نقش روابط خانواده و همسالان در رفتار ضد اجتماعی نوجوانان: مقایسه چهار گروه قومی

Abstract The dominant theories about the development of antisocial behaviour during adolescence are based almost entirely on research conducted with mainstream, white, middle-class adolescents. The present study addresses this significant gap in the literature by examining whether the same model of family and peer influence on antisocial behaviour is applicable to adolescents belonging to different ethnic groups. The sample included 603 adolescents (318 females and 285 males) from four ethnic groups: 68% of adolescents were Dutch, 11% were Moroccan, 13% were Turkish and 8% were Surinamese. The questionnaires assessing antisocial behaviour, quality of parent–adolescent relationship and involvement with deviant peers were completed by adolescents individually at schools. Results show few ethnic differences in the mean level of the assessed constructs: adolescents from different ethnic groups show similar levels of antisocial behaviour, are to a similar degree satisfied with their relationships with parents, disclose as much information to them, and do not differ in their involvement with deviant peers. However, the associations of parent and peer relations with antisocial behaviour differed across the ethnic groups.

Introduction In research on developmental psychopathology, and in clinical work, the child's family is frequently considered the most important factor in the development of antisocial behaviour. Empirical studies have identified a wide range of family factors, including low socioeconomic status, living in a single parent family, marital discord, etc., as consistent covariates of such problems during adolescence (see for reviews Dishion, French, & Patterson, 1995; Lahey, Waldman, & McBurnett, 1999). Among these factors, it appears from recent studies that the quality of the parent–adolescent relationship, a factor that is most proximal to adolescent's everyday experience, bears an especially strong association to antisocial behaviour (Vitaro, Brendgen, & Tremblay, 2000; Bird et al., 2001; Deković, Janssens, & van As, 2003). The quality of the parent–adolescent relationship is a broad construct that refers to an emotional climate or “atmosphere” in the relationship between the parent and the adolescent. As such, the quality of the relationship is seen as a fundamental aspect of the relationship and as the context in which interactions between the parent and adolescent take place (see also Darling & Steinberg, 1993; Dishion et al., 1995). Research findings indicate that a negative relationship quality (i.e. a high level of conflict and a low level of emotional bonding) between adolescents and parents is related to higher levels of externalizing problem behaviour (Deković, 1999). Similarly, Snyder and Huntley (1990) report that the relationship between antisocial adolescents and their parents appears to be characterized by a lack of intimacy, a lack of mutuality, and more blaming, anger, and defensiveness than in normal families. When the parent–adolescent relationship is characterized by negativity, adolescents are probably less likely to internalize parental values and norms. Because parents provide support for conventional behaviour and sanctions against conduct problems, a positive relationship with parents may function as a protection against antisocial behaviour and delinquency. Moreover, adolescents who have a positive relationship with their parents are more likely to communicate with their parents, to tell them about their daily activities, and to disclose to them their thoughts and feelings. Recent findings (Kerr & Stattin, 2000; Stattin & Kerr, 2000) have shown that adolescent disclosure was the strongest predictor of parental knowledge about their adolescent's whereabouts, and parental knowledge has been consistently linked to a low level of antisocial behaviour (for a review see Dishion & McMahon, 1998; for empirical examples see Dishion, Patterson, Stoolmiller, & Skinner, 1991; Jacobson & Crockett, 2000). Therefore, adolescent openness and communicativeness towards their parents seem to be closely related to involvement in antisocial behaviour. Indeed, poor communication with parents appeared to be an important predictor of adolescent delinquency (Cernkovich & Giordiano, 1987), whereas higher levels of adolescent disclosure were found to correspond to lower levels of norm breaking (Stattin & Kerr, 2000) and delinquency (Kerr & Stattin, 2000). The determinants of problem behaviour are not limited to the family. As children approach adolescence, they spend increasing amounts of time with their peers without adult supervision (Mounts & Steinberg, 1995) and peers become the most important reference group for adolescents (Hartup, 1999). An aspect of peer relations that has emerged as the most prominent predictor of several kinds of problem behaviour is the association with deviant peers. Research findings showed that adolescents who are