Implications of Interparental Conflict for Adolescents’ Peer Relationships: A Longitudinal Pathway through Threat Appraisals and Social Anxiety Symptoms

Abstract

The goal of this study was to broaden the developmental understanding of the implications of interparental conflict (IPC) and threat appraisals of conflict for adolescents’ relationships with peers. Guided by the cognitive contextual framework and evolutionary perspectives, we evaluated a developmental model in which adolescents who are exposed to IPC perceive these conflicts as threatening to their well-being or that of their family. In turn, threat appraisals of IPC increase risk that adolescents experience worries and fears about the peer context (i.e., social anxiety), leading to decreased support from friends and increased feelings of loneliness and engagement with antisocial peers. Autoregressive analyses were conducted with a sample of 768 two-parent families across four measurement occasions. Exposure to IPC was related to increases in youths’ perceived threat, which increased their risk for social anxiety symptoms. Consistent with our hypothesis, heightened social anxiety symptoms undermined youths’ subsequent functioning in the peer context. Specifically, youth with greater adolescent social anxiety symptoms experienced increased feelings of loneliness and decreased perceptions of friendship support. Significant indirect effects were substantiated for adolescent loneliness and friendship support. Findings did not vary as a function of adolescent gender. The findings highlight the enduring implications of IPC and threat appraisals of IPC for youths’ functioning, which can be expanded beyond broad measures of youth psychopathology, and the critical role of social anxiety symptoms as an explanatory mechanism in this process.

Introduction

Considerable research now documents the implications of contentious and poorly resolved interparental conflicts (IPC; e.g., hostility, shouting, name-calling, and physical aggression) for a broad range of youth difficulties (Harold & Sellers, 2018). In particular, important theoretical frameworks, such as the cognitive-contextual framework (Grych & Fincham, 1990) and emotional security hypothesis (Davies & Martin, 2013) have supported research illustrating how youths’ evaluations of conflict represent salient risk mechanisms explaining why exposure to IPC affects emotional and behavioral functioning among children and adolescents. To date, however, much of the research on processes involving IPC and youths’ subjective evaluations of IPC have focused primarily on internalizing and externalizing symptoms or a narrow range of youth psychological and behavioral indicators.

Studying IPC and evaluations of IPC (during adolescence in particular) calls for consideration of a broader developmental framework. Specifically, conceptualizations of the potential consequences of IPC for youth need to be expanded to developmentally salient outcomes that are not encapsulated in measures of psychological functioning. For instance, a recent study examined the effects of IPC on adolescent substance use—an outcome with robust consequences for health, safety, and well-being well into adulthood (Fosco & Feinberg, 2018). Findings indicated that adolescents who perceive IPC as more threatening are more likely to escalate cigarette use through high school. Other work has shown that IPC and perceived threat of IPC are associated with decreases in adolescents’ self-efficacy, which undergoes substantial evolvement during this developmental period (Fosco & Feinberg, 2015). Moreover, other research has as illuminated processes linking IPC to adolescents’ friendship quality and social competence via youths’ emotional security (Davies, Martin, & Cummings, 2018).

The overarching goal in this study was to expand this body of research examining the implications of IPC for specific, developmentally salient, and understudied youth outcomes, thereby broadening the developmental understanding of the potential consequences of IPC. Important questions remain about how and why youths’ evaluations of IPC, and particularly perceived threat, might explain the effects of IPC on youths’ peer relationships. Guided by the cognitive-contextual framework (Grych & Fincham, 1990; Fosco, DeBoard, & Grych, 2007) and evolutionary perspectives of social anxiety (Gilbert, 2014), we tested a developmental model hypothesizing that adolescents who are exposed to more frequent, hostile, and poorly resolved IPC are more likely to perceive these conflicts as threatening to their well-being or that of their family (Grych & Fincham, 1993). We expected that greater perceived threat of IPC in the family context has consequences for increases in adolescent social anxiety symptoms, and in turn, these symptoms contribute to decreased perceptions of friendship support and increased loneliness and engagement with antisocial peers.

Mechanisms Explaining the Implications of IPC for Peer Relationships

Broadly speaking, family relationships provide the important foundation for the development of interpersonal skills that youth bring into the peer context, and thus, are critical influences on the quality of adolescents’ relationships with peers (Buehler, Benson, & Gerard, 2006; Buehler, Franck, & Cook, 2009; Mak, Fosco, & Feinberg, 2018). Theories of coercive family interactions have posited transmission processes of hostile family interactions to youth behaviors and interaction styles outside of the family context (Patterson, 2005). However, empirical work has provided little evidence to support direct influences of IPC on peer relationship outcomes (e.g, Johnson, La Voie, & Mahoney, 2001; Underwood, Beron, Gentsch, Galperin, & Risser, 2008).

Rather, theory and research suggest that the influences of IPC on youth’s maladaptive social skills and lower quality relationships (Buehler et al., 2009) operate through associations with youth’s individual and relational characteristics (Lindsey, MacKinnon-Lewis, Campbell, Frabutt, & Lamb, 2002; Schwarz, Stutz, & Ledermann, 2012; Stocker & Youngblade, 1999). For example, the development of the early adult romantic relationships (DEARR) model (Bryant & Conger, 2002) suggests that experiences in the family shape youth’s cognitions or beliefs about relationships which, in turn, influence youths’ romantic interactions and the quality of these relationships. Extending this model to peer relationships in general, and applied to experiences of IPC, we can postulate that youth who witness contentious and hostile interparental interactions are more likely to form beliefs that interpersonal relationships are characterized by discord, coercion, and dissatisfaction. These beliefs, in turn, undermine how youth approach and behave in interpersonal relationships outside of the family context and contribute to lower quality relationships (Bryant & Conger, 2002).

Similarly, two theories—the cognitive contextual framework (Grych & Fincham, 1990) and the emotional security hypothesis (Davies & Martin, 2013)—specify that youths’ subjective evaluations of IPC explain the cascading effects of IPC on youth functioning (Davies & Martin, 2013; Davies & Woitach, 2008; Fosco & Grych, 2007; Grych & Cardoza-Fernandez, 2001; Grych & Fincham, 1990). Subjective evaluations of IPC are discussed in terms of cognitive appraisals of IPC (e.g., perceived meaning and implications of IPC) in the cognitive contextual framework (Grych & Fincham, 1990) and regulatory/response patterns that help identify and neutralize threats to security in the emotional security hypothesis (Davies & Martin, 2013). Central to the cognitive contextual framework are threat appraisals which encompass youths’ evaluations that IPC poses a risk to youths’ or their family’s well-being. Concerns about IPC might be general or they might be specific to worries that IPC might lead to divorce, violence, or a redirection of hostility towards the adolescents. Over time, youth who are chronically exposed to IPC may develop persistent concerns about their safety and well-being which are generalized to other interpersonal interactions within and outside the family (Grych et al., 2000; Grych, Raynor, & Fosco, 2004).

