Coping Responses Moderate Prospective Associations between Marital Conflict and Youth Adjustment

Abstract

Children's engaged coping responses to family conflict were examined as moderators of the prospective association between marital conflict in middle childhood and internalizing and externalizing symptoms in adolescence. Youth and their mothers participated in four waves of data collection (one-year intervals from T1 to T3, five-year interval between T3 and T4). The final analytic sample included 304 participants (51% boys; 66% European American, 34% African American). Participants were approximately 8 and 16 years old at T1 and T4, respectively. A multi-informant, longitudinal design was used to address study aims. Mothers reported on marital conflict (T1 to T3) and externalizing problems (T1 to T4); youth reported on coping responses to family conflict (T3) and internalizing symptoms (T1 to T4). Primary (e.g., problem-solving) and secondary (e.g., cognitive reappraisal) engaged coping were computed as proportion scores (out of all coping responses). Towards identifying unique effects, path models controlled for internalizing when predicting externalizing symptoms, and vice versa. Primary and secondary engaged coping emerged as moderators. In the context of marital conflict, higher levels of secondary engaged coping protected against, whereas lower levels of secondary engaged coping increased risk for, externalizing problems. Conversely, lower levels of primary and secondary engaged coping protected against, whereas higher levels of primary and secondary engaged coping increased risk for, internalizing symptoms in the context of marital conflict. Findings contribute to the small literature on the moderating role of coping in the context of marital conflict, providing further insight into the prediction of unique externalizing and internalizing symptoms.

Children are commonly exposed to marital conflict, which refers to physical, verbal, or psychological aggression between partners (Hamby, Finkelhor, Turner, & Ormrod, 2011; Straus, Hamby, Boney-McCoy, & Sugarman, 1996). The concurrent and prospective risks of marital conflict on children's emotional and behavioral adjustment are well-documented in middle childhood (El-Sheikh, Keiley, Erath, & Dyer, 2013; Kouros, Cummings, & Davies, 2010) and adolescence (Buehler & Gerard, 2013) (for a review see Cummings & Davies, 2010). Yet, research efforts have identified factors (e.g., child characteristics, parenting) that may moderate this association, such that the level of risk for adjustment problems associated with marital conflict varies across children (Cummings & Davies, 2002; 2010; El-Sheikh & Erath, 2011). One potentially important variable that may function as a protective or vulnerability factor is children's coping responses, or “conscious volitional efforts to regulate emotion, cognition, behavior, physiology, and the environment in response to stressful events or circumstances” (Compas et al., 2001, p. 89). The present study examined youths’ voluntary engaged coping with family conflict as a moderator of the prospective association between marital conflict in middle childhood and adjustment in adolescence.

There are several influential theoretical models that propose how marital conflict may directly influence children's psychological adjustment, yet each differs in its relative emphasis on behavioral, cognitive, and emotional domains of influence. At the behavioral level, social learning theory (Bandura, 1977) proposes that children exposed to marital conflict may model conflict behaviors and display aggressive behaviors in subsequent interpersonal interactions. Further, according to the cognitive-contextual framework (Grych & Fincham, 1990), as well as the emotional security theory (Davies Harold, Goeke-Morey, & Cummings, 2002), exposure to destructive forms of marital conflict (intense, frequent, unresolved) may contribute to children's development of negative appraisals and cognitive representations, such as heightened perceptions of threat and self-blame. Marital conflict may also disrupt children's sense of emotional security in the marital relationship (Cummings & Davies, 1994; Davies & Martin, 2014), as evident in negative emotional (fear, distress, threat) and behavioral (avoidance, aggression, intervention) reactivity to marital conflict. Collectively, these frameworks propose that negative cognitive, behavioral, and emotional responses to marital conflict can undermine children's psychological adjustment. Yet, coping responses may moderate the effects of martial conflict, such that some coping strategies provide at least partial protection against adjustment problems, whereas other coping strategies increase vulnerability (Compas et al., 2001; Cummings & Davies, 2010).

Consistent with a well-established taxonomy, coping is conceptualized as voluntary engagement and disengagement coping responses that refer to efforts to manage conflict experiences (Compas et al., 2001). The use of the term “voluntary” to characterize our operational definition of the coping construct is common in much of the literature (Compas et al., 2001; Connor-Smith et al., 2000), though we acknowledge that involuntary responses to stress may also affect coping responses. Voluntary disengagement refers to strategies that are directed away from the stressor, such as avoidance or denial, whereas voluntary engagement refers to strategies that seek to influence stressful events, conditions or emotions, as well as attempts to adapt to the environment (Compas et al., 2001). Voluntary engagement strategies can be further differentiated as primary or secondary control responses (Connor-Smith et al., 2000), which is the focus of the present study. Primary engagement strategies seek to influence events or conditions (e.g., problem-solve) or directly regulate emotions in response to stress, whereas secondary engagement involves attempts to adapt to the environment through cognitive reappraisal. Specific to the context of marital conflict, primary engaged coping strategies may refer to planning a strategy to reduce conflict in the future or telling a parent or friend about feelings related to conflict. Secondary engaged coping may refer to reappraising conflict in relatively benign (or less threatening) terms.