involved with deviant peers exhibit more norm-breaking behaviour (Brendgen, Vitaro, & Bukowski, 2000), more substance use (Aseltine, 1995), more school problems (Berndt & Keefe, 1995), and more antisocial behaviour in general (Dishion et al, 1991). An explanation offered for these findings (Dishion, Andrews, & Crosby, 1995) is that deviant peers provide opportunities to engage in antisocial behaviour and supply the adolescents with attitudes, motivation, and rationalization to support such behaviour. In addition, exposure to deviant peers may foster antisocial behaviour through positive reinforcement and through modelling of new types of problem behaviour. In sum, both the family and peers play major roles in explanations of adolescent antisocial behaviour. Research findings up to now seem to support the model of family and peer influence presuming that a negative, conflictual parent–adolescent relationship contributes to adolescents’ problem behaviour directly as well as indirectly through deviant peer associations (Kim, Hetherington, & Reiss, 1999). However, the dominant theories about the nature and impact of these relationships are based almost entirely on studies conducted with white, Western, middle-class adolescents from the dominant culture. The generalizability of this model to members of different ethnic groups needs to be tested. Besides providing information about the generalizability of a theory, it is important to answer the question about the appropriateness of interventions for minority adolescents and their families. Minority children constitute rapidly growing segments of the populations in most Western countries. Available data show that ethnic minority adolescents in The Netherlands run considerably higher risks than Dutch adolescents for maladaptive functioning and mental health problems (Junger & Hean Marshall, 1997). In the present study we address this significant gap in the literature on the role of family and peers in adolescence by testing for ethnic group differences in mean levels of antisocial behaviour, perceptions of the quality of the parent–adolescent relationship, and involvement with deviant peers in four ethnic groups living in The Netherlands: Dutch, Turkish, Moroccan, and Surinamese. The examination of these mean differences is a first step. The next question focuses on a more complex and interesting issue: whether the same theoretical model of parent and peer influence on problem behaviour is applicable to different ethnic groups. Such an approach can increase our understanding of how social relations affect developmental outcomes in different cultural settings and has been recommended recently by several researchers (Rowe, Vazsonyi, & Flannery, 1994; Garcia Coll, Akerman, & Cicchetti, 2000; Greenberger, Chen, Beam, Whang, & Dong, 2000). With few exceptions (Smith & Krohn, 1995; Lamborn, Dornbusch, & Steinberg, 1996), most of the studies that have examined similar questions have compared adolescents from one minority group with adolescents from the majority cultures (see for example Forehand, Miller, Dutra, & Chance, 1997). By including four, rather than two groups, the present study can provide a better test of generalizability. We selected three minority groups, Moroccan, Turkish, and Surinamese, as appropriate populations for studying these issues for the following reasons. First, they are the three largest ethnic groups in The Netherlands. Second, these ethnic groups are distinctively different from the Dutch majority group in their cultural values. In contrast to the more individualistic Dutch culture where autonomy and independence from others is highly valued, these three groups value interpersonal relations, collectivism, conformism and social harmony (Janssens, Pels, Deković, & Nijsten, 1999). Still, there are substantial differences among these three ethnic minorities in historical background, immigration history, and religious values. For example, both Moroccan and Turkish immigrants came to the Netherlands in the 1970s for economic reasons, probably with little knowledge regarding Dutch culture. Surinam, on the other hand, was until 1975 a Dutch colony, and immigrants were acquainted with the Dutch culture and language before immigration. Generally speaking, both Moroccan and Turkish immigrants are Islamic, but Surinamese immigrants have more diverse religious backgrounds (about 50% being Christian, 25% Hindu, 20% Muslim and 5% not religious). Finally, Surinamese is a more visible ethnic group in a predominantly white society. They might, therefore, be more exposed to discrimination. If the same findings were obtained in such diverse samples, it would give strong support for the generalizability of the theory. Due to the limited research on this issue, it is difficult to formulate specific hypotheses about differences and similarities. It is possible that an identical pattern of association exists in all ethnic groups, as has been found in studies by Greenberger and Chen (1996) and Rowe et