دلبستگی والدین و بزهکاری نوجوانان چینی: نقش واسطه ای عدم مشارکت اخلاقی

Abstract There is substantial literature documenting the negative association between secure parental attachment and lower adolescent delinquency, but little is known about the mediating mechanisms (i.e., how does parental attachment relate to delinquency?) underlying this relation. The present study examined whether secure parental attachment would be indirectly related to lower adolescent delinquency through lower adolescent moral disengagement. A total of 1766 adolescents (44% male; mean age = 14.25 years, SD = 1.54) living in an urban area of southern China completed anonymous questionnaires regarding parental attachment, moral disengagement and delinquency. After controlling for gender, age, socioeconomic status, and school variable, it was found that secure parental attachment was negatively associated with adolescent delinquency and this negative association was fully mediated by the extent of adolescent moral disengagement. These findings contribute to an understanding of the mechanisms underlying the development of adolescent delinquency and have important implications for intervention.

Introduction During the past several decades, increases in adolescent delinquency and its detrimental effects on adolescents' adjustment and public safety have received substantial attention from criminologists, sociologists, and psychologists (Agnew, 2008 and Howell, 2003). Involvement in delinquent behavior leads to many negative consequences, such as poor academic performance, disturbances in social relationships, and even serious, violent, and chronic offenses (Loeber & Farrington, 2000). Thus, it is essential to understand contextual and psychological antecedents in order to identify targeted interventions for adolescent delinquency. The role of parental attachment in adolescent delinquency From an ecological perspective, the family has an enduring effect on the development of delinquent behaviors (Bronfenbrenner, 1979). There is considerable evidence indicating that various aspects of family dysfunction play important roles in the emergence of adolescent delinquency (Hoeve et al., 2009 and Hoeve et al., 2012). Among these factors, parental attachment is considered to be particularly important (Bowlby, 1973 and Hirschi, 1969). Parental attachment refers to an affectional bond between parents and children (Bowlby, 1988). Two theories from the criminological and psychological literature explain the relation between parental attachment and delinquency. Hirschi's (1969) social control theory asserts that adolescents with strong positive affectional bonds to parents are more likely to care about the normative expectations of their parents, which keep them from delinquent behaviors. Bowlby's (1988) attachment theory assumes that children form “internal working models of relationships” based on their attachment experiences with attachment figures (e.g., parents, peers). Children who are in an insecure attachment relationship form maladaptive internal working models of others (e.g., people are untrustworthy or hostile), that lead them to show little affection or concern for others and engage in high levels of problem behaviors to fight against this hostile environment (Bowlby, 1973). There is considerable evidence suggesting that the quality of parental attachment is a powerful predictor of adolescents' delinquency (Allen et al., 2002 and Higgins et al., 2010). For instance, Allen et al. (2002) found that insecure attachment to parents predicted increasing delinquency during mid-adolescence. In another longitudinal study, Higgins et al. (2010) also found that lower or decreasing levels of secure parental attachment throughout adolescence were associated with higher levels of delinquent behaviors. In addition, a meta-analysis indicated that the overall mean effect size for the association between insecure parental attachment and delinquency was significant, with r = .18 ( Hoeve et al., 2012). Although previous studies have examined the effect of parental attachment on adolescent delinquency, most of them have been conducted in western countries such as the United States. Arnett (2008) pointed that the U. S. represents less than 5% of the world's population, but over 98% of developmental research is based on children in that country. Thus, we were interested in examining whether the reported association between secure parental attachment and adolescent delinquency could be generalized to other countries such as China. Given that attachment is generally seen as a universal aspect of the human experience (Bowlby, 1979), it likely will have a similar role in parent-adolescent relationships across different countries and cultures. Cai (2010) proposed that the theory of “internal working models of relationships” likely functions similarly for youth in Chinese culture as it does for youth in Western cultures. Thus, we hypothesized that the protective effect of secure parental attachment on delinquency would also be found among Chinese adolescents. However, so far few studies in China have tested the role of secure parental attachment in adolescent delinquency (Liu, 2011 and Yang et al., 2010). In fact, the features of Chinese culture offer an intriguing research setting for exploring this issue. Since 1980, the Chinese government has implemented the one-child policy and most families only have one child. Parents are inclined to invest a great deal of care and attention in their only child, resulting in forming more interactions and close ties between parents and children (Feng, Poston Jr., & Wang, 2014). Therefore, the parent-child relationship continues to be dominant in adolescents' interpersonal life and plays a very important role in the adjustment of Chinese adolescents. Based on the above theoretical analyses and empirical evidence, we used a relatively large sample of Chinese adolescents to examine the association between secure parental attachment and delinquency. The mediating role of moral disengagement Although the direct link between parental attachment and adolescent delinquency was demonstrated in previous research, there is little known about the mediating mechanisms through which parental attachment is associated with adolescent delinquency. Recently, there has been interest in gaining a better understanding of the mediating variables that may be involved in this relation (e.g., Troxel et al., 2013 and de Vries et al., 2015). Determining the roles of mediators in the relationship between parental attachment