Peer selection and influence of delinquent behavior of immigrant and nonimmigrant youths: Does context matter?

Abstract

This study examines selection and influence related to delinquent behaviors of immigrant and nonimmigrant adolescents attending three majority-immigrant schools (54% to 65.2% immigrant) and four minority-immigrant schools (11.1% to 25.1% immigrant) in one community. The sample included 1,169 youths (50.4% male; 24.2% immigrant) initially between the ages of 12 and 16 years (M =13.92, SD = 0.85). Results showed that immigrant and nonimmigrant adolescents were similar to their peers on delinquent behaviors, and peer selection and social influence operated in a complementary manner to explain this similarity. The processes did not differ between immigrants and nonimmigrants or between school contexts, suggesting that immigrants do not differ from nonimmigrants on either the prevalence or the processes behind delinquency.

Keywords

adolescence delinquency friendship selection immigrant and nonimmigrant youth peer influence

The image of immigrant youths in Europe today is often stereotypic. Media reports of riots in suburban areas of large western European cities have portrayed immigrants as angry, stone-throwing youths who set cars on fire and attack the police and fire fighters. As research and official records support that immigrants and youths from deprived areas are overrepresented in crime statistics (Brottsförebyggande rådet, 2005, report no. 17), it is easy to get the impression that delinquency is primarily a problem of immigrant youths living in segregated areas, only committing crime together with other immigrants. However, does this image correspond to reality? Can the delinquency of immigrant youths be explained by processes in their peer networks and by the contexts in which these peer formations exist? The aim of the present study was to address these questions by examining peer selection and influence processes related to delinquency of immigrant and nonimmigrant youths in schools in which immigrants were the majority and in schools in which immigrants were the minority.

Why look at peer processes when exploring delinquency?

Peers play an important role in the development of most behaviors during adolescence, and especially in the development of delinquency. Most delinquent acts are done together with peers (Kiesner, Kerr, & Stattin, 2004), and several studies have shown that peers tend to be similarly delinquent (Dishion, Spracklen, Andrews, & Patterson, 1996; Kiesner, Cadinu, Poulin, & Bucci, 2002; Snijders & Baerveldt, 2003). This similarity is referred to as homophily, and describes the tendency for friends to be more similar to each other than non-friends (Cairns & Cairns, 1994; Kirke, 2004). Two processes underlie homophily: peer selection and social influence (Burk, Steglich, & Snijders, 2007; Kandel, 1978). Peer selection refers to the process of people selecting peers based on pre-existing similarities (Byrne, 1971). Social influence, also termed socialization or assimilation, refers to the process of peers becoming more similar over time due to indirect and direct forms of social influence (Kandel, 1978).

Both social influence and selection have been found to play roles in the development of delinquency during early and mid-adolescence (Burk, Kerr, & Stattin, 2008; Burk et al., 2007). More importantly, the two processes have been suggested to work in a complementary manner. Psychosocial theories (Erickson, Crosnoe, & Dornbusch, 2000; Oetting & Beauvais, 1987) suggest that youths initiate friendships with similarly-delinquent peers (selection). Once friendships are established, youths within peer clusters will influence each other to become more similar in delinquent behaviors. Thus there are reasons to expect that both selection and influence effects operate in the development of delinquency with initial selection and subsequent influence.

Why would the peer processes related to delinquency differ between immigrants and nonimmigrants?

It has previously been suggested that there are no reasons to assume that the social processes behind delinquency would differ between immigrants and nonimmigrants (Moffitt, 2006). However, even though the processes might be the same, there might be reasons to expect differences between immigrants and nonimmigrants in the salience of these processes. In terms of selection effects, the probability of choosing deviant peers could be higher for immigrants for two reasons. First, immigrants more often live in disadvantaged neighborhoods where both the frequency and severity of delinquency are higher than in privileged neighborhoods (Sampson & Groves, 1989). Thus, there are more delinquent youths in these kinds of neighborhoods and the chances of choosing a delinquent peer might be higher. Second, for immigrants there is an additional selection effect, as several studies have shown a preference for intraethnic over interethnic peer formations (Baerveldt, van Duijn, Vermeij, & van Hemert, 2004; McPherson, Smith-Lovin, & Cook, 2001; Titzmann, Silbereisen, & Schmitt-Rodermund, 2007). As immigrants have been found to be, on average, more delinquent than nonimmigrants, this increases the chance that immigrants, by choosing other immigrants as their peers, would be more likely than nonimmigrants to choose delinquent peers. In terms of influence, previous findings suggest that immigrants might be more inclined to be influenced by peers because of their lower status. Immigrants and ethnic minorities often rank themselves lower on subjective social status than do majorities (Reitzel et al., 2010) and, as perceived popular adolescents are most influential (Wilson, Karimpour, & Rodkin, 2011), immigrant youths might be more easily influenced than nonimmigrant youths. To summarize, the social processes of selection and influence in the development of delinquency might be the same regardless of immigration status; however, previous findings suggest that these processes might be more pronounced among immigrant than nonimmigrant youths.