A handful of studies have found that youths’ emotional reactivity in response to IPC is associated with the qualities of youths’ friendships and their perceptions of friendship support (Buehler et al., 2009; Cook, Buehler, & Blair, 2013; Davies, Martin, & Cummings, 2018). No studies to our knowledge, however, have tested youths’ cognitive appraisals of conflict, namely perceived threat, as potential mechanisms. Additionally, it remains unclear why and how perceived threat in response to IPC might generalize to adolescents’ peer relationships. In order to narrow this gap in understanding, we drew upon evolutionary perspectives (Gilbert, 2014) to examine social anxiety symptoms as a specific and developmentally salient linking mechanism in this process. Social anxiety symptoms are defined as avoidance and fear of social situations and interactions where negative evaluation might occur (Rapee & Heimberg, 1997).

An accumulating number of cross-sectional and longitudinal studies show that greater threat appraisals of IPC are associated with elevated adolescent risk for anxiety and internalizing disorders (Fosco & Grych, 2007; Gerard, Buehler, Franck, & Anderson, 2005; Grych et al., 2000; Grych, Harold, & Miles, 2003; Jouriles, Spiller, Stephens, McDonald, & Swank, 2000; Kerig, 1998; Kim, Jackson, Conrad, & Hunter, 2008; Mueller, Jouriles, McDonald, & Rosenfield, 2015). These findings hold even after accounting for other cognitive appraisals, such as self-blame (Fosco & Bray, 2016; Fosco & Grych, 2008; Gerard et al., 2005; Grych et al., 2003; Schlomer, Cleveland, Vandenbergh, Fosco, & Feinberg, 2015). Most the work in this area has focused on early to middle childhood, but recent studies have replicated these findings in early to middle adolescence, suggesting that IPC and threat appraisals remain salient risks during the adolescent developmental period (Fosco & Feinberg, 2015; Fosco & Lydon-Staley, 2017). However, consideration of how IPC and threat appraisals are associated with youth’s experiences and relationships with peers necessitates examinations of mechanisms that encompass skills and abilities to navigate interactions and relationships outside the family context, such as social anxiety symptoms. In comparison to general anxiety and internalizing problems, social anxiety is context-specific and reflects the increasing importance that adolescents attribute to peers and how they are perceived in the social realm (Collins & Laursen, 2004). Social anxiety symptoms increase during early adolescence (Weems & Costa, 2005; Westenberg, Drewes, Goedhart, Siebelink, & Treffers, 2004), and during early to mid-adolescence, clinically relevant levels of symptoms manifest, with social anxiety becoming the third most prevalent mental health disorder during this developmental period (Kessler et al., 2012; Merikangas et al., 2010).

Evolutionary perspectives related to attachment theory (Bowlby, 1973; Gilbert, 2001; 2015) highlight the central role of family relationships in shaping youths’ cognitive frameworks, or “schemas,” of interpersonal relationships. These schemas influence youth’s understanding, expectations, and interpretations of themselves and others in relationships within and outside family contexts. Schemas are an adaptive tool for self-protection; facilitating self-monitoring and mobilization patterns (i.e., specific behavioral responses) that reduce youth exposure to potential harm (broadly defined) (Davies & Sturge-Apple, in press; Gilbert, 2014; 2015). Framed within this perspective, youth who are frequently exposed to hostile IPC and who appraise this conflict as more threatening will be more likely to develop negative cognitive schemas that interpersonal relationships and interactions are competitive and conflictual (Gilbert, 2015). Expectations that interpersonal relationships are competitive and coercive are central to social anxiety and contribute to hypervigilance to potential “harm,” conceptualized in this perspective as rejection, negative evaluations, and loss of social rank. Accordingly, hypervigilant individuals face an increased likelihood that they will become preoccupied with how they present themselves in social situations, evaluate themselves critically, and anticipate negative evaluation from others. Moreover, they are at increased risk for employing submissive or avoidant behaviors to reduce exposure to potential harm (Gilbert, 2014). Thus, we hypothesize that adolescents who are frequently exposed to IPC and appraise this conflict as threatening are “primed” to appraise other interpersonal interactions as competitive and threatening and, in response, will be hypervigilant to how they behave and are perceived and will avoid social interactions as a method of self-preservation.

In turn, mounting evidence suggests that heightened social anxiety symptoms are associated with declines in adolescents’ abilities to engage successfully in the peer context (Alden & Taylor, 2004). Evolutionary perspectives of social anxiety suggest that the mobilization strategies that protect socially anxious individuals from threat (i.e., submissiveness, fear, avoidance) undermine the potential resources that they can offer others, making socially anxious individuals’ unattractive affiliative partners (Gilbert, 2014). This is particularly likely during adolescence when peer affiliation is highly salient. Consequently, social anxiety limits opportunities for youth to develop high quality friendships, contributing to greater risk for peer alienation (Tillfors, Persson, Willen, & Burk, 2012), lower support (La Greca & Lopez, 1998), and greater loneliness (Mak et al., 2018; Prinstein & La Greca, 2002; Stoeckli, 2009). Recent formulations of evolutionary perspectives of social anxiety also suggest, however, that a sole focus on avoidance or ineffective social behaviors “fails to capture the full range of social behavior associated with SA [social anxiety]” (Brosnan, Tone, & Williams, 2017, pp. 105). Rooted in social anxiety is preoccupation with avoiding competitive situations where negative evaluation and loss of social rank might occur, while at the same time, a strong desire to be liked and accepted (Gilbert, 2014). Antisocial peer groups tend to be smaller and less connected (Kreager, Rulison, & Moody, 2011), which might contribute to greater availability for friendships and opportunities for less competitive social affiliation (Gilbert, 2014; Rudolph et al., 2014). These characteristics might allow socially anxious youth to affiliate with antisocial peers more readily and enhance perceptions of “being in control,” increase feelings of social attractiveness, and ultimately, avoid losing social rank. Indeed, studies show early adolescents experiencing internalizing symptoms are more likely to befriend peers who smoke, drink, and use marijuana (Siennick, Widdowson, Woessner, & Feinberg, 2016), and youth with social anxiety disorders have a heightened risk for conduct disorders and antisocial behaviors, which frequently coincide with antisocial-peer affiliation (Marmorstein, 2006; Sareen, Stein, Cox, & Hassard, 2004).

Robustness of Hypotheses

To provide a more robust examination of the proposed pathway, we were careful to consider the potential bi-directionality among variables, as well as the possibility that the strength or significance of associations might vary as a function of youth gender. The focus of this study was examining a developmental process of IPC to adolescent loneliness, friendship support, and affiliation with antisocial peers via elevated threat appraisals of IPC and social anxiety symptoms. However, some scholars posit that individual characteristics, such as social anxiety, influence threat appraisals of IPC, reflecting more global response patterns to interpersonal relationships (Cook et al., 2017). Additionally, other research suggests that lower quality peer relationships exacerbate adolescent vulnerability for social anxiety symptoms (Alden & Taylor, 2004; Mak et al., 2018; Tillfors et al., 2012). Thus, we were careful to also account for the effect of social anxiety symptoms on subsequent threat appraisals to IPC, as well as the effect of peer relationships on subsequent social anxiety symptoms.