Generally, research has linked engagement strategies (primary and secondary coping) with better adjustment, particularly in relatively controllable situations (Compas et al., 2001; Skinner & Zimmer-Gembeck, 2007). However, a limited number of studies have found that in relatively uncontrollable situations such as marital conflict, primary engagement strategies (i.e., problem-solving) were linked with poorer adjustment, whereas secondary engagement strategies (Jaser et al., 2007) and disengagement strategies were associated with better adjustment (for a reviews see Compas et al., 2001 and Seiffge-Krenke, 2011). Thus, the types and effectiveness of coping strategies may vary depending on the controllability of the stressor, as well as the context (family, academic, peer) of the stressor (Compas et al., 2001; Jaser et al., 2007). Further, as children get older, they develop a better capacity to cope with stress in general and are also able to employ different coping strategies based on the type of stressor (Zimmer-Gembeck & Skinner, 2011).

Indeed, several studies have examined coping strategies as moderators of associations linking marital conflict with internalizing and externalizing problems during childhood or adolescence. Among these studies, there is mixed evidence for the protective and vulnerability functions of both engaged and disengaged coping strategies in the context of marital conflict. For instance, marital conflict was concurrently associated with externalizing behaviors among children who reported low (Rogers & Holmbeck, 1997) and high (Nicolotti et al., 2003) levels of support-seeking, more primary engaged coping, as well as more disengaged coping (Nicolotti et al., 2003). Yet another study found no moderation effects for support seeking or primary engaged coping over time (Shelton & Harold, 2007). Additionally, marital conflict predicted concurrent and later internalizing problems among youth who reported less social support-seeking and more disengaged coping strategies (Nicolotti et al., 2003; Rogers & Holmbeck, 1997; Santiago & Wadsworth, 2009), but not among youth who reported less disengaged coping strategies (Nicolotti et al., 2003; Santiago & Wadsworth, 2009). Further, youth were at increased risk for internalizing problems if they reported high levels of venting negative emotions (Shelton & Harold, 2007) and more primary engaged coping (among boys only; Nicolotti et al., 2003). Of note, in these studies coping responses were assessed in general (e.g., Shelton & Harold, 2007) as well as specifically in the context of marital or family conflict (e.g., Nicolotti et al., 2003; Rogers & Holmbeck, 1997; Santiago & Wadsworth, 2009) with a focus on primary but not secondary engagement coping.

It is also noteworthy that most of these studies examined both internalizing and externalizing outcomes. Results were somewhat similar across internalizing and externalizing outcomes, and these variables were highly correlated with one another, consistent with abundant evidence of moderate to high correlations between internalizing and externalizing symptoms (Youngstrom, Loeber, & Stouthamer-Loeber, 2000). Given the substantial overlap between internalizing and externalizing symptoms, it is possible that the moderating role of coping for each specific outcome is obscured by overlap with the other outcome.

The moderating role of coping in the context of marital conflict may also be obscured by examining primary and secondary engagement coping together rather than separately. Given the relatively uncontrollable nature of marital conflict for children, primary engaged strategies may be less effective at reducing adjustment problems, as found in prior research (Nicolotti et al., 2003; O'Brien, Margolin, & John, 1995; see also Compas et al., 2001). For example, youths’ attempts to problem-solve or intervene with marital conflict may further expose them to parents’ verbal or physical aggression, lead to triangulation in parents’ conflict (e.g., forced to take sides; Buehler & Welsh, 2009), or increase spillover of parents’ conflict to parent-child interactions, elevating youths’ risk for internalizing and externalizing problems (e.g., Jouriles, Rosenfield, McDonald, & Muller, 2014). However, findings are not entirely consistent in the literature (e.g., Shelton & Harold, 2007).

Secondary engaged coping may operate differently. For instance, secondary coping or cognitive reappraisals may help to reduce the influence of marital conflict on externalizing problems by disrupting negative or hostile attributions (Grych & Fincham, 1990) as well as aggressive behaviors (Bandura, 1977) that often stem from marital conflict. Similarly, cognitive reappraisal may help to reduce internalizing symptoms in the context of family stress by reducing negative thoughts and feelings (e.g., Seiffge-Krenke, 2011). However, such cognitive engagement may have the unintended consequence of increasing attention to marital conflict, which may increase internalizing symptoms, particularly if marital conflict persists. Indeed, studies of emotional insecurity in the family have found that insecurity characterized by disengagement was linked with conduct problems, whereas insecurity characterized by preoccupation was linked with internalizing symptoms (Cummings, Koss, & Davies, 2015; Forman & Davies, 2005). Thus, distinguishing primary and secondary engaged coping may clarify the roles of engaged coping as a moderator of associations between marital conflict and adolescent internalizing and externalizing symptoms.