and delinquency would be important to advance our understanding of delinquency and provide guidance for interventions to reduce it (Fearon, Bakermans-Kranenburg, Van Ijzendoorn, Lapsley, & Roisman, 2010). In the present study, we aimed to test a conceptual model in which parental attachment would be indirectly related to Chinese adolescent delinquency through moral disengagement (see Fig. 1). Conceptual model showing the proposed relationship between parental attachment ... Fig. 1. Conceptual model showing the proposed relationship between parental attachment and adolescents' delinquency as mediated by moral disengagement. Figure options The concept of moral disengagement is grounded in social cognitive theory (Bandura, 1986 and Bandura et al., 1996) and refers to the “individual's tendency to use mechanisms conducive to a selective disengagement of moral censure” (Paciello, Fida, Tramontano, Lupinetti, & Caprara, 2008, p. 1288). According to social cognitive theory, people always have personal standards of moral behavior that deter them from misbehavior, and misbehavior that is in contrast with personal standards will result in self-censure and guilt (Detert, Treviño, & Sweitzer, 2008). However, the above process can be deactivated selectively through moral disengagement (Bandura, 1999). Individuals with morally disengaged attitudes find ways of justifying immoral behaviors and feel freed from the self-sanctions and guilt induced by their immoral behaviors. Therefore, moral disengagement is considered to be a risk factor for the emergence and development of delinquent behaviors (Bandura et al., 1996, Bandura et al., 2001 and Shulman et al., 2011). The cross-cultural generalizability of social cognitive theory has been demonstrated in terms of its explanatory power and predictive power (Bandura, 2002). Children with morally disengaged attitudes show high engagement in misconduct regardless whether they come from individualistically oriented or collectivistically oriented cultures (Bandura et al., 1996, Bandura et al., 2001, Shulman et al., 2011 and Yang and Wang, 2012). For example, Bandura et al. (2001) found that moral disengagement contributed to U. S. adolescent delinquency over time; similarly, Shulman et al. (2011) found that a reduction in moral disengagement helped to speed decline in self-reported and official records of delinquent behaviors among adolescent male felony offenders in the U. S. In the empirical studies from China, moral disengagement has also been shown to be a critical mechanism underlying delinquent behaviors for adolescents: those adolescents with high levels of moral disengagement show more problem behaviors (Yang and Wang, 2011 and Yang and Wang, 2012). Despite the crucial role that moral disengagement is likely to play in the emergence and development of delinquent behaviors, little is known about its developmental precursors (Hyde, 2007) and to date, only a few studies have focused on demographic, contextual, or individual factors as antecedents (Detert et al., 2008 and Hyde et al., 2010). For instance, Hyde et al. (2010) found significant associations between rejecting parenting, neighborhood impoverishment, and child empathy on the one hand, and later moral disengagement on the other. The present study aims to expand the limited research in this area by investigating the association between parental attachment and moral disengagement. Parental attachment is viewed as an important contributor to children's moral development (Dunn, 2006). One way that attachment might shape moral development is through the influence of internal working models, which in the case of insecure attachment could be seen as similar to biased social information processing (e.g., the world is harsh and uncaring). These biased perspectives may contribute to adolescents dismissing their moral standards, and may generate new assumptions about the acceptability of problem behaviors (Fontaine, Fida, Paciello, Tisak, & Caprara, 2014). This possibility is also consistent with social control theory (Hirschi, 1969), which proposes that the strong affectional bonds (one aspect of secure attachment) between parents and children may encourage adolescents to conform to parental expectations and norms, thus being less likely to develop morally disengaged attitudes that may violate their parents' expectations and disrupt the affection bonds with their parents. According to these theoretical analyses, parental attachment is likely to be a precursor of moral disengagement. As far as we know, no study to date has examined moral disengagement as a mediator in the link between parental attachment and delinquency. Previous studies have indicated that moral disengagement serves as a mediator that helps explain the associations between family risks (e.g., negative parenting) and problem behaviors (Hyde et al., 2010 and Pelton et al., 2004). However, compared with negative parenting, parental attachment is a proximal predictor of delinquent behaviors (Bosmans, Braet, Van Leeuwen, & Beyers, 2006). Thus, we hypothesized that moral disengagement would also mediate the association between parental attachment and adolescent delinquency. In other words, we proposed the following conceptual model: parental attachment would be indirectly related to adolescent delinquency through moral disengagement in Chinese adolescents (see Fig. 1). In summary, this study makes several important contributions to literature. First, most of the previous studies linking secure parental attachment and adolescent delinquency were carried out in the western context, and this study examined whether such an association could also be generalized to Chinese culture. Second, although there is substantial literature documenting the negative association between secure parental attachment and adolescent delinquency, the underlying mechanisms linking parental attachment to adolescent delinquency are not well understood. To address this gap, this study aimed to examine moral disengagement as a mediator. The focus on moral disengagement is aligned with a recent call to investigate individual cognitive processes (Hyde, 2007 and Hyde et al., 2010) as a plausible reason for the association between contextual risk factors (e.g., insecure parental attachment, negative parenting) and adolescent delinquency. Knowledge about internal cognitive mechanisms in delinquent youth may increase our understanding of how contextual risk factors (e.g., insecure parental attachment) are translated into different kinds of delinquent behaviors (Hyde, 2007).