Moreover, we considered whether the strength or significance of associations vary as a function of youth gender. Many studies have examined adolescent gender differences in youths’ responses to IPC, citing gender differences in relationship orientation and socialization that make adolescent girls more vulnerable to the deleterious effects of parental discord (Davies & Lindsey, 2001; Grych et al., 2004). Similarly, Grych, Raynor, and Fosco (2004) found that the association between exposure to IPC and threat appraisals was stronger for girls than for boys. Grych et al. (2003) found that perceived threat mediated the association between IPC and internalizing problems for girls but not for boys. Many studies, however, have failed to find gender differences in associations between IPC, threat appraisals, and various indicators of adolescent adjustment, including peer relationship outcomes (Buehler et al., 2009; Davies & Lindsay, 2001; Fosco & Feinberg, 2015; Grych et al., 2000; McDonald & Grych, 2006). Gender differences in associations between social anxiety and peer relationship outcomes are inconsistent (La Greca & Lopez, 1998; Flanagan, Erath, & Bierman, 2008). Given the inconsistent evidence, we did not formulate specific hypotheses regarding potential adolescent gender differences in the significance or strength of structural paths.

Summary of the Current Study

In an effort to expand the implications of IPC to a broader developmental framework, this study examines the effects of IPC and threat appraisals of IPC on adolescents’ experiences in the peer context. Guided by developmental saliency and evolutionary frameworks (Gilbert, 2014), we examined social anxiety symptoms as a key mechanism in this developmental process that emerges in response to elevated threat appraisals of IPC and undermines youths’ peer relationships. Specifically, we hypothesized that adolescents who are exposed to more frequent IPC experience increased threat appraisals of IPC, which, in turn, increase adolescent risk for social anxiety symptoms. We expected that heightened social anxiety symptoms undermine friendship support and increase adolescent loneliness and affiliation with antisocial peers. This model was tested using four waves of longitudinal data and an auto-regressive design to assess change in these variables over time. Moreover, this study employed a thorough test of the hypothesized associations by accounting for potential bi-directional associations among threat appraisal, social anxiety symptoms, and peer relationship outcomes and examining whether findings vary as a function of adolescent gender.

Method

Procedure

Participants were from PROSPER (PROmoting School-community-university Partnerships to Enhance Resilience): a partnership-based delivery system for evidence-based preventive interventions (Spoth, Greenberg, Bierman, & Redmond, 2004). A three-component community-university partnership model, including local community teams, state-level university researchers, and a Prevention Coordinator team in the land grant university Extension System, guided the implementation of evidence-based interventions. Prevention Coordinators were liaisons between community-based teams and university researchers, providing continuous, proactive technical assistance to the community teams. Community teams comprised of an Extension staff team leader, a public school representative co-leader, representatives of local human service agencies, and other local community stakeholders (e.g., youth and parents).

Participants in PROSPER were from 28 rural and small-town communities in Iowa and Pennsylvania. Communities that included: (a) school district enrollment from 1,300 to 5,200, and (b) at least 15% of the student population eligible for free or reduced-cost lunches, were eligible for participation (for more information, see Spoth, Guyll, Lillehoj, Redmond, & Greenberg, 2007). Participating students were in 6^th grade at Wave 1 (W1) of data collection. A total of 10,849 students across two cohorts (spaced one year apart) completed baseline assessments (approximately 90% of those eligible). On average, 88% of all eligible students completed follow-up surveys across eight data points (Spring 6^th to 12^th grades), with slightly higher rates of participation at earlier data collection points.

Communities were blocked on school district size (enrollment) and geographic location, and then they were randomly assigned to the partnership intervention and “normal programming” comparison conditions. Community teams selected and oversaw the implementation of a family and a school evidence-based intervention in their communities from a menu of evidence-based programs. Communities administered these programs for two successive cohorts of 6^th graders. For the family-focused intervention, all community teams selected the Strengthening Families Program: For Parents and Youth 10–14 (SFP 10–14), which occurred in 6^th grade. The school-based intervention was implemented for the same cohort of youth in 7^th grade. Four community teams selected Life Skills Training, four selected Project Alert, and six selected the All Stars curriculum (for more detail, see Spoth, et al., 2007).

A random sample of 2,267 families from the second cohort also were invited to complete in-home family assessments. In-home assessments were completed by 980 (43%) of families. The in-home assessments included a family composition interview and written questionnaires completed independently by the adolescent, mother, and, if present, father. Approval for PROSPER was obtained by the Iowa State University and Pennsylvania State University Institutional Review Boards. Comparisons of youth in the in-home sample with the original sample at baseline indicated that they were similar. Youth in the in-home sample were not different from the general population at Wave 1 on receipt of free or reduced lunch (33.6 vs. 33.0% respectively), living with two biological parents (59.3 vs. 62.5%), race (88.6 White vs. 86.5% White), or gender (49.5 vs. 46.8% male). Youth in the in-home sample were slightly less likely to engage in delinquent behavior than those in the original sample (F(1, 27) = 18.32, p < .01, d = .15), suggesting that the in-home subsample may be at slightly lower risk for problem behavior.

Participants

The focus of this study was adolescent exposure to interparental conflict; thus, we selected the 768 two-parent families that participated in the in-home family assessment for analyses. Two-parent families were defined as: a) married and living with their spouse, or b) living with someone in a steady, marital-like relationship. Data about the length of their relationship were not available. Retention rate was 80% (N = 611) through Wave 4. Over four waves of data, 91% of couples remained together. Data were collected in the Fall of 6^th grade and the Spring of 6^th through 8^th grades. Adolescents were 11.3 years old, on average, during 6^th grade. Males comprised 47% of youth. Adolescents were White (89%), Hispanic (6%), African American (1%), Asian (1%), or Other (3%). Female caregivers identified their relationship to the target adolescent as “mother” (94.9%), “stepmother” (1.3%), and other parental figures (3.8%; e.g., parents’ significant other, foster parent). Male caregivers identified their relationship to the target adolescent as “father” (75.3%), “stepfather” (16.9%), and other parental figures (7.8%).

Measures

Interparental conflict (Fall 6^th grade)

Mothers and fathers responded to seven items assessing the frequency of conflict behaviors over the past month on a 7-point scale (always = 1 to never = 7). Sample items included: “criticize your ideas” and “hit, push, grab, or shove you.” Items were reverse-coded so that higher values reflect more frequent conflict. This scale is correlated with marital dissatisfaction and marital distress (Cui & Conger, 2009). Mother and father scales were computed as item averages and scaled so that higher values reflected more interparental conflict. Mother and father report on interparental conflict were each a manifest indicator on the latent variable. Cronbach’s alphas for mother and father reports of their own and their partners’ conflict ranged from .83 to .90. Youth report on interparental conflict was the third manifest indicator of the latent variable. Youth reported on a single item, “Thinking about your parents or guardians, how often would you say they argue or disagree with each other?” Response options ranged from 1 (always) to 5 (never) and scores were reverse-coded so that higher values reflected more frequent interparental conflict.