The Present Study

The present study examined whether youths’ engaged coping responses to family conflict moderated prospective associations between parental marital conflict in middle childhood and internalizing and externalizing symptoms in adolescence. We differentiated primary and secondary engaged coping to provide a more nuanced examination of the protective/vulnerability function of different types of engaged coping in the context of a relatively uncontrollable stressor. Further, to account for the total number of coping strategies (engaged, disengaged), we utilized proportion scores (Connor-Smith et al., 2000). Building on prior research examining internalizing and externalizing symptoms separately, we aimed to predict “pure” internalizing and externalizing symptoms by controlling for internalizing symptoms when predicting externalizing symptoms, and vice versa (Keiley, Lofthouse, Bates, Dodge, & Pettit, 2003).

We anticipated that secondary engaged coping would attenuate the association between marital conflict and unique externalizing symptoms because cognitive reappraisal may help to disrupt the mechanisms (e.g., hostile attributions, negative appraisals) by which marital conflict contributes to externalizing problems. Additionally, although secondary engaged coping may attenuate the association linking marital conflict and internalizing symptoms, it is also possible that secondary engaged coping may fail to reduce feelings of depression and anxiety stemming from marital conflict, and instead might draw more attention to marital conflict, increasing the risk of internalizing symptoms. Given the mixed evidence in the literature regarding the moderating role of primary engaged coping (risk vs null effects), we made no hypothesis regarding this coping strategy.

Method

Participants

Children and their parents participated in four waves of data collection from childhood to adolescence as part of the Family Stress Study. There were one-year intervals between T1 to T3 and a five-year interval between T3 and T4. Data collection occurred in 2005 (T1), 2006 (T2), 2007 (T3), and 2012-2013 (T4). Children and their parents were recruited from elementary schools in the Southeastern United States. At T1, children in second- or third-grade from two parent homes (cohabiting for at least two years) were eligible for participation. To reduce potential confounds, children diagnosed with a developmental or learning disability, attention deficit hyperactivity disorder, or a chronic illness were not eligible to participate.

At T1, participants included 251 school-aged children (122 boys, 129 girls; M age = 8.23 years, SD = .72) and their mothers. A majority of children (74%) lived with both biological parents and 26% lived with the mother and stepfather/partner. Of children who participated at T1, 86% (N = 217) participated at T2 (106 boys, 111 girls; M age = 9.31 years, SD = .79), and of those who participated at T2, 84% (N = 194) participated at T3 (92 boys, 102 girls, M age = 10.28 years, SD = .99). Lastly, 83% (N = 199) of children who participated at T1, T2, or T3 participated at T4 (93 boys, 106 girls, M age = 15.27 years, SD = .86). To increase sample size because of a five-year lag between T3 and T4 and a loss of participants, an additional 53 families were recruited to participate at T4 (25 boys, 28 girls, M age = 15.26 years, SD = .90). These participants were recruited from the same school systems as the original sample. Participants who participated for the first time at T1 versus T4 did not differ on representation of sex or race/ethnicity, T4 externalizing problems, or T4 internalizing symptoms. All available data at each wave were used in analyses. The final analytic sample included a total of 304 adolescents.

Reasons for attrition over time included a lack of interest in participating and geographic relocation. Compared with participants who returned, participants who did not return at T2 reported higher depressive symptoms at T1 (M = 7.49, SD = 6.00; M = 10.53, SD = 6.43, respectively; t = 2.72, p < .01). Additionally, compared with participant who returned, participants who did not return at T4 had lower income at T3 (M = 4.25, SD = 1.41; M = 3.28, SD = 1.51, respectively; t = −3.11 p < .01). No other differences on demographics or primary study variables emerged for participants who dropped out across T1 to T4.

Based on T1 data (and T4 for the new subsample), 66% of children were European American and 34% were African American, which is representative of the community. Across T1 to T4, the percent of families reporting the following ranges of annual income was: 13% to 17% for income < $20,000; 31% to 47% for $20,000 - $50,000; and 21% to 26% for $50,000 - $75,000. Finally, families reporting income > $75,000 ranged between 16% and 30%.

Procedure

The study was approved by the university's institutional review board. Consent and assent for participation were obtained during a laboratory visit. Youth and parents also completed questionnaires in separate rooms during the same session.

