Parent–Adolescent and Interparental Conflicts

Adolescence is a period of dynamic growth in the development of abstract thinking and of greater self-awareness of one’s sexuality, identity, and autonomy. Adolescence also is marked by biological, cognitive, and sociological changes that affect the quality of relationships with parents (Comstock, 1994). Research increasingly supports the association between mental health and the quality of parent–adolescent relationships (E. Kim & Cain, 2008; S. L. Wong, 2000). For example, although the intensity of conflict between parents and their children increases during an adolescent’s transition into adulthood, with possible gender differences (Allison & Schultz, 2004), parent–adolescent conflict has been linked to adolescent anxiety and depressive symptoms (Adams & Laursen, 2007), low self-esteem (Hu & Ai, 2016), suicidal ideation (Au et al., 2009), and delinquency (Doorn et al., 2008).

Although parent–adolescent conflict is a common phenomenon across cultures, KA adolescents of immigrant parents may experience additional stress related to intergenerational issues (Park, 2015). Such issues may be related to incompatible goals as immigrant parents and children acculturate at different rates and in different ways (Hwang, 2006). According to the acculturative family distancing model (Hwang, 2006), a theoretically derived construct, immigrant parents and their children may be prone to greater family conflicts when family members distance from one another emotionally, cognitively, and behaviorally as a “consequence of differences in acculturative processes and cultural changes” (Hwang, 2006, p. 389). The breakdown in communication and incongruent cultural values is hypothesized to increase over time, causing greater emotional, cognitive, and behavioral distancing between parents and their children, thereby increasing the risk for family conflict, which has been linked to youth depression (Hwang et al., 2010) and depressive symptoms in KA adolescents specifically (Huh-Kim, 1998).

In addition, exposure to interparental conflict and violence can have a detrimental impact on children’s emotional and behavioral adjustment (G. H. Chung et al., 2009) and can contribute to parent–adolescent conflict that is greater in frequency and/or intensity. Adolescents exposed to parental conflict exhibit significantly higher levels of internalizing and externalizing behaviors than those from low-conflict families (Gerard et al., 2006). Yet, scant research examines this relationship among KA families.

Coping and Adolescents

Coping is an important variable between perceived stress and emotional adjustment (Orth et al., 2009). Understanding the relationship between coping responses and psychological well-being has implications for the development of appropriate interventions (R. M. Lee et al., 2005), as different coping strategies are associated with various psychological and physical health outcomes (E. C. Chang, 1996). Findings are inconsistent concerning the effects of coping on adolescents’ psychological well-being (R. M. Lee et al., 2005), and little is known about the effects of coping strategies on the mental health of KA adolescents.

For Asian American families with collectivist worldviews and a conception of self that is interdependent with others, interpersonal conflict reflects a disharmony between people who are inevitably interconnected (Chun et al., 2006). Therefore, the aim of this resolution is to manage conflict without damaging relationships by “riding it out” and reinforcing interdependence with a long-term view of the relationship in mind. In facing difficult interpersonal problems, traditional Asian values place importance on maintaining peace and emphasize personal sacrifice and perseverance over potentially destabilizing one’s relationships with others (P. T. P. Wong & Ujimoto, 1998).

A study of Chinese American college students challenged the assumption that problem-focused engagement coping strategies (e.g., problem solving, cognitive restructuring, express emotions and support seeking) are more adaptive than problem-focused disengagement coping strategies (e.g., problem avoidance, wishful thinking, self-criticism, and social withdrawal) when dealing with life’s stressors (E. C. Chang, 2001). Chang found that problem-focused engagement coping strategies were not necessarily more adaptive than problem-focused disengagement coping strategies when dealing with external stressors; that is, higher use of problem-avoidance coping in Asian American college students did not result in lower life satisfaction or greater depressive symptoms as it did for white students. Based on these findings, E. C. Chang (2001) argued that there are discernable differences between Asian American and white students’ use of coping strategies and these strategies may differentially influence psychological outcomes. For Asian American youth, distancing/avoidant coping strategies may have an indirect effect on depressive symptoms through appraisal processes that are culturally affirming to the self. However, further research is needed to illuminate the effects of disengagement coping on psychological outcomes among Asian Americans and, more specifically, among KA youth.