Threat appraisals of IPC (Fall and Spring 6^th grade)

Adolescents completed four items adapted from the Children’s Perceptions of Interparental Conflict Scale (CPIC; Grych et al., 1993) that assess their beliefs that interparental conflict may have negative consequences for them, their parents, or their family. A sample item was, “When my parents argue, I’m afraid that something bad will happen.” Items were rated on a 5-point scale (strongly agree = 1 to strongly disagree = 5). Items were reverse scored so that higher values reflected greater perceived threat. Cronbach’s alphas for the total scale at W1 and W2 were .86 and .87, respectively.

Social anxiety symptoms (Fall 6^th grade and Spring 7^th grade)

Adolescent social anxiety symptoms were a latent variable with two manifest indicators. Adolescents completed the 18-item Social Anxiety Scale for Adolescents (SAS-A; La Greca & Lopez, 1998) and indicated how true each item was for themselves on a 5-point scale (not at all = 1 to all the time = 5). Although previous research has sometimes treated the SAS-A as having three subscales (fear of negative evaluation and two social avoidance subscales), exploratory and confirmatory factor analyses conducted with the current sample indicated that two factors, a fear of negative evaluation and a social avoidance subscale, were a better fit in this sample for both measurement occasions. The first manifest indicator was an 8-item fear of negative evaluation (FNE) subscale. A sample item was “I worry about being teased.” The second manifest indicator was a 10-item social avoidance (SA) subscale. A sample item was “I feel shy around people I don’t know.” The latent variable was scaled so that higher values reflected greater social anxiety symptoms. Cronbach’s alphas for the FNE and SA subscales ranged from .87 to .93.

Antisocial peers (Fall 6^th grade and Spring 8^th grade)

Adolescents responded to three items on a 5-point scale (strongly disagree = 1 to strongly agree = 5) that assessed their perception of their friends’ antisocial behaviors. Items were “These friends sometimes get into trouble with police,” “these friends sometimes break the law,” and “these friends don’t get along very well with their parents.” Items were adapted from Conger et al. (1991) and has demonstrated acceptable reliability (.85). A confirmatory factor analysis was conducted to validate the factor structure, with results indicating that the model fit the data well at each time point. Each item was used as a manifest indicator on the latent variable. Cronbach’s alphas for the scales during 6^th grade and 8^th grades were .79 and .82, respectively.

Loneliness (Fall 6^th grade and Spring 8^th grade)

Adolescents completed three items from the four item School Loneliness Questionnaire (Asher & Wheeler, 1985). Previous research has assessed and demonstrated the utility of the three item scale using factor analyses (McCloskey & Stuewig, 2001). Items were “I’m lonely at school,” “I feel left out of things at school,” and “I feel alone at school” (strongly disagree = 1 to strongly agree = 5). A confirmatory factor analysis was conducted to validate the factor structure, with results indicating that the model fit the data well at each time point. Each item was used as a manifest indicator on the latent variable. Cronbach’s alphas for the three-item scales were .93 and .94 during 6^th grade and 8^th grades, respectively.

Friendship support (Fall 6^th grade and Spring 8^th grade)

Adolescents completed five items from the 46-item Friendship Quality Questionnaire-Revised (FQQ) on a 5-point scale (not at all true = 1 to really true = 5) (Rose & Asher, 1999). The FQQ was designed to capture a broad range of best friendship qualities including companionship/recreation, validation/caring, help/guidance, intimate disclosure, conflict/betrayal, and conflict resolution and has demonstrated strong reliability and validity (Rose & Asher, 1999). In the present study, the FQQ was adapted to assess perceived support from friendships more generally rather than support from a single best friendship. Sample items included, “My friends care about me,” and “I can count on my friends when I need them.” A confirmatory factor analysis was conducted to validate the factor structure, with results indicating that the model fit the data well at each time point. Each item was used as a manifest indicator on the latent variable. Cronbach’s alphas were .78 and .79 during 6^th grade and 8^th grades, respectively.

Covariates (Fall 6^th grade)

Several covariates were included in the structural model based on previous research using subsamples from PROSPER finding that these variables are associated with missing data patterns (blinded for review). Parent education was scored from 0 to 20: 0 (no grade completed), 1–16 (each year until completion of a bachelor’s degree), 17 (education beyond a bachelor’s degree), 18 (master of science/master of arts), 19 (education beyond a master’s degree), to 20 (doctoral degree). Family income was scored from 1 to 11 in $10,000 intervals (1 = $0 −10,000 to 11 = above $100,000). Biological parent was coded to indicate whether both parents were biological parents (1) or not (0). We also included intervention condition to indicate whether families were in the intervention (1) or control (0) condition of the larger study.

Analysis Plan

Analyses were conducted usingMplus Version 8.1 (Muthen & Muthen, 2015). Three fit indices were used to assess the acceptability of each analytic model: the chi-square statistic, comparative fit index (CFI), and root-mean-square error of approximation (RMSEA). A nonsignificant chi-square statistic indicated good model fit. However, due to the large sample size, a significant chi-square was expected for most models. Therefore, other fit indices also were examined. Adequate model fit was indicated by CFI values of .90 to .95 (Hu & Bentler, 1999) and RMSEA values ranging from .06 to .08 (Browne & Cudeck, 1993). Good model fit was indicated by CFI values greater than .95 and RMSEA values less than .05 (Browne & Cudeck, 1993). The significance level for all estimates was p < .05. Missing values were addressed using maximum likelihood estimation methods, a preferred technique for producing estimates with minimal bias (Schlomer, Bauman, & Card, 2010).

Given that this study examined constructs at multiple time points and with adolescent boys and girls, as a preliminary step, we examined measurement invariance across time and adolescent gender. First, we examined the fit of models with freely estimated parameters (configural invariance). We then compared the configural model to the metric model in which factor loadings were constrained to be equal across time or gender. When comparing nested models, a ACFI of .01 or greater was used as the threshold for invariance, which is the preferred approach (compared to chi-square tests) with larger samples (i.e., N > 300; Cheung & Rensvold, 2002).

Indirect effects were assessed using a bias-corrected bootstrapping procedure (5,000 draws) which produces a 95% confidence interval. A 95% confidence interval that does not contain zero indicates a significant mediating effect (Preacher et al., 2007). We also examined whether the pattern of results was consistent across adolescent boys and girls using multi-group invariance tests. We compared a model where all structural paths were constrained to be equal across groups to a model in which paths were freely estimated across boys and girls. Group comparisons were evaluated by change in the CFI. A ACFI of .01 or greater indicated a significant change in model fit due to parameter constraints (Cheung & Rensvold, 2002).