Measures

Marital conflict (T1-T3)

Mothers completed the well-established Conflict Tactics Scale (CTS2) (Straus et al., 1996). Mothers reported about their own psychological/verbal (e.g., “I insulted or swore at my partner”; 12 items) and physical (e.g., “I pushed or shoved my partner”; 8 items) aggression towards their spouse in the past year, as well as their spouses’ psychological/verbal and physical aggression towards them. Items were rated on a 7-point scale; a score of 0 = did not happen in the past year; 1 = once; 2 = 2 times; 3 = 3 - 5 times; 4 = 6 - 10 times; 5 = 11 - 20 times; 6 = > 20 times. Approximately 12% to 24% of mothers reported some degree of physical aggression during the study waves. Mothers also reported a wide range of psychological/verbal aggression across T1-T3 with scores ranging from 0 to 4.38. Mothers’ reports about their own and partners’ physical (or psychological) aggression were summed at each time point and averaged. Physical (rs = .22 to .55, ps < .01) and psychological (rs = .49 to .63, ps < .001) aggression were stable over time; αs over time for physical and psychological aggression ranged from .83 to .93 and .70 to .90, respectively. Additionally, physical and psychological aggression were highly correlated (r = .64, p < .001). Thus, physical and psychological aggression scores from T1-T3 were averaged to create a single marital conflict score, reflecting cumulative interparental marital aggression in childhood.

Engaged coping (T3)

Youth completed the Responses to Stress Questionnaire- Family Conflict, which has established psychometric properties (RSQ; Connor-Smith et al., 2000). Participants were given a list of items about what people might do, think, or feel when something stressful happens, and were specifically asked about how much they do or feel these things when there are problems in their family such as arguments or disagreements. They rated each item on a 4-point scale (1 = not at all to 4 = a lot). The RSQ is comprised of the following voluntary coping scales (3 items per scale): Problem Solving, Emotional Regulation, Emotional Expression, Cognitive Restructuring, Positive Thinking, Acceptance, Avoidance, Denial, Distraction, and Wishful Thinking. These scales make up three voluntary coping factors, including Primary and Secondary Engagement and Disengagement (Connor-Smith et al., 2000).

To validate the voluntary coping factor structure of the RSQ with the current sample, confirmatory factor analyses were conducted. Consistent with Connor-Smith et al. (2000), three factors were derived: primary engagement – problem solving, emotional regulation, and emotional expression (factor loadings ranged from .76 to .81); secondary engagement – cognitive restructuring, positive thinking, and acceptance (factor loadings ranged from .41 to .77); and disengagement – avoidance, denial, wishful thinking (factor loadings ranged from .66 to .76). Inconsistent with Connor-Smith et al. (2000), in which distraction loaded on secondary rather than disengaged coping, our analyses revealed that the distraction scale loaded similarly on secondary engaged (.75) and disengaged (.73) coping. Thus, it is unclear where the distraction scale best fits in our sample. Further, given the conceptual and empirical overlap of the distraction scale with multiple coping dimensions (e.g., Compas et al., 2001; Connor-Smith et al., 2000), distraction was not included as part of either coping factor or the total coping score.

Proportion scores for primary and secondary engaged coping responses were created by dividing the mean of primary or secondary engaged coping responses by the sum of the means of all voluntary coping responses (i.e., means of primary and secondary engagement and disengagement coping). This approach accounts for all voluntary coping strategies and provides an estimate of the relative degree to which a particular voluntary coping strategy is used; αs for primary and secondary engaged and disengaged coping were .82, .70, and .76, respectively. Coping was only assessed during late childhood (T3). The proportion of primary and secondary engaged coping will be referred to as primary and secondary engaged coping hereafter.

Externalizing problems (T1-T4)

Mothers reported on youth externalizing symptoms using the Personality Inventory for Children II (PIC2; Lachar & Gruber, 2005). The Scale has 24 items that assess youth impulsivity, aggression, disruptive behavior, noncompliance, and delinquency. Items were rated as true or false. Externalizing symptoms were highly stable over time (rs ranged from .49 to .95, ps < .001; αs ranged from .76 to .97 across all waves). Percentages of youth with borderline or clinical levels of externalizing symptoms (T > 60) ranged from 7% to 9% across waves. Externalizing symptoms at T1-T3 were averaged to create a childhood externalizing score that was used to control for autoregressive effects. In addressing our research questions, T4 externalizing scores were examined as outcomes. Per standard guidelines for longitudinal analyses, raw scores for externalizing behaviors were used.

Internalizing symptoms (T1-T4)

Youth completed the well-established 27-item Children's Depression Inventory (CDI; Kovacs, 1992); one item regarding suicidal ideation was excluded. Participants were asked to select one of three possible options per item (e.g., “I am sad once in a while, I am sad many times, or I am sad all the time”), which were rated on a 3-point scale (0 = absence of symptom to 2 = definite symptom). The CDI scores were stable (rs = .37 to .42, ps < .001) and internally consistent (αs = .83 to .95) across time. The percentage of youth with clinically significant levels of depression (scores > 20) ranged from 2% to 7% across waves.