The Mediating Effect of Self-Esteem

Many diathesis-stress models use dispositional risk factors (e.g., self-esteem, self-efficacy, and cognitive attribution style) and environmental stress (e.g., family environment) to explain the development of depression (Hu & Ai, 2016; Orth et al., 2009). Self-esteem and self-efficacy are two aspects of self-concept. Self-esteem first conveys one’s sense of self-worth and value by others, and self-efficacy describes a sense of control, positive coping responses, and belief that challenges can be overcome (Benight & Bandura, 2004; Rosenberg, 1965). High self-esteem and self-efficacy can attenuate the effects of stress on psychological outcomes, such as depression (C. A. Lee et al., 2014). According to Orth and colleagues (2009), individuals with low self-esteem may be more prone to depression because they have fewer coping resources at their disposal when faced with challenging life circumstances. Individuals with high self-esteem, on the other hand, are buffered from depression because they are assumed to have better coping resources. Although various diathesis-stress models of depression have been proposed (Orth et al., 2009), specific pathways leading to depression have not been adequately investigated, especially among minority and immigrant youth. One study recently investigated a specific pathway to depression in a sample of Chinese youth and found that self-esteem partially mediated the relationship between stressful family relationships and depression (Hu & Ai, 2016). Another study of Chinese primary and secondary school students found that self-esteem fully mediated the effects of social support on depressive symptoms, but self-efficacy did not (C. W. Chang et al., 2018). The current study extends previous work by examining the direct and indirect effects of self-esteem and problem-focused disengagement coping on depression among KA adolescents.

Gender Differences in Family Conflict and Depression

The literature on gender differences in depression as a function of family dynamics among Asian Americans is scant; therefore, we examined the broader empirical literature with non-Asian American populations.

As noted by Davies and Lindsay (2004), research in this area is generally mixed. In one study, Essex and colleagues (2003) found that girls were more likely than boys to show an increase in depressive symptoms in the presence of family conflict. Similar results were reported by Davies and Lindsay (2004). They examined the association between interparental conflict and internalizing symptoms (withdrawal, anxiety, and depression) in a sample of 924 children ages 10–15. They found a significant interaction between interparental conflict and binary gender, indicating a stronger association between these variables for girls. They also tested and found support for the hypothesis that girls’ greater vulnerability to conflict is due to their communal disposition (e.g., interpersonal connectedness and concern for welfare of others). The authors concluded that the socialization of girls to traditional gender roles may account for their higher levels of distress stemming from discord or conflict in family relationships (Davies & Lindsay, 2004; Lewis et al., 2015).

Gender studies in immigrant youth fairly consistently show that girls tend to have greater domestic responsibilities than boys (Espiritu, 2001). For example, girls were asked more frequently than boys to translate; to cook; to attend to medical, financial, and legal matters; and to act as surrogate parents. There is also evidence that immigrant parents tend to be more restrictive of their daughters’ activities outside the home, such as spending time with friends and participating in after-school programs (Espiritu, 2001; Olson, 1997). Studies show that these gendered patterns of socialization of children can strain the parent–child dyad (Williams et al., 2002). For example, in a study of Asian American college students, female students reported significantly greater conflict with their parents over dating and marriage than male students (R. H. Chung, 2001). The author indicated that the sources of family conflict may differ for Asian American boys and girls, in part, due to cultural considerations. However, in a separate study of Chinese immigrant youth, no gender differences were observed in relation to family conflict (Florsheim, 1997).