Results

Preliminary Analyses

Missing data patterns were examined using the Little’s MCAR test, which indicated that data were not missing completely at random, (x² (2052) = 2459.78, p < .001). Study condition, living with two biological parents, parent education, and family income were significantly correlated with missingness. Thus, these variables were included as auxiliary variables in the model. A series of t-tests also were conducted to examine differences on all variables in the model and covariates between those remained versus dropped out of the study. Participants who remained in the study at W2 were more likely to be in the intervention condition (M = .59; t = 2.12, p < .05) and to have parents with more education (M = 13.39; t = 3.88, p < .05) than those who did not (M’s = .38 and 12.12, respectively). There were no differences on variables or covariates for participants who remained in the study at W3 or W4 versus those who did not.

Descriptive statistics and intercorrelations among variables are presented in Table 1. Correlations were in the expected directions, and correlations within construct variables were high. Construct covariances are shown in Table 2. Across mother, father, and adolescent reports of IPC, correlations ranged from .41 to .58 (p’s < .001). To confirm that construct measurements were equivalent across time and adolescent gender, we tested for configural and metric invariance. For the longitudinal model, the configural factor model fit the data adequately (χ² (570) = 1481.85, p < .001; CFI = .93; RMSEA = .05). We then compared a model where factor loadings were constrained to be equal across time to the configural model, with results supporting metric invariance (ΔCFI = .002). Results were similar for measurement invariance tests of youth gender. The configural model for boys and girls yielded an acceptable fit (χ² (1148) = 2012.72, p < .001; CFI = .94; RMSEA = .04), and model comparisons yielded support for metric invariance (ΔCFI = 0).

Table 1.

Descriptive Statistics and Intercorrelations between Variables

Variables	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23
Interparental conflict
1. MR1	-
2. FR1	.58	-
3. YR1	.42	.41	-
Perceived threat
4. item 11	.24	.23	.30	-
5. item 21	.23	.27	.28	.65	-
6. item 31	.22	.26	.33	.63	.58	-
7. item 41	.23	.25	.27	.65	.59	.58	-
8. item 12	.19	.24	.24	.44	.42	.37	.42	-
9. item 22	.19	.21	.26	.39	.50	.36	.38	.67	-
10. item 32	.21	.25	.26	.37	.33	.43	.36	.63	.61	-
11. item 42	.21	.24	.18	.40	.38	.31	.49	.68	.61	.54
Social anxiety
12. FNE1	.04	.11	.21	.32	.27	.34	.29	.28	.25	.27	.27	7
13. SA1	.07	.10	.16	.27	.33	.28	.29	.31	.28	.23	.27	7 .72	-
14. FNE3	.09	.17	.26	.30	.26	.29	.29	.35	.31	.34	.29	.52	.44	-
15. SA3	.11	.17	.21	.22	.28	.24	.24	.35	.33	.26	.25	.46	.55	.72	-
Antisocial peers
16. item 11	.10	.00	.06	.15	.19	.08	.05	.03	.09	.10	.11	.03	.02	.02	.07	-
17. item 21	.07	.01	.05	.08	.14	.04	.04	.03	.09	.09	.09	.04	.04	.01	.06	.75	-
18. item 31	.11	.03	.18	.13	.11	.15	.13	.03	.07	.12	.10	.14	.06	.11	.07	.45	.47
19. item 14	.00	−.02	.05	.05	.01	.04	.04	.03	.06	.14	.13	00	−.03	−.01	.04	.29	.27	.24	-
20. item 24	.05	−.01	.03	−.02	−.01	−.01	−.01	.02	.01	.12	.10	−.03	−.05	.00	−.02	.28	.30	.21	.82
21. item 34	.07	.05	.13	.12	.10	.11	.12	.12	.11	.19	.15	.10	.03	.12	.06	.15	.19	.27	.52	.53	-
Loneliness
22. item 11	.05	.10	.15	.22	.22	.26	.22	.14	.16	.16	.14	.39	.29	.31	.27	.05	.04	.07	.04	.01	.11	-
23. item 21	.04	.07	.13	.19	.22	.24	.22	.15	.20	.17	.17	.40	.33	.30	.31	.04	.02	.11	.08	.04	.13	.79	-
24. item 31	.03	.08	.12	.19	.21	.24	.20	.11	.15	.17	.13	.38	.29	.27	.26	.05	.01	.10	.04	.01	.12	.84	.80
25. item 14	−.01	.02	.09	.11	.16	.18	.19	.13	.13	.21	.06	.21	.18	.28	.25	.06	.07	.16	.16	.12	.24	.17	.23
26. item 24	.03	.06	.15	.12	.16	.16	.15	.16	.15	.23	.09	.26	.24	.33	.29	.04	.06	.11	.12	.12	.21	.23	.29
27. item 34	−.02	.05	.11	.13	.22	.22	.17	.15	.14	.19	.06	.22	.19	.30	.27	.02	.01	.09	.10	.08	.20	.19	.24
Friendship suport
28. item 11	−.07	−.03	−.09	−.13	−.12	−.07	−.08	−.12	−.12	−.08	−.15	−.18	−.15	−.13	−.12	−.08	−.06	−.10	−.17	−.14	−.11	−.20	−.21
29. item 21	−.10	−.06	−.15	−.16	−.12	−.21	−.13	−.16	−.15	−.15	−.15	−.25	−.18	−.17	−.12	−.03	−.04	−.11	−.03	−.06	−.14	−.27	−.24
30. item 31	−.06	−.03	−.08	−.05	−.09	−.04	−.03	−.09	−.07	−.04	−.04	−.11	−.08	−.06	−.09	−.07	−.09	−.05	−.12	−.11	−.02	−.22	−.16
31. item 41	−.11	−.06	−.15	−.19	−.14	−.14	−.09	−.15	−.14	−.14	−.12	−.23	−.18	−.15	−.12	−.08	−.08	−.14	−.09	−.08	−.10	−.24	−.21
32. item 51	−.03	−.01	−.13	−.12	−.07	−.14	−.11	−.10	−.10	−.14	−.06	−.25	−.21	−.15	−.16	−.09	−.09	−.16	−.08	−.0.6	−.07	−.24	−.21
33. item 14	−.01	−.02	−.11	−.06	−.14	−.08	−.08	−.04	−.08	−.06	−.04	−.08	−.08	−.18	−.12	−.12	−.10	−.13	−.08	−.14	−.10	−.14	−.12
34. item 24	−.02	−.05	−.13	−.07	−.10	−.09	−.09	−.02	−.05	−.09	−.06	−.12	−.09	−.17	−.13	−.07	−.05	−.14	−.16	−.18	−.17	−.14	−.14
35. item 34	.00	−.05	−.09	−.03	−.11	−.07	−.03	−.09	−.10	−.09	−.04	−.05	−.08	−.13	−.19	−.12	−.10	−.09	−.21	−.20	−.11	−.09	−.14
36. item 44	−.08	−.07	−.12	−.05	−.15	−.14	−.11	−.12	−.10	−.12	−.08	−.10	−.12	−.17	−.16	−.04	−.05	−.11	−.09	−.17	−.12	−.11	−.12
37. item 54	−.05	−.03	−.11	−.05	−.09	−.09	−.11	−.08	−.07	−.13	−.06	−.20	−.14	−.23	−.14	−.05	−.05	−.14	−.18	−.22	−.28	−.15	−.16
Covariates
32. educ	.06	−.05	−.01	−.15	−.25	−.12	−.15	−.11	−.18	−.10	−.13	−.12	−.16	−.06	−.08	−.12	−.11	−.06	−.14	−.06	−.08	−.07	−.10
33. inc	−.10	−.10	−.03	.20	.28	.09	.14	−.11	−.18	−.06	−.11	−.13	−.18	−.13	−.15	−.10	−.13	−.06	−.14	−.13	−.16	−.11	−.10
34. bio	−.03	−.04	−.09	−.13	−.15	−.07	−.09	−.09	−.15	−.12	−.07	−.13	−.13	−.06	−.10	−.10	−.10	−.07	−.11	−.10	− .15	−.06	−.05
35. cond	.02	.00	.03	.03	.05	.05	.03	−.01	.01	−.02	.05	.00	.05	.00	−.04	.00	.02	.03	−.11	−.06	−.04	.01	−.02
36. sex	−.04	−.04	−.04	.01	.02	.01	.00	.03	.02	.00	−.01	−11	−.08	.01	−01	.12	.11	.05	.05	.07	−.09	.00	.03
Variables	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	M	SD	N
Interparental conflict
1. MR1																			2.03	.77	760
2. FR1																			2.08	.73	641
3. YR1																			2.44	.85	708
Perceived threat
4. item 11																			2.57	1.29	737
5. item 21																			2.06	1.26	736
6. item 31																			2.14	1.21	736
7. item 41																			2.20	1.37	736
8. item 12																			2.46	1.30	636
9. item 22																			1.95	1.24	636
10. item 32																			2.17	1.22	636
11. item 42																			2.09	1.33	632
Social anxiety
12. FNE1																			2.19	.89	739
13. SA1																			2.21	.76	739
14. FNE3																			2.14	.90	628
15. SA3																			2.10	.74	628
Antisocial peers
16. item 11																			1.31	.75	701
17. item 21																			1.34	.78	701
18. item 31																			1.51	.87	699
19. item 14																			1.50	.90	618
20. item 24																			1.61	.99	614
21. item 34																			1.78	.99	612
Loneliness
22. item 11																			1.56	.91	734
23. item 21																			1.64	.96	734
24. item 31	-																		1.53	.92	733
25. item 14	.18	-																	1.56	.85	610
26. item 24	.25 .82	-																	1.63	.88	610
27. item 34	.23	.87	.83	-															1.53	.85	610
Friendship support
28. item 11	−.22	−.06	−.07	−.07	-														4.38	.85	739
29. item 21	−.25	−.08	−.08	−.12	.43	-													4.14	1.10	737
30. item 31	−.22	−.08	−.08	−.09	.43	.36	-												3.53	1.30	739
31. item 41	−.20	−.06	−.06	−.05	.47	.48	.45	-											4.21	1.09	733
32. item 51	−.24	−.10	−.11	−.12	.38	.46	.32	.48	-										4.04	1.07	740
33. item 14	−.14	−.24	−.26	−.22	.23	.13	.19	.17	.13	-									4.49	.74	610
34. item 24	−.12	−.25	−.23	−.22	.26	.18	.19	.20	.20	.45	-								4.24	1.03	610
35. item 34	−.11	−.23	−.23	−.24	.22	.14	.35	.17	.14	.44	.35	-							3.85	1.20	610
36. item 44	−.14	−.22	−.26	−.26	.24	.19	.19	.18	.19	.47	.45	.49	-						4.36	.90	610
37. item 54	−.17	−.27	−.30	−.24	.21	.19	.15	.21	.25	.46	.51	.36	.52	-					4	1.01	610
Covariates
38. educ	−.07	−.02	−.04	−.01	.09	.02	.02	.14	.03	.02	−.03	.03	.03 .03	-					13.25	2.19	755
39. inc	−.09	−.07	−.11	−.08	.14	.07	.03	.13	.09	.02	.06	.05	.03	.05	.50	-			6.10	2.63	702
40. bio	−.07	−.02	−.05	−.05	.08	.13	.01	.08	−.01	.00	−.01	.00	−.06	−.04	.12	.25	-		.67	.47	745
41. cond	−.03	−.03	−.04	.01	.02	−.04	.04	.02	.01	−.03	−.06	.02	−.08	−.08	.04	.01	.05	-	.59	.49	767
42. sex	.01	.03	−.03	−.01	−.17	−.08	−.24	−.10	−.11	−.23	−.13	−.33	−.15	−.08	.03	.07	.05	−.03	.50	.62	768