al. (1994). But there are also indications in the literature that the same “risk factors” have different effects in different groups, a so-called “process x context interaction” phenomenon. For example, it has been shown that a high degree of parental control, which is associated with problem behaviour in samples of middle class families, has a positive effect on adolescents who grow up in disadvantage neighbourhoods (Deater-Deckard, Dodge, Bates, & Pettit, 1998). These studies, however, examined different socioeconomic contexts within the same culture, rather than cross-ethnic differences, and therefore these results cannot be overly generalized. On the other hand, ethnic minority families are disproportionally represented in lower socioeconomic status and it is possible that similar interaction occurs with ethnicity as well. In the present study, we expect that cross-ethnic variations will result in stronger relations between parent-child relations and adolescent behaviour for two reasons. First, family loyalty and a strong sense of obligation to meet family expectations that exists in these ethnic groups (Phalet & Schönpflung, 2001) suggests that the quality of the parent–adolescent relationship might be more strongly related to adolescent adjustment than in Dutch culture, where parents expect that children develop independence from the family at an early age (Deković, Noom & Meeus, 1997). Second, there is evidence that immigration increases distance between values of parents and the values of adolescents, especially for values dealing with conformity, tradition, and openness to change (Knafo, & Schwartz, 2001). This intergenerational conflict due to migration may result in greater family upheaval and a consequent increase in adolescent problem behaviour. For both of these reasons, a stronger association between parent–adolescent relationships and adolescent antisocial behaviour might be expected for ethnic minorities than for the Dutch group. The first reason suggests that this could be due to the positive aspects of the relationship, whereas the second reason emphasizes negative aspects of the relationship. It seems important, therefore, to assess both positive and negative aspects of the parent–adolescent relationship. On the other hand, there are reasons to expect a weaker association between parent–adolescent relationships and adolescent adjustment among ethnic minorities. Because parents from ethnic minority groups lack knowledge regarding the dominant culture, and have different values and child-rearing practices, they may be less able to support their adolescents and prepare them for functioning in the new society. Because of this, adolescents from minority groups may be more peer oriented, and more vulnerable to influences outside the family. In fact, empirical findings show that adolescents from minority groups identify more with their peers and spend more unsupervised time with them than adolescents from the majority group (Giordiano, Cernkovich & DeMaris, 1993). In other words, the lack of support within the family makes it more likely that ethnic minority adolescents will depend on peers for affiliation and support. Previous research also has suggested that peer influences might be especially strong in rapidly changing groups, where “old ways” are no longer sufficient for successful adaptation in the future (Greenberger et al., 2000). This higher involvement with peers might result in a higher association between peer relations and problem behaviour in ethnic minority groups than in the Dutch group. In sum, the current study seeks to examine whether parents and peers play similar roles in the development of adolescent antisocial behaviour for adolescents from different ethnic backgrounds. We focus on middle adolescence, when many changes occur. Parental control tends to decrease and family relations tend to become more mutual. Adolescents develop interests and concerns outside the family context: they become increasingly more involved with peers and their peer relations change in quality, towards more intimacy and disclosure (Hartup, 1999). In this period there is also a significant increase of adolescent problem behaviour (Loeber, Green, Lahey, Frick, & McBurnett, 2000). Moreover, it has been found that cultural differences in the level of adolescent problem behaviour do not emerge until middle adolescence (Greenberger et al., 2000). This developmental period, therefore, seems to be the most appropriate for examining the above question. Studies on the development of antisocial behaviour have focused almost exclusively on males and often have been conducted with selected samples: adolescents who live in high-risk neighbourhoods, youth offenders or clinical samples (e.g. Stoolmiller, 1994), which further limits the generalizability of the findings. Thus, the present study extends previous work by examining correlates of antisocial behaviour in an ethnically diverse community sample, including adolescents of both gender.