Results Preliminary analyses Descriptive statistics for youths' scores on each delinquent behavior are provided in Table 1. The most common delinquent behavior was cheating on school tests (24.40%) and the least common was stealing from others or shops (1.60%). In addition, the risks of violent acts (10.50%), property damage (8.40%), and alcohol use (8.20%) were relatively high. In the total sample, 36.20% (n = 639) of the adolescents reported at least one kind of delinquent behavior in the previous six months. Table 1. Descriptive statistics for youths' scores on each delinquent behavior. Specific delinquent behavior Distribution of responses (%) None One or more times Robbed, extorted or threatened someone 98.30% 1.70% Ran away from home 96.90% 3.10% Sleepover without guardians' approval 95.50% 4.50% Stolen from home 97.00% 3.00% Skipped classes without an excuse 97.40% 2.60% Stolen from others or shops 98.40% 1.60% Alcohol use 91.80% 8.20% Cheated on school tests 75.60% 24.40% Tobacco use 96.70% 3.30% Gambling 96.80% 3.20% Property damage 91.60% 8.40% Violent acts 89.50% 10.50% Note. N = 1766. Table options Table 2 contains univariate statistics and bivariate correlations for all study variables. Parental attachment was negatively associated with adolescent delinquency, r = −.20, p < .001, indicating that higher levels of secure parental attachment were associated with lower levels of delinquency. In addition, moral disengagement was positively associated with delinquency, r = .39, p < .001, indicating that higher levels of moral disengagement was associated with higher levels of delinquency. Finally, higher levels of secure parental attachment were associated with lower levels of moral disengagement, r = −.33, p < .001. Table 2. Univariate and bivariate statistics for all study variables. Variables 1 2 3 4 5 6 1. Gender – 2. Age −.15∗∗∗ – 3. SES −.05∗ .03 – 4. Parental attachment −.16∗∗∗ .08∗∗ .16∗∗∗ – 5. Moral disengagement .30∗∗∗ −.02∗ −.05∗ −.33∗∗∗ – 6. Delinquency .22∗∗∗ −.14∗∗∗ −.05∗ −.20∗∗∗ .39∗∗∗ – M .44 14.25 .00 3.62 1.91 1.12 SD .50 1.54 1.00 .69 .56 .32 Note. N = 1766. SES = socioeconomic status. Gender was dummy coded such that 0 = female and 1 = male. *p < .05. **p < .01. ***p < .001. Table options Measurement model test Prior to testing the mediation model, a measurement model was first established for the latent variables of parental attachment and moral disengagement using confirmatory factor analysis. The measurement model demonstrated a good fit to the data: CFI = .96, TLI = .94, RMSEA = .07, SRMR = .04. The standardized factor loadings of each indicator on its corresponding factor were significant at p < .001 (see Table 3). Table 3. The measurement model: Latent variable factor loadings. Variables Standardized loading coefficients Parent Attachment Trust .78∗∗∗ Communication .78∗∗∗ Alienation .71∗∗∗ Moral Disengagement Moral justification .76∗∗∗ Euphemistic language .83∗∗∗ Advantageous comparison .78∗∗∗ Displacement of responsibility .71∗∗∗ Diffusion of responsibility .57∗∗∗ Distorting consequences .78∗∗∗ Attribution of blame .74∗∗∗ Dehumanization .78∗∗∗ Note. N = 1766. ***p < .001. Table options Testing for mediation effect To test the mediation hypothesis, a structural equation model examined whether the latent variable of moral disengagement mediated the relationship between the latent variable of parental attachment and the measured variable of delinquency (see Fig. 2). We included adolescents' gender, age, family socioeconomic status, and school variable as covariates in all analyses. Model of the full mediating effect of moral disengagement. Parameter estimates ... Fig. 2. Model of the full mediating effect of moral disengagement. Parameter estimates are standardized coefficients, and standard errors are given in parentheses. Significant paths are indicated by asterisks (***p < .001). For simplicity of presentation, error variances and control variables are not shown. Figure options First, we tested the direct effect of parental attachment on delinquency in the initial model (not depicted). The results indicated that this direct-effect model fit well to the data: CFI = .94, TLI = .90, RMSEA = .07, SRMR = .05. Secure parental attachment