Note. Bolded p <.05 MR = mom report; DR = dad report; YR = youth report; FNE = fear of negative evaluation; SA = social avoidance; bio = biological parent; condition = study condition. Superscripts indicate wave of data collection.

Table 2.

Covariance Estimates between Latent Variables

Latent construct	1	2	3	4	5	6	7	8	9	10	11	12	13
1. W1 Interparental conflict	-
2. W1 Perceived threat	.43	-
3. W1 Social anxiety	.15	.44	-
4. W1 Antisocial peers	.07	.15	.06	-
5. W1 Loneliness	.11	.30	.45	.05	-
6. W1 Friendship support	−.15	−.23	−.33	−.14	−.37	-
7. W1 Education	−.06	−.21	−.15	−.13	−.08	.11	-
8. W1 Income	−.11	−.22	−.18	−.15	−.12	.15	.51	-
9. W1 Bio parent	−.06	−.14	−.15	−.11	−.06	.09	.12	.24
10. W1 Condition	−.01	.06	.03	.02	−.01	.02	.04	.01	.05	-
11. W4 Antisocial peers	-	-	-	-	-	-	-	-	-	-	-
12. W4 Loneliness	-	-	-	-	-	-	-	-	-	-	.13	-
13. W4 Friendship support	-	-	-	-	-	-	-	-	-	-	−.24	−.34	-

Note. Bolded p < .05

Associations among IPC and Adolescent Peer Relationships

Results from the structural model indicated that model fit was good (χ²(584) = 1248.91, p = .00, CFI = .95, RMSEA = .04; see Figure 1). Adolescent threat appraisals and adolescent social anxiety symptoms demonstrated moderate levels of stability over time. The standardized regression estimate for adolescent threat appraisals from Fall to Spring of 6^th grade was .53 (p < .001). The standardized regression estimate for social anxiety symptoms from 6^th to 7^th grades was .56 (p < .001). Peer relationship qualities demonstrated lower levels of stability from 6^th to 8^th grades, with standardized estimates ranging from .15 to .38 (p’s < .001).

Figure 1.