Additionally, youth completed the reliable and valid Revised Children's Manifest Anxiety Scale (RCMAS; Reynolds & Richmond, 1978) at each study wave. Twenty-eight items (e.g., “I worry a lot of the time”) assessing physiological anxiety, social concerns, and worry were rated on a 2-point scale (0 = no to 1 = yes). The RCMAS was stable (rs = .37 to .56, ps < .001) and internally consistent (αs = .90 to .92) across study waves. The percentage of youth with high levels of anxiety (scores > 2 SDs from the mean) ranged from 2% to 5% across waves.

Youths’ reports on the CDI and RCMAS were moderately correlated at each study wave (rs ranged from .44 to .52, ps < .001); these scores were standardized and averaged to yield a composite score for youth internalizing symptoms at each wave. Internalizing scores at T1-T3 were averaged to create a childhood internalizing score that was used to control for autoregressive effects. T4 internalizing scores were examined as outcomes in the models.

Controls

Some variables were associated with primary model variables and were controlled in analyses: youth sex (girls = 0; boys = 1); age (T1); ethnicity (0 = European American; 1 = African American); and annual family income (income; T1-T3); the latter was scored on a 6-point scale (1 = less than $10,000 to 6 = more than $75,000).

Plan of Analysis

To reduce outlier effects, values of study variables that exceeded three standard deviations (SDs) were recoded as the highest observed value within three SDs. In total, 22 values were recoded, including six values for psychological aggression, four for physical aggression, three for childhood externalizing, six for childhood internalizing, and three for adolescent internalizing. None of the study variables were skewed.

Path models were fit in AMOS (Arbuckle, 2012) to examine whether history of exposure to marital conflict, either primary and secondary engaged coping,¹ and the interaction between marital conflict and coping predicted change in externalizing and internalizing symptoms over time, controlling for autoregressive effects. Full information maximum likelihood estimation was used to handle missing data. Externalizing and internalizing symptoms in adolescence (T4) were examined separately, controlling for the other (composite of T1-T4) to predict pure externalizing and pure internalizing symptoms.² To remove shared variance of prior and concurrent symptoms of the alternative outcome, a composite of scores across all waves were created for each outcome. Control and predictor variables were centered for analyses. Control variables, which included youth sex, ethnicity, age (T1), and income (T1-T3) were allowed to predict the primary study predictors (i.e., marital conflict, engaged coping, their interaction, autoregressive effects); significantly correlated controls were allowed to covary. Although not a primary focus of this study since we utilized proportion scores for primary and secondary engaged coping, disengaged coping was also examined as a moderator of the associations between marital conflict and adolescent adjustment. Such findings may be useful for comparisons with prior studies examining disengaged coping in the context of marital conflict.

Significant interactions were plotted at high (+1 SD) and low (−1 SD) levels of the predictor and moderator (Aiken & West, 1991). The Preacher, Curran, and Bauer's (2006) interaction utility was used to plot interactions using estimates from the fitted models.

Results

Preliminary Analyses

Descriptive statistics and correlations are presented in Table 1. Marital conflict in childhood (T1-T3) was positively correlated with externalizing problems in childhood (T1-T3) and adolescence (T4). However, marital conflict was not correlated with internalizing symptoms in childhood or adolescence. Further, externalizing problems in childhood and adolescence were correlated; a similar association was observed for internalizing symptoms. Further, primary and secondary engaged coping were negatively associated.

Older children reported more secondary engaged coping at T3 and lower levels of childhood internalizing symptoms. Income was negatively correlated with marital conflict and adolescent externalizing problems. T-tests were conducted to examine sex and ethnic differences across study variables. Compared with girls, boys reported lower levels of internalizing symptoms in adolescence (M_boys = −.23, SD = .89; M_girls = .21, SD = .94; t = 3.72, p < .001).

Predicting Adolescent Externalizing Problems

Primary engaged coping

This model was a good fit to the data, χ² = 8.13, df = 8, p = .42; χ²/df = 1.02; RMSEA = .01 ns, and explained 33.2% of the variance in externalizing problems. The autoregressive effect was significant (B = .71, SE = .08, β = .57, p < .001). However, the independent and interactive associations between marital conflict and primary engaged coping did not predict externalizing problems in adolescence.

Secondary engaged coping

The model (Figure 1) was a good fit to the data, χ² = 8.64, df = 8, p = .37; χ²/df = 1.08; RMSEA = .02 ns and accounted for 35.1% of the variance in externalizing problems. As shown in Figure 1, the autoregressive effect was significant. Controlling for the autoregressive effect and demographic variables, marital conflict and secondary engaged coping did not individually predict externalizing problems in adolescence. However, marital conflict interacted with secondary engaged coping to predict change in externalizing problems over time.