was negatively associated with delinquency (β = −.17, SE = .02, p < .001). Second, the mediation model also demonstrated an adequate fit to the data: CFI = .93, TLI = .91, RMSEA = .07, SRMR = .05. As presented in Fig. 2, higher levels of secure parental attachment predicted less moral disengagement (β = −.33, SE = .04, p < .001), and more moral disengagement was associated with higher levels of delinquent behaviors (β = .37, SE = .03, p < .001). The relationship between parental attachment and adolescent delinquency became non-significant after moral disengagement was added to the model (β = −.04, SE = .02, p > .05). Therefore, the association between parental attachment and delinquency was fully mediated by moral disengagement and the indirect effect was significant, ab = −.13, SE = .01, 95% CI = [−.10, −.04], providing support for the hypothesis. It is noteworthy that we use the term “mediate” only in the statistical sense, as the present cross-sectional data do not allow causal conclusions. However, statistical mediation gives us a clue about causal mediation. Supplementary analyses Consistency of findings across genders Because gender differences have been reported in moral disengagement and delinquent behaviors (Lahey et al., 2000, Yang et al., 2010 and Yang et al., 2010), the mediation model was examined across genders. We conducted a series of subgroup analyses (females vs. males). In the first model, all parameters were estimated freely for both subgroups. In the second model, we constrained the three paths pertaining to the mediation effect (a, b, and c', in Fig. 1) to be equivalent across gender. A chi-square difference test was used to compare the unconstrained and constrained models. The unconstrained model was not significantly different from the constrained model (Δχ2/Δdf = 2.61/3, p > .05), indicating this mediation model was not moderated by gender (results are available upon request). Moderated mediation Some previous research has found that there are age or SES differences in moral disengagement and adolescents' delinquency (Lahey et al., 2000, van der Laan et al., 2010, Yang et al., 2010 and Yang et al., 2010). Thus, we examined age and SES as moderators by testing moderated mediation models. That is, we explored the possibility that the indirect effect of parental attachment on adolescents' delinquency through moral disengagement differs based on age or SES. More specifically, we tested whether age or SES moderates the path from parental attachment to moral disengagement and/or the path from moral engagement to delinquency. The results indicated that the path from parental attachment to moral disengagement was not moderated by age (b = .03, SE = .02, p > .05) or SES (b = −.02, SE = .03, p > .05). In addition, the path from moral disengagement to delinquency was also not moderated by age (b = −.04, SE = .02, p > .05) or SES (b = −.02, SE = .05, p > .05). Therefore, the indirect effect of parental attachment on adolescents' delinquency through moral disengagement was not moderated by age or SES. Negative binomial analyses Because a large proportion of our participants reported not engaging in any delinquent behavior, the distribution of delinquency was skewed (skewness = 6.54, SE = .06) and overdispersed (variance greater than the mean; variance = 14.44, mean = 1.42). This type of skewed distribution is common for self-reported delinquent behaviors ( Trinkner, Cohn, Rebellon, & Van Gundy, 2012). Thus, we ran an additional set of negative binomial (NB) analyses in Mplus 7.0 ( Long, 1997 and Muthén and Muthén, 1998-2012). 1 These analyses substantively replicated our results (see Table 4). Table 4. Negative binomial regressions estimate the direct and mediation effects from parental attachment to adolescent delinquency. Direct and mediation effects Negative binomial Estimate SE Direct effect Parental attachment → Delinquency −.50∗∗∗ .08 Mediation effect Parental attachment → Moral disengagement −.33∗∗∗ .03 Moral disengagement → Delinquency .54∗∗∗ .06 Parental attachment → Delinquency −.26∗∗∗ .06 Parental attachment → Moral disengagement → Delinquency −.18∗∗∗ .04 Note. N = 1766. Parameter estimates are standardized coefficients. ***p < .001.