Mediational Model of Threat and Social Anxiety Pathways from Interparental Conflict to Peer Relationships

Note. Model fit: χ²(584) = 1248.91, p < .001; CFI = .95; RMSEA = .04. Covariances were estimated and disturbance terms between dependent variables correlated but are not shown for clarity of presentation. Path coefficients reflect standardized betas. Solid lines are statistically significant paths (p < .05). Dotted lines are not statistically significant. Paths from interparental conflict (IPC) and peer relationships at W1 to peer relationship outcomes at W4 were estimated but were not significant and are not shown for clarity of presentation. Standardized beta coefficients for IPC to antisocial peers, loneliness, and friendship support were −.01, −.04, and −.09, respectively. Endogenous variables were regressed on all covariates but are not shown for clarity of presentation.

Although bivariate correlations indicated that indicators of IPC were significantly correlated with some of the peer relationship outcomes, in the full structural model, IPC and adolescent threat appraisals were not directly associated with any peer-related outcomes. Adolescents in families with more frequent IPC during the Fall of 6^th grade reported increases in threat appraisals (b = .21, β = .11, p < .05, 95% C.I. [.01, .21]). In turn, adolescents who experienced higher levels of threat appraisals during the Spring of 6^th grade reported increases in social anxiety symptoms (b = .15, β = .22, p < .001, 95% C.I. [.12, .32]) (6^th to 7^th grades). Adolescents who experienced heightened social anxiety symptoms during 7^th grade reported decreased friendship support (b = −.13, β = −.21, p < .001, 95% C.I. [−.32, −.09]) and increased loneliness (b = .31, β = .30, p < .001, 95% C.I. [.18, .40]) (6^th to 8^th grades). Social anxiety symptoms were not significantly associated with changes in adolescent affiliation with antisocial peers (b = −.06, β = −.06, p = .22, 95% C.I. [−.18, .06]). A Wald test for parameter constraints revealed that the path from social anxiety symptoms to loneliness was significantly stronger than the path to friendship support (χ²(1) = 33.94, p < .001). We examined the indirect effects from interparental conflict to peer outcomes through increases in adolescent threat appraisals and social anxiety symptoms. Consistent with our hypothesis, there was a significant indirect effect of interparental conflict on decreased friendship support (b = −.01, β = −.01, 95% C.I. [−.02, −.001]) and increased loneliness (b = .01, β = .01, 95% C.I. [.001, .02]) through threat appraisals and social anxiety.

In the same model, we also examined potential bidirectional associations among threat appraisals, social anxiety symptoms, and peer relationships outcomes. Heightened social anxiety symptoms during the Fall of 6^th grade were significantly associated with increased appraisals of IPC as threatening (b = .15, β = .11, p < .05, 95% C.I. [.003, .21]). Friendship support, loneliness, and affiliation with antisocial peers during the Fall of 6^th grade were not significantly associated with increased social anxiety symptoms in the Spring of 6^th grade. We also tested whether there were adolescent gender differences in the structural model. A multi-group invariance test indicated that there were no differences in the structural paths for boys and girls (ACFI < .01), suggesting that the model fit equally well for boys and girls. Additional tests revealed that these findings were the same with and without the covariates in the model.

Discussion

Extant literature examining the implications of IPC for adolescents’ peer relationships is limited and has rarely been guided by important advancements in theory and research showing that adolescents’ subjective evaluations of IPC are crucial to our understanding of the implications of IPC for youth outcomes (Harold & Sellers, 2018). Consequently, developmental understanding of the implications of IPC and youths’ subject evaluations of IPC has been narrowed to a few, broad indicators of youth psychopathology, and there is an important gap to be filled by research explaining how experiences of IPC might be associated with more specific and developmentally salient youth outcomes. The current study narrowed this gap by drawing on the cognitive contextual framework (Grych & Fincham, 1990) and evolutionary perspectives (Gilbert, 2014) to examine a developmental process linking exposure to IPC to adolescents’ experiences of friendship support, loneliness, and antisocial peer affiliation via heightened threat appraisals of IPC and social anxiety symptoms. Specifically, we hypothesized that adolescents who are exposed to more frequent, hostile, and poorly resolved IPC are more likely to appraise these conflicts as threatening (Grych & Fincham, 1993). In turn, we expected that greater perceived threat of IPC in the family context increases adolescent risk for social anxiety symptoms which contribute to decreases in friendship support and increases in loneliness and antisocial peer affiliation. Our test of hypotheses was strengthened by an auto-regressive design using four waves of longitudinal data, consideration of bidirectional effects, and an examination of potential gender differences.

Overall, the results were consistent with our hypotheses. Exposure to hostile IPC was related to increases in adolescents’ threat appraisals of IPC—a finding that replicates a growing body of research (e.g. Fosco & Grych, 2008; Gerard et al.,; Grych et al., 2003; Schlomer et al., 2015) and supports a main assertion in the cognitive contextual framework that youth who are frequently exposed to IPC are increasingly likely to appraise that this conflict poses a risk to themselves or their family (Davies & Woitach, 2008; Fosco et al., 2007). In previous research, the reasons for how or why threat appraisals of IPC might affect youths’ relationships in the peer context have been less clear, and our findings show that threat appraisals of IPC are not directly associated with peer relationship outcomes. Rather, our results point to social anxiety symptoms as an explanatory mechanism in the process linking IPC and threat appraisals of IPC to peer outcomes. Considered within an evolutionary framework, schemas of interpersonal relationships that are formed in the context of IPC might underlie the linkages between threat appraisals of IPC and social anxiety symptoms (Gilbert, 2015). Although the nature of interpersonal fears and concerns might be context-specific, the findings suggest important inter-context continuations of fearful and wary cognitive appraisals of interpersonal relationships. Specifically, in the face of IPC, youth who appraise this conflict as threatening might become “primed” to expect that interpersonal relationships are competitive and coercive—an expectation that is at the core of social anxiety which contributes to (a) hypervigilance, (b) preoccupations with how one behaves and is evaluated in social situations, and (c) avoidance of social interactions as a method of preventing and evading potential rejection (Gilbert, 2014). Other studies have found connections between threat appraisals of IPC and internalizing problems or general anxiety and depression (e.g., Fosco & Bray, 2016; Fosco & Grych, 2008), but this is the first study to our knowledge to examine linkages with social anxiety—a context-specific and developmentally salient outcome that reflects the growing salience that adolescents attribute to peer relationships and the necessity of their skills and ability to engage in these relationships effectively.