Research Questions

The present study examined the prevalence of depressive symptoms among KA adolescents and explored the complicated relationships among family conflict (i.e., interparental conflict and parent–adolescent conflict), problem-focused disengagement coping, depressive symptoms, self-esteem, and gender in KA adolescents. The structural model is presented in Figure 1. We sought to answer the following research questions:

Study Design, Sample, and Data Collection Procedure

This study used a cross-sectional survey design and a convenience sampling strategy. Eligibility criteria included (a) have at least one parent who immigrated from Korea, (b) reside with at least one parent, (c) maintain contact with both parents, (d) 12–18 years of age, (e) able to read and understand English, and (f) reside in New York or New Jersey at the time of investigation. The researcher visited several KA religious organizations and one social service agency run by a pastor to explain the study purpose. Upon gaining permission to recruit potential participants, the researcher explained the study to groups of adolescents. Those who were interested in participating were given printed information explaining the study’s purpose, those under age 18 were given a consent form to be signed by their legal guardian and an assent form to sign themselves, and those who were 18 were given a consent form to sign. Data were collected from June to September 2014 after obtaining approval from the New York University Committee on Activities Involving Human Subjects and from the cooperating sites.

Potential participants who returned all signed forms to the researcher were given a questionnaire, which was self-administered in a designated room at each facility and took approximately 20–25 minutes to complete. Prior to beginning the survey, participants were informed that they were free to withdraw without reprisal and could discontinue participation at any time. As an incentive, they were entered into a raffle for $10 gift cards, regardless of survey completion. Of the 353 survey respondents, one withdrew due to minor discomfort during the survey, and 13 did not complete the survey, for a total of 339 KA adolescent participants (276 from religious organizations and 63 from a social service agency). The response rate was 96%.

Power Analyses

Power analyses were conducted using the software G*Power 3.1 (Faul et al., 2009) to determine the number of participants needed to detect effects due to the independent variables. Given that the statistical power for this study was set to 0.80, a medium effect size ( $f^2=0.15$) with 10 predictors for each analysis, and a criterion for significance of 0.05, a sample of 119 was needed to detect an effect.

Measures

Data Analytic Plan

Descriptive statistics for sociodemographic characteristics (i.e., age, gender, generational status, living arrangement, and parents’ education) were examined. We then applied linear regression to compare group mean differences of the major variables among KA adolescents, controlling for covariates (i.e., age, generational status).

We used structural equation modeling to determine whether self-esteem mediated the major independent variables (i.e., interparental conflict, parent–adolescent conflict, problem-focused disengagement coping) and depressive symptoms in KA adolescents. Analyses of total and indirect effects and the magnitude of the effects of the mediator were performed using a percentile based bootstrap method (Biesanz et al., 2010). Gender differences in the structural model were tested using a multigroup structural equation modeling strategy and a scaled chi-square difference test (Brown, 2006).

We estimated model fit using the following goodness-of-fit indices: chi-square fit statistics ( ${\chi }^2\left[p>.05\right]$), comparative fit index ( $\mathrm{CFI}>0.95$), standardized root mean residual ( $\mathrm{SRMR}\le 0.08$), and root mean square error of approximation ( $\mathrm{RMSEA}<0.06$) with a 90% confidence interval (CI; Schreiber et al., 2006). We used Mplus 8 and SPSS 22.0 for all analyses.

The proportion of missingness in the data ranged from 0 to 3.5%, with an average of 1.7%. Although there is no established cutoff point on an acceptable percentage of missing data, less than 5% is considered inconsequential for valid statistical inferences (Schafer, 1999), and a less rigorous missing data method can be applied (Little et al., 2014). To explore missing data mechanisms, we performed Little’s missing completely at random test; results showed that missing data were not missing completely at random, indicating that they were either missing at random or missing not at random, also called nonignorable missingness (Garson, 2015). Considering the low proportion of missingness in the data and the missing data mechanism, we handled missing data using full information maximum likelihood (FIML) in Mplus (Enders, 2010). Although FIML assumes that data are missing at random, it provides superior performance to listwise and pairwise deletion methods even with nonignorable data (Wothke, 1998). FIML does not impute or delete missing values, but it uses all available data, including observed responses in the incomplete cases (Mazza et al., 2015).