Figure 1.

Examination of secondary engaged coping as a moderator of the association between marital conflict and adolescent externalizing problems. Unstandardized and standardized coefficients (in parentheses) are provided. Child gender, age (T1), ethnicity, and income (T1-T3) were allowed to predict endogenous variables (not pictured). Exogenous variables were allowed to correlate as were residual variances among endogenous variables. Statistically significant lines are solid, whereas non-significant lines are dotted. T1-T3 = average of data collected at Time 1, Time 2, and Time 3; T3 = data collected at T3; T4 = data collected at T4.

*p < .05. ***p < .001.

As shown in Figure 2, tests of simple slopes revealed that marital conflict in childhood was positively associated with externalizing problems in adolescence among youth who reported lower levels of secondary engaged coping (at the non-significant trend level; B = .13, SE = .07, p < .10). Among youth who reported higher levels of secondary engaged coping, marital conflict was negatively associated with externalizing problems (B = −.17, SE = .08, p < .05). At higher levels of marital conflict, predicted means of adolescent externalizing problems were higher for those reporting lower (M = 5.10) than higher (M = 3.41) levels of secondary engaged coping.

Figure 2.

Mother-reported marital conflict in childhood (T1-T3) predicting mother-reported externalizing problems (T4), controlling for autoregressive effects, at lower and higher levels of secondary engaged coping with family conflict (T3).

Secondary analyses

Disengaged coping also served as a moderator of the association between marital conflict and externalizing problems (B = 2.58, SE = 1.01, β = .19, p < .05). Tests of simple slopes revealed a positive association between marital conflict and externalizing problems among adolescents who reported high levels of disengaged coping (B = .12, SE = .06, p < .05), but a negative association among adolescents who reported low levels of disengaged coping (B = −.16, SE = .08, p < .05).

Predicting Adolescent Internalizing Symptoms

Primary engaged coping

This model yielded an acceptable fit to the data, χ² = 17.47, df = 8, p < .05; χ²/df = 2.18; RMSEA = .06 ns and explained 13.6% of the variance in adolescent internalizing symptoms. The autoregressive effect was significant (B = .31, SE = .08, β = .29, p < .001). Although main effects of marital conflict and primary engaged coping did not emerge, an interaction emerged predicting adolescent internalizing symptoms (B = .49, SE = .22, β = .18, p < .05). Tests of simple slopes revealed that marital conflict was associated with more internalizing problems among adolescents who reported high (B = .04, SE = .01, p < .001) but not low (B = −.01, SE = .01, p = .31) primary engaged coping.

Secondary engaged coping

The model (Figure 3) yielded an acceptable fit to the data, χ² = 18.12, df = 8, p < .05; χ²/df = 2.27; RMSEA = .07 ns and accounted for 17.5% of the variance in adolescent internalizing symptoms. The autoregressive effect of childhood internalizing symptoms was significant. Marital conflict and secondary engaged coping did not individually predict internalizing symptoms in adolescence (Figure 3). However, marital conflict interacted with secondary engaged coping to predict change in internalizing symptoms over time.

Figure 3.

Examination of secondary engaged coping as a moderator of the association between marital conflict and adolescent internalizing symptoms. Unstandardized and standardized coefficients (in parentheses) are provided. Child gender, age (T1), ethnicity, and income (T1-T3) were allowed to predict endogenous variables (not pictured). Exogenous variables were allowed to correlate as were residual variances of endogenous variables. Statistically significant lines are solid, whereas non-significant lines are dotted. T1-T3 = average of data collected at Time 1, Time 2, and Time 3; T3 = data collected at T3; T4 = data collected at T4.

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

As shown in Figure 4, and similar to the pattern of associations found with primary engaged coping, marital conflict in childhood was positively associated with internalizing symptoms in adolescence among youth who reported higher levels of secondary engaged coping (B = .05, SE = .01, p < .001), but not among youth who reported lower levels of secondary engaged coping (B = −.02, SE = .01, p < .10). At higher levels of marital conflict, predicted means of adolescent internalizing symptoms were higher for those reporting higher (M = .51) than lower (M = −.23) levels of secondary engaged coping.

Figure 4.

Mother-reported marital conflict in childhood (T1-T3) predicting adolescent-reported internalizing symptoms (T4), controlling for autoregressive effects, at lower and higher levels of secondary engaged coping with family conflict (T3).

Secondary analyses

Disengaged coping also moderated the association linking marital conflict with internalizing symptoms (B = −.87, SE = .20, β = −.32, p < .001). Tests of simple slopes revealed that marital conflict was associated with more adolescent internalizing symptoms among youth who reported low disengaged coping (B = .07, SE = .01, p < .001). Conversely, among adolescents who reported high disengaged coping, a negative association emerged between marital conflict and adolescent internalizing symptoms (B =−.03, SE = .01, p < .01).