Results indicated that increased social anxiety symptoms, subsequently, were associated with undermined friendship support and increased feelings of loneliness, but were not related to antisocial peer affiliation. These findings point to important specificity in the implications of social anxiety symptoms for peer relationship outcomes and as a linking mechanism in our developmental model. Unsurprisingly, and consistent with previous research, these results suggest that the social avoidance characterizing social anxiety might impede opportunities to connect with others, resulting in increased loneliness (Mak et al., 2018; Prinstein & La Greca, 2002; Stoeckli, 2009). Fearful and avoidant characteristics of socially anxious youth might also be perceived as unusual and unattractive to peers during a developmental period in which peer relationships are considered highly salient; thus, peers become less interested in establishing and maintaining close, connected relationships with socially anxious youth (Gilbert, 2014). Follow up tests indicated that coefficients for associations between social anxiety and loneliness were significantly stronger than those for friendship support. The significant difference in the strengths of these associations suggests that social anxiety has stronger consequences for ones’ own feelings and satisfaction in peer relationships than for how one is treated (or is perceived to be treated) by friends. It is important to note, however, that this study used youth report on friendships, and it should be investigated whether these findings hold when peer reports are used.

Although we did not find significant associations for antisocial peer affiliation, the link between social anxiety and antisocial peer affiliation warrants attention in future research. The null findings might be explained by important heterogeneity among individuals with social anxiety (Kashdan & Hofmann 2008). It might be that only a subgroup of adolescents with social anxiety affiliate with antisocial peers, as research has identified a subgroup of individuals with social anxiety symptoms who also engage in risk-prone and disinhibited behaviors (Kashdan & Hofmann, 2008). In contrast, other subgroups of youth with social anxiety might be too fearful of the consequences and pressures to engage with antisocial peers (Hodgins, Barbareschia, & Larsson, 2011). Additionally, findings are not conclusive regarding the indirect effects of IPC on antisocial peer affiliations via other linking mechanisms. Future research should explore other mechanisms, such as aggressive behavior, which have received considerable support as risks for antisocial peer affiliation (Dodge, Coie, & Lynam, 2006). Moreover, given that existing research has most consistently linked threat appraisals of IPC to internalizing problems, whereas several studies have linked self-blame (another cognitive appraisal) to externalizing problems (Fosco & Grych, 2008; Grych et al., 2003), future research should also consider self-blame as a potential explanatory mechanism of the effects of IPC on antisocial peer affiliations.

Tests of indirect effects were significant, aligning our findings with previous research and theory on IPC (Bryant & Conger, 2002; Grych & Fincham, 1990). The deleterious effects of exposure to IPC extend over a three-year period to undermine youths’ friendship support and feelings of loneliness via increases in threat appraisals and social anxiety symptoms. Thus, these results support that conceptualizations of the enduring implications of IPC for youth can be expanded beyond broad measures of youth psychopathology to also include youths’ experiences and relationships in the peer context. Moreover, findings that threat appraisals of IPC and social anxiety symptoms are key mechanisms implicate these problems as important areas for future intervention. Randomized interventions indicate that training individuals to disengage from negative social cues and threat-related cognitions successfully reduces social anxiety (Schmidt, Richey, Buckner, & Timpano, 2009). Altering youth’s appraisals that hostile IPC is threatening may not be advisable, given the role of appraisals in warning and protecting youth from violence that may result from escalating conflict. As an alternative, intervening on IPC, helping youth differentiate threat in families from threat in other contexts, and providing youth with strategies to alter their negative interpretations of interactions with peers (Aune & Stiles, 2009; Klein et al., 2015) might help youth cope in the face of worrisome interactions, leading to better individual functioning and success in other relationship contexts.

Finally, we also tested the robustness of associations by accounting for potential bidirectional associations among social anxiety, threat appraisals, and peer relationships. Results indicated that youth with heightened social anxiety experienced increases in threat appraisals over time. These results provide further support for the idea that fearful and wary cognitive appraisals carry across contexts and suggest child-driven effects in which youths’ negative experiences in the peer context change how they interact and perceive family relationships at home. Similarly, other scholars suggest that social anxiety symptoms might simply reflect more global response patterns to interpersonal relationships; thus, threat perceptions in one context would be continuous across contexts and mutually influential (Cook et al., 2017). A growing number of studies show that adolescents’ behaviors, and particularly externalizing problems, have implications for parents’ behaviors and family relationships (e.g., De Haan, Soenens, Dekovic & Prinzie, 2013). It will be interesting for future research in this area to consider the consequences of social anxiety symptoms for family relationships (e.g., blinded for review) and adolescents’ evaluations of family interactions. It also is important to note that findings did not vary as a function of adolescent gender. It appears that a similar developmental process characterizes how adolescent boys and girls carry-over conflictual relationship dynamics among their parents to their expectations in peer relationships. Despite greater prevalence of social anxiety among girls than among boys (Merikangas et al., 2010), our results suggest that social anxiety symptoms may be fruitful intervention targets for both adolescent boys and girls.

Limitations

Although this study includes many strengths, important limitations should be discussed. First, this study used a community sample of mostly rural or small-town, White families. This sample limits the generalizability of the findings to other racial and ethnic groups. This work needs to be replicated among samples of more diverse socio-economic backgrounds and other ethnicities. Second, only adolescent report was used for reports of threat appraisals, social anxiety, and the three aspects of peer relationships. This increases the potential for inflated associations due to shared method variance (Shadish, Cook, & Campbell, 2001); however, it is likely that adolescents are the best reporters of their perceptions of threat in response to IPC, and studies have indicated that adolescents are the best reporters of their own internalizing symptoms (i.e., social anxiety symptoms) (Grills & Ollendick, 2003; Lagattuta, Sayfan, & Bamford, 2012). Third, data on adolescents’ self-blame in response to IPC was not available. Self-blame has been linked to adolescents’ experiences in other relational contexts (Stocker & Youngblade, 1999); thus, future research should aim to include different types of adolescent appraisals of interparental conflict in order to more fully capture the processes linking IPC to adolescents’ peer relationships. Finally, our autoregressive controls for threat, social anxiety, and peer relationships outcomes were all examined during the Fall of 6^th grade, creating differences in the amount of change that was available to predict across variables. We note that the one year stability estimate for social anxiety is .65. Similarly, one year stability estimates for peer relationship outcomes increase to .56. In terms of auto-regressive tests, scholars have suggested that, “the designer must choose a lag that will provide sufficient time for the effect to occur” (p. 272; Selig & Little, 2012). Thus, employing a time lag of one year might lead to an underestimation of the effect of threat on social anxiety, or social anxiety on peer relationships, simply because sufficient time was not allowed for the effect to occur. Nonetheless, future research should carefully consider the timing of effects and the potential benefits of cross-lagged designs in future research.

Conclusion

In testing this developmental process of IPC to peer relationships outcomes, we help broaden our awareness of the developmental implications of IPC. Our findings identified important and novel connections between threat appraisals of IPC and social anxiety symptoms and identified these links as key mechanisms that explain the effects of IPC on adolescents’ peer relationships. Specifically, findings demonstrated that IPC threat appraisals have important implications for adolescents’ fear and avoidance in the peer context, which later undermines their ability to engage effectively in peer relationships. Findings from this study provide evidence that equipping youth with strategies to identify and cope with fears and avoidant behaviors in the social context (Aune & Stiles, 2009) may interrupt the disruptive consequences of IPC and help protect and promote youths’ positive peer relationships.