Descriptive Analyses

Table 1 presents the sample characteristics and prevalence of depressive symptoms in our sample of KA adolescents. Mean age was 14.99 ( $\mathit{SD}=1.67$), and more than half of respondents (53.1%) were girls. Two-thirds of the sample (66.4%) were second-generation (U.S. born), and almost all (90.0%) lived with both parents. More than half reported that their mother (51.9%) and/or father (59.9%) were at least college graduate or above. More than half (53.7%) reported problematic levels of depressive symptoms (a cutoff score of 16 or above).

Mean Values of Major Variables and Group Mean Comparisons by Gender

Table 2 presents the mean values of major variables for all KA adolescents and the group mean comparisons of the major variables by gender, as determined by linear regression. Age and generation status were included as covariates in these comparisons. For all KA adolescents, mean values for interparental conflict and parent–adolescent conflict were 0.82 ( $\mathit{SD}=0.49$) and 2.67 ( $\mathit{SD}=0.90$), respectively. The mean value for problem-focused disengagement coping was 3.20 ( $\mathit{SD}=0.77$). The mean values for self-esteem and depressive symptoms were 2.75 ( $\mathit{SD}=0.55$) and 0.91 ( $\mathit{SD}=0.56$), respectively.

There were significant gender differences in the mean values of parent–adolescent conflict ( $t=1.97$, $p<.05$), self-esteem ( $t=5.66$, $p<.001$), and depressive symptoms ( $t=4.05$, $p<.001$). For boys, parent–adolescent conflict ( $M=2.77$) and self-esteem ( $M=2.93$) values were significantly higher than those for girls ( $M=2.58$ and 2.60, respectively). Conversely, the mean value for depressive symptoms was higher among girls ( $M=1.03$) than boys ( $M=0.78$).

Structural Equation Model

We performed an initial exploration of the model in Figure 1 to identify any sources of poor fit that might require revision. This initial model yielded a poor fit to the data with respect to overall global fit indices ( ${\chi }^2\left(\mathit{df}=3\right)=29.071$, $p<.05$; $\mathrm{CFI}=0.873$; $\mathrm{SRMR}=0.056$; $\mathrm{RMSEA}=0.160$; 90% CI [0.11, 0.22]). Fit statistics (the absolute standardized residuals and modification indices) suggested the need for model respecification. Based on prior theory and evidence (E. Kim & Cain, 2008; Huh-Kim, 1998; Gerard et al., 2006), we revised the initial model by adding two paths with the largest modification indices: from interparental conflict to depressive symptoms and from parent–adolescent conflict to depressive symptoms. The revised structural model, including covariates (age, gender, generation), yielded a good model fit for the global fit indices ( ${\chi }^2\left(\mathit{df}=1\right)=1.703$, $p>.05$; $\mathrm{CFI}=0.998$; $\mathrm{SRMR}=0.008$; $\mathrm{RMSEA}=0.046$; 90% CI [0.00, 0.16]). Fit indices revealed no theoretically meaningful points of poor fit. Figure 2 displays the unstandardized path coefficients for KA adolescents. All path coefficients were statistically significant: from interparental conflict, parent–adolescent conflict, and problem-focused disengagement coping to self-esteem (path $\mathrm{coefficients}=\mathrm{-0.16}$, −0.09, and −0.14, respectively); from interparental conflict and parent–adolescent conflict to depressive symptoms (path $\mathrm{coefficients}=0.21$ and 0.11, respectively); and from self-esteem to depressive symptoms (path $\mathrm{coefficient}=\mathrm{-0.55}$).