Discussion

Although the association between marital conflict and youths’ emotional and behavioral maladjustment is well-documented, studies have revealed moderators that elucidate for whom and under which conditions these associations are stronger or weaker (Cummings & Davies, 2010). The present study examined primary and secondary engaged coping with family conflict as moderators of the associations between marital conflict during middle childhood and adjustment problems in adolescence, controlling for earlier levels of adjustment problems as well as the alternative adjustment outcome. Utilizing a multi-informant and longitudinal design, secondary but not primary engaged coping in late childhood moderated the prospective association between marital conflict in childhood and externalizing problems in adolescence, and primary and secondary engaged coping moderated the link between marital conflict and internalizing symptoms. Results revealed two different patterns of associations in the prediction of unique externalizing and internalizing symptoms. Lower levels of secondary engaged coping exacerbated the association between marital conflict and externalizing problems (trend level), whereas higher levels of secondary engaged coping protected against higher externalizing problems over time. In contrast, lower levels of primary and secondary engaged coping protected against higher internalizing symptoms in the context of marital conflict, whereas higher levels of primary and secondary engaged coping exacerbated the link between marital conflict and internalizing symptoms over time.

Secondary engaged coping responses, or adapting to the situation through cognitive reappraisal, may protect against externalizing problems in the context of marital conflict because positive reappraisals, as assessed in the present study, are typically associated with better youth adjustment (e.g., Compas et al., 2001; Connor-Smith et al., 2000). It is also possible that this adaptive coping strategy may generalize to other contexts of stress beyond the family or prevent the spillover of family conflict to other domains of functioning (Erath, Bub, & Tu, 2014), reducing externalizing behaviors. In addition, more secondary engaged coping with marital conflict may be protective against externalizing problems specifically because cognitive reappraisal might prevent or disrupt the mechanisms by which marital conflict contributes to externalizing problems. That is, cognitive reappraisal and positive thinking might reduce the risk of negative behavioral and cognitive processes (negative/hostile appraisals or attributions) linked to marital conflict (Grych & Fincham, 1990).

In contrast, less secondary engaged coping functioned as a risk factor for externalizing behaviors associated with marital conflict. Similarly, towards clarifying inconsistencies in the literature, additional analyses conducted with disengaged coping revealed that higher levels of disengaged coping increased risk for, whereas lower levels of disengaged coping protected against, externalizing problems in the context of marital conflict. These findings are consistent with some evidence in the literature that more disengaged (e.g., avoidant) coping exacerbated, whereas less disengaged coping protected against, the effects of marital conflict on externalizing problems (Nicolotti et al., 2003). Similarly, disengagement from the family system has been linked with more conduct problems among young adolescents (Cummings et al., 2015). Thus, these findings appear to suggest that more engaged coping strategies, specifically secondary engagement, can help to disrupt the mechanisms through which martial conflict contributes to externalizing problems.

Whereas cognitive reappraisal about family conflict decreased adolescents’ vulnerability to externalizing behaviors, secondary engaged coping increased adolescents’ vulnerability to internalizing symptoms in the context of marital conflict, which is inconsistent with some findings in the literature revealing beneficial effects of secondary coping in the context of other forms of family stress (e.g., parental depression, illness; Seiffge-Krenke, 2011). Youths’ attempts to reframe marital conflict in more positive terms may draw more awareness and attention to parental marital conflict, resulting in more internalizing symptoms. Indeed, there is some evidence that emotional insecurity within the family characterized by preoccupation predicted increased risk for internalizing symptoms (Cummings et al., 2015). Our findings suggest that even though cognitive reframing may disrupt the mechanisms by which marital conflict influences externalizing problems, engagement coping through cognitive reframing may not reduce youths’ sense of helplessness or lack of control in alleviating marital conflict, thus contributing to feelings of depression and anxiety (O'Brien et al., 1995).

Conversely, less secondary engaged coping may protect against increased internalizing symptoms in the context of marital conflict because less secondary engagement might indicate less attention to the problem (e.g., Rhoades, 2008). The extent to which less secondary engaged coping functions as a risk or protective factor might be explained by what youth do or think when they are less cognitively engaged with marital conflict (e.g., distract themselves with something else; Rhoades, 2008), which was not examined in this study.

Further, consistent with the literature (Nicolotti et al., 2003) and the findings with secondary engaged coping, primary engaged coping moderated the association between marital conflict and internalizing symptoms such that a stronger association emerged among adolescents who reported more primary engaged coping. In the context of marital conflict, a relatively uncontrollable stressor, primary engagement strategies (e.g., problem-solving) may give youth a false sense of control, which can contribute to depression and anxiety when their involvement does little to actually change the situation (Compas et al., 2001). Similarly, additional analyses examining disengaged coping revealed that lower levels of disengaged coping exacerbated and higher levels of disengaged coping attenuated the association between marital conflict and internalizing symptoms. These findings suggest that less engaged or more disengaged coping may be protective against internalizing symptoms in the context of marital conflict.