Results Ethnic, age and gender differences Table 1 shows the means and standard deviations of all assessed variables separately for the four ethnic groups. Table 1. Means and standard deviations of assessed variables Variables Dutch Moroccan Turkish Surinamese Total M S.D. M S.D. M S.D. M S.D. M S.D. 1. Antisocial behaviour 1.43 0.54 1.37 0.65 1.51 0.60 1.48 0.51 1.44 0.56 2. Positive quality 3.06 0.95 3.04 0.80 2.92 0.87 3.00 0.98 3.03 0.92 3. Negative quality 1.98 0.70 1.90 0.67 2.21 0.74 1.99 0.70 2.00 0.70 4. Adolescent disclosure 2.79 0.63 2.71 0.63 2.60 0.68 2.83 0.61 2.76 0.64 5. Deviant peers 1.69 0.64 1.68 0.67 1.71 0.82 1.86 0.83 1.71 0.68 Table options First, we tested for ethnic group differences in antisocial behaviour, parent–adolescent relationships, and peer deviance. Although our main focus was on ethnic group differences, the existing literature suggests that it is important to include age and gender as well. Therefore, we conducted 4 (ethnicity)×3 (age: 14, 15 and 16 years)×2 (gender) multivariate analyses of variance (MANOVA) for all the variables, followed by univariate ANOVA and post hoc Bonferonni test. The results of these analyses are presented in Table 2. Table 2. Ethnic, gender, and age differences in assessed variables Ethnicity Age Gender Interactionsa F η2η2 FF η2η2 FF η2η2 F η2η2 Multivariateb 1.01 0.01 1.99* 0.02 6.61*** 0.06 Univariate 1. Antisocial behaviour 0.93 0.01 2.54 0.01 14.48*** 0.03 3.79*c 0.01 2. Positive quality 0.75 0.00 0.94 0.00 0.98 0.00 3. Negative quality 3.19* 0.02 0.27 0.00 1.78 0.00 4. Adolescent disclosure 0.95 0.01 0.20 0.00 0.38 0.00 5. Deviant peers 0.19 0.00 3.95* 0.01 20.54*** 0.04 *** p<0.001. ** p<0.01. * p<0.05. a Only significant interactions are reported. b Wilks’Lambda. c Age×gender interaction. Table options The MANOVA revealed a significant main effect of age and gender, but not of ethnicity. None of the interactions were significant in multivariate analysis. The univariate analyses provided a more detailed look at the patterning of group differences. The ANOVA for the antisocial behaviour measure, showed a significant main effect of gender. As could be expected, boys show higher levels of antisocial behaviour (M boys=1.54, M girls=1.34). In addition, a significant age×gender interaction was found for antisocial behaviour. Post hoc Bonferonni test indicated that younger boys and girls (ages 14 and 15) do not differ in level of antisocial behaviour. Significant gender differences emerged only by the age of 16 (M boys=2.08; M girls=1.38). Two measures of the parent–adolescent relationship, the positive quality and adolescent disclosure, did not produce any significant main effects or interactions in the univariate analyses. An ANOVA with the third measure, negative quality of parent–adolescent relationship, showed a significant effect of ethnicity. Post-hoc analyses with Bonferonni test indicated that Turkish adolescents report significantly more negative aspects in the parent–adolescent relationship than the other three groups (see Table 1). Involvement with deviant peers showed a significant main effect of age, with older adolescents reporting increasingly more peer deviance, and a main effect of gender, with boys being more often involved with peers who also show higher levels of deviance (M=1.82), than girls (M=1.62). Comparisons of the variances on the five measures across ethnic groups, based on Levene's test for equality of variance, revealed no significant differences between the groups. That is, all four groups showed a similar degree of variance on all measures. In sum, a small but significant effect of ethnicity was found only in the univariate analyses for one of the five variables: negative quality of the parent–adolescent relationship. Relations between antisocial behaviour, parent–adolescent relationship and deviant peers The second aim of this study was to examine whether the associations between antisocial behaviour and the measures of parent–adolescent relationships and deviant peers vary as a function of ethnicity. First, we examined the bivariate correlations among the variables for the total sample (Table 3). In general, all of the correlations were statistically significant and in the expected direction. The three assessed aspects of the parent–adolescent relationships were related to both adolescent antisocial behaviour and involvement with deviant peers. These three scales were moderately intercorrelated, indicating that each scale contributed unique information about the quality of the parent–adolescent relationship. Involvement with deviant peers showed a strong positive relationship with adolescent engagement in antisocial behaviour. This