextual factors such as parenting.

Furthermore, while a diathesis-stress model was not supported here, it is still possible that a pessimistic general cognitive style serves as a vulnerability factor for children who are witness to escalating IPC. For example, it is possible that a pessimistic general cognitive style may predict which children are likely to experience negative cognitions specific to conflict (e.g., thoughts of self-blame) that have been addressed in previous research (Gerard et al. 2005; Grych et al. 2000; Grych et al. 2003) and, in turn, which children will experience clinically significant depression in response to high conflict divorce. Future studies might include multiple measures of cognitions (both general and those specific to the conflict), as well as additional variables not included in this study, in a longitudinal design that includes data from three or more time points over a period of longer than one year to better test the longitudinal hypotheses proposed and explored here. This would also allow for more sophisticated data analysis (i.e., longitudinal growth curve modeling).

We hope that this study might serve as a step toward developing family interventions that are appropriate for low-income urban pre-adolescents from high conflict homes, as well as their parents. While the findings of this study seem to indicate that IPC, parenting, and cognitions are related to depression in similar ways among this population as in studies with middle-class samples, we would caution against assuming that an intervention addressing these areas would be equally effective with minority families as it would be with majority families. More research needs to be done to replicate these findings and studies need to include other factors that might be specific to low-income urban families and that might affect treatment. For example, a recent study (Cardemil et al. 2008) found that child immigration history is related to more frequent negative automatic thoughts and that immigration history interacts with cognitive style to predict depression.

In sum, findings of this study provide further support for the argument that IPC has few if any unique effects on children’s well-being. Instead, it seems to be negative changes in the home environment that often, though not always, accompany increasing conflict in combination with children’s cognitive style that predict an increase in child depressive symptoms. This study adds to the literature on IPC that suggests that it is not simply the occurrence of conflict in the home, but more importantly, the qualities of the conflict and the impact of conflict on parent-child relationships that interact with individual child level factors to affect children’s well-being.