The different patterns of associations found for youth adjustment outcomes may be related to different processes by which primary and secondary engaged coping moderates the link between marital conflict and internalizing compared to externalizing symptoms. It is also possible that these differences may be due in part to different informants for these constructs. Further research is needed to replicate and identify processes that may help to explain these findings.

Our aforementioned interpretations of study findings regarding the moderating role of engaged (and disengaged) coping in the context of marital conflict should be considered tentative in light of the evidence in the literature. Some differences in findings from the present study compared with the literature may be because of how coping was assessed (individual scale scores versus proportion score), the context in which coping responses were measured (general, marital or family conflict), and the examination of internalizing symptoms with or without control for externalizing symptoms, and vice versa.

Significant bivariate associations emerged linking marital conflict with more externalizing problems in childhood and adolescence; however, limited main effects emerged in path models, which are not inconsistent with existing literature (Cummings & Davis, 2010). This may be a result of the relatively robust autoregressive effects of childhood externalizing symptoms on adolescent symptomatology, as well as the prediction of unique externalizing symptoms. As expected, there was moderate overlap between externalizing and internalizing symptoms over time, consistent with the literature (Youngstrom et al., 2008).

Although replication and further research will be important, our findings may have implications for clinicians or intervention programs targeting marital conflict in families. Intervention programs have aimed to promote effective coping strategies, such as regulating emotions and behaviors, among children exposed to interparental conflict with limited success (Cummings, Faircloth, Mitchell, Cummings, & Schermerhorn, 2008; Mitchell, McCoy, Cummings, Faircloth, & Cummings, 2009). Our findings suggest that helping youth to reframe their thinking about the negative circumstances in more positive or benign ways might reduce the risk of externalizing problems. However, given that these secondary (and primary) coping strategies may function as a risk factor for internalizing symptoms, the inclusion of other approaches may be needed to help reduce symptoms of depression and anxiety.

There are several limitations of the present study and directions for future research. Although the present study employed a multi-informant design, we did not have multiple informants reporting on all constructs. Additionally, the community sample is characterized by mostly non-clinical levels of adjustment problems, which are comparable to normative samples from middle childhood to adolescence (e.g., Kovacs, 1992; Lachar & Gruber, 2005; Reynolds & Richmond, 2008), and modest proportions of primary and secondary engaged coping responses. Further, although a wide range of marital conflict was reported, physical aggression was not highly prevalent across all waves, similar to rates reported in other community samples (e.g., Kelly & El-Sheikh, 2011; Slep & O'Leary, 2005; for national rates see Black et al., 2011). Thus, the results of the present study will not necessarily generalize to higher-risk populations. In addition, whereas marital conflict was represented by a composite score of cumulative conflict across three waves, coping with family conflict was only available at the third wave. Although coping is relatively stable in late childhood and early adolescence (Valiente et al., 2014), future studies should consider how coping may change over time (e.g., Goeke-Morey, Papp, & Cummings, 2013), and how changes in coping may function as a moderator in the context of marital conflict. The examination of youths’ perceptions of and actual exposure to marital conflict as well as specific domains of marital conflict (e.g., finances, in-laws, child-rearing) in future work would help to further elucidate links with youths’ adjustment.

Although we examined coping as the moderator of associations, alternative moderator models are possible, such as involuntary physiological responses to stress (El-Sheikh & Erath, 2011) or cognitive appraisals (e.g., perceptions of threat or blame; Grych & Fincham, 1990), and the consideration of the interplay among these responses to stress should be considered (Erath & Tu, 2014; Mueller, Jouriles, McDonald, & Rosenfield, 2014). Further, we examined youths’ coping responses to family conflict more broadly, such as disagreements, arguments, or problems in the family (potentially including youth conflict with parents and/or siblings), rather than coping with marital conflict exclusively, which should be considered in future studies. Further examining the cognitive, behavioral, and emotional mechanisms by which coping may function as a vulnerability/protective factor in the context of marital conflict may be a worthwhile future direction, potentially helping to inform intervention programs.

Despite these limitations, this multi-informant, longitudinal study advances our understanding of which adolescents are more or less susceptible to adjustment problems over time in the context of marital conflict. Findings extend the literature by taking a context-specific approach to assess coping with family conflict, as well as by examining internalizing and externalizing symptoms independent of one another. Overall, the findings from this study indicate that the protective and vulnerability function of primary and secondary engaged coping in the context of marital conflict may depend on the outcome of interest.