pattern of associations, showing significant associations among predictors (indicators of the parent–adolescent relationship), mediator (deviant peers) and the outcome (antisocial behaviour), meets preconditions necessary for the test of mediational models (Baron & Kenny, 1986). Table 3. Interrelationship among measures for total sample 1 2 3 4 5 1. Antisocial Behaviour 1.00 2. Positive Quality −0.13* 1.00 3. Negative Quality 0.33** −0.30** 1.00 4. Adolescent Disclosure −0.31** 0.48** −0.26** 1.00 5. Deviant Peers 0.72** −0.12* 0.35** −0.26** 1.00 ** p<0.001. * p<0.01. Table options Next, we evaluated the fit of the “direct and indirect effect model” for the total sample. Analyses began with the fully saturated model in which all paths among the variables were assessed. One path coefficient, from positive quality of parent–adolescent relationship to antisocial behaviour, was not significant (t=1.48, ns). This path was deleted and the model was reassessed. The model produced an adequate fit, χ2 (1)=2.19, p=0.14; RMSEA=0.045, NNFI=0.99, CFI=1.00. Multigroup analysis: comparison of four ethnic groups After evaluating the overall fit of the model in the total sample, we used multigroup comparisons to examine the extent to which this model is consistent across ethnic groups. The covariance matrices for each ethnic group were fitted to the common model. First, all parameters were constrained to be equal across groups (Model 1: constrained model). The fit indices of this constrained model were then compared with those of Model 2: unconstrained model (in which path coefficients were estimated separately within each group). The fit indices of each tested model and the results of model comparison are summarized in Table 4. Table 4. Comparative fit of tested models Model fit indices Model comparison test df χ2 p RMSEA NNFI CFI Δdf Δχ2 p(d) Model 1: Constrained model 40 76.78 0.000 0.081 0.91 0.96 Model 2: Unconstrained model 4 4.66 0.320 0.017 0.99 1.00 Model 1 versus Model 2 36 72.12 <0.001 Table options Results of χ2 differences test comparing a constrained model with the unconstrained model showed that the unconstrained model fit the data significantly better (i.e. path coefficients, as set, differ across the groups). In other words, multigroup comparisons showed that the magnitude of the path coefficients linking parent–adolescent relationship and peers to antisocial behaviour varied across ethnic groups. Fig. 1 shows the result of model fitting separately for each ethnic group. Dashed lines indicate non-significant paths. Some of the findings are of particular interest. First, the quality of parent–adolescent relations and involvement with deviant peers accounted for a significant amount of variance in adolescent antisocial behaviour in each of the ethnic groups. However, the percentage of variance accounted for was much smaller for Surinamese adolescents (23%) than for other the three groups (60%, 47%, and 67%, respectively for Dutch, Moroccan and Turkish adolescents). Second, some of the interrelations among the three aspects of parent–adolescent relations seem to be stronger in the samples of Dutch and Moroccan, than in the samples of Turkish and Surinamese adolescents. In the latter two groups, the negative quality of the parent–adolescent relations was unrelated to the positive quality and to adolescent disclosure. Third, only in the Dutch group both direct and indirect effects of the parents on antisocial behaviour reached significance. In Moroccan, Turkish and Surinamese samples direct parental effect did not reach significance, which is probably due to the differences in sample sizes between the groups. However, the magnitude of beta coefficients for the path linking negative quality and antisocial behaviour, were comparable for Dutch, Moroccan an Turkish sample (0.08, 0.06, and 0.07, respectively). Comparable magnitude of beta coefficients for the path between disclosure and antisocial behaviour, were also found for Dutch, Turkish and Surinamese adolescents (−0.07, −0.07, and −0.06, respectively). Disclosure seems to be more important for Moroccans (−0.16), whereas negative quality seems to be less important for Surinamese adolescents (0.01). Finally, the magnitude of the path coefficient from deviant peers to antisocial behaviour was comparable for Dutch, Moroccan and Turkish group, but this path seems to be much smaller for Surinamese. Path coefficients obtained through the LISREL analysis in sample of (A) Dutch, ... Fig. 1. Path coefficients obtained through the LISREL analysis in sample of (A) Dutch, (B) Moroccan, (C) Turkish, and (D) Surinamese adolescents. Paths with non-significant coefficients are represented by dashed lines.