Table 3 shows direct, indirect, and total effects using bootstrapping (2,000 iterations). Significance tests for the total effects used the logic of the joint significance test (MacKinnon et al., 2002). The path coefficients for every path in the implied mediational chains were statistically significant for the mediator (all critical ratios $\left[\mathrm{CR}\right]>1.96$, $p<.05$). The effects of interparental conflict and parent–adolescent conflict on depressive symptoms were partially mediated by self-esteem (the significant direct and indirect effects are shown). The effect of problem-focused disengagement coping on depressive symptoms was fully mediated by self-esteem (the significant indirect effect is shown in Table 3). The total effect of interparental conflict on depressive symptoms was estimated to be 0.29 ( $\mathrm{CR}=4.63$, $p<.001$, 95% CI [0.17, 0.41]), and 0.08 of the units of change were attributable to the mediational chain through self-esteem ( $\mathrm{CR}=2.48$, $p<.05$, 95% CI [0.02, 0.15]). The total effect of parent–adolescent conflict on depressive symptoms was estimated to be 0.16 ( $\mathrm{CR}=4.98$, $p<.001$, 95% CI [0.10, 0.22]), and 0.05 of the units of change were attributable to the mediational chain through self-esteem ( $\mathrm{CR}=2.74$, $p<.01$, 95% CI [0.02, 0.09]). The total effect of problem-focused disengagement coping on depressive symptoms was estimated to be 0.08, and all the units of change were attributable to the mediational effect through self-esteem ( $\mathrm{CR}=3.53$, $p<.001$, 95% CI [0.03, 0.12]).

We tested the hypothesized structural model for KA boys and girls using a multigroup structural equation modeling method. The overall fit of the model was good ( ${\chi }^2\left(\mathit{df}=2\right)=2.724$, $p>.05$; $\mathrm{CFI}=0.997$; $\mathrm{SRMR}=0.011$; $\mathrm{RMSEA}=0.046$; 90% CI [0.00, 0.17]). Focused fit indices showed no theoretically meaningful points of poor fit. Figure 3 presents the unstandardized path coefficients for KA boys and girls separately. All the path coefficients in the model were constrained as equal to test for invariance, and they were similar between the two groups, which indicates no gender differences in the structural model.

Limitations

This study has several limitations. First, convenience sampling introduced self-selection bias, which limits the generalizability of findings. Participants were recruited mainly from Korean ethnic churches in New Jersey. These participants live in ethnically dense communities or have easy access to other KA peers, which may affect the results. Along with geographic limitations, the sample is skewed toward adolescents who have two parents, one or both with a college or graduate degree, as well as toward those with Protestant religious affiliations. It is possible that a different sample of KA adolescents would have yielded different results. Second, in addition to self-report data from adolescents, future research should include parents’ reports of parent–child conflict and interparental conflict to develop a more comprehensive picture of family conflicts. Third, cross-sectional survey data make it difficult to address the causal associations between the key variables of interest. Longitudinal research is needed to detect developmental changes in the target population and to examine the mediational effects of self-esteem on the relationships between interparental conflict, parent adolescent conflict, problem-focused disengagement coping, and depressive symptoms among KA adolescents across time.

Implications for Practice and Conclusion

Our results offer insight into the mental health status of KA youth of immigrant families by examining the understudied domains of family conflicts and coping strategies, and their relationship to self-esteem and depression. Findings suggest that greater family conflicts and disengagement coping strategies can affect self-esteem and depressive symptoms among KA adolescents. Clinical interventions for KA adolescents and their families should be considered in the broader context of acculturation gap, loss of common language and discrepant values between the immigrant parents and their children and its impact on family relationships. Prevention and intervention strategies may include immigrant-family-specific parenting and youth programs, if possible, offered through Korean cultural schools and religious organizations, as these institutions are highly respected in KA communities. Family-centered programs should focus on helping families anticipate and navigate intergenerational acculturation issues by normalizing these experiences and teaching immigrant families the skills to strengthen intergenerational communications and parent–child bonding. Additionally, it would be instructive to provide resources and supports for immigrant parents to help their children cultivate positive self-esteem, self-efficacy, and effective coping strategies in light of intergenerational tensions stemming from acculturation issues.