Why look at context differences?

There are also reasons to assume that these processes would depend on the context in which they occur. A study on ethnically-diverse classrooms showed that selection among classmates was affected by the proportion of minority members in the neighborhood, with majority members showing less in-group homophily and minority members showing more in-group homophily with increasing numbers of minority members in the neighborhood (Vermeij, van Duijn, & Baerveldt, 2009). This indicates that selection processes vary with the ethnic composition of the context. If and how influence effects vary with the context has, to our knowledge, not been investigated previously.

In the current study we explore the processes related to the delinquency of immigrant and nonimmigrant adolescents in two different school contexts. This is done in four steps. First, we test whether immigrant adolescents engage in more delinquent behaviors than nonimmigrants in different school contexts (immigrant majority and immigrant minority). Based on previous studies we expect immigrant youths and youths in disadvantaged neighborhoods to be more delinquent. Second, we ask whether immigrant adolescents are more likely than nonimmigrant adolescents to select delinquent peers; and third, we explore whether immigrant adolescents are more likely than nonimmigrant adolescents to adopt the delinquent behaviors of their peers. These two questions have not been explored previously. Existing literature indicates that immigrant youths are more delinquent on average than nonimmigrant youths, and immigrant youths have been found to prefer other immigrants as their friends. Therefore we expect to find that immigrant youths are more likely than nonimmigrant youths to select delinquent peers and to be more influenced by their peers’ behavior than nonimmigrant youths. Finally, we ask whether peer selection and social influence processes related to delinquency differed between schools in which immigrants were in the majority and those in which they were in the minority. Based on previous studies on the importance of the context, we anticipated contextual differences, but refrained from formulating specific predictions regarding context.

Because of a growing body of research showing that first generation immigrant adolescents show more positive adaptation than later generation immigrants and natives (e.g., Sam, Vedder, Ward, & Horenczyk, 2006; van Geel & Vedder, 2010), we ran all analyses separating first and second generation immigrants.

Method

Participants and procedure

The sample was from a cohort-sequential study conducted in a city in central Sweden (around 132,000 inhabitants). The target sample consisted of all students attending classrooms of grades 7 to 9 in seven public schools. At Time 1 (T1) a total of 1,559 youths participated and 1,485 youths participated one year later at Time 2 (T2). At the start of the project in 2007, 18.4% of participants reported immigrant backgrounds. This proportion was higher than the average in Sweden at that time (17.3%; Statistiska centralbyrån [SCB], 2010). The town is segregated with some public housing areas that are largely composed of immigrants and some areas with very few immigrants. The seven schools were chosen on the basis of their ethnic composition with the aim of having a wide range of different schools. This resulted in three schools with a majority of immigrant youths (54.3% to 65.2%), and four schools with a minority of immigrant youths (11.1% to 25.1%).

The analytic sample consisted of 1,169 youths (590 males and 579 females) with data from both waves. The number of youths included at each school ranged from 73 (43% of the total sum of pupils in the school) to 323 (94% of the total number of pupils). At the onset of the study, participants ranged in age from 12 to 16 years (M = 13.92, SD = 0.85). In the analytic sample, 323 youths were either born in a country other than Sweden (first generation = 46.4%) or had both parents born outside of Sweden (second generation = 53.6%). Their countries of origin were Middle Eastern countries (38.4%), former Yugoslavian countries (18.3%), Somalia and Eritrea (11.8%), Eastern Europe (5%), Asian countries (6.8%), Western Europe (1.2%), and other countries (18.5%). Forty youths did not indicate their countries of origin.

Questionnaires were administered by trained research assistants during regular school hours. No school staff members were present. The participants were assured of the confidentiality of their answers. Participation was voluntary and the adolescents themselves could decline participation in the classroom. The parents could withdraw permission for their children’s participation at any time by sending in a prepaid post card (1% did so). All pupils, regardless of whether they were participating or not, had to stay in the classroom for one and a half hours. In the middle of the session there was a break with refreshments, and most pupils had sufficient time to take additional breaks. Participants who needed more time or got tired could get an additional hour and a half or could finish the questionnaires when the research assistants visited the school a second time. To avoid fatigue effects, half of the sample started with one questionnaire and the other half started with another questionnaire. The order of the scales was also counterbalanced between each wave. The questionnaires were administered in Swedish, but youths in need of help due to language difficulties had research assistants reading and explaining each question to them. The study was approved by the Regional Research Ethics Committee.

Measures

Peer nominations

Participants were asked to nominate three important peers in their schools. These could be friends, siblings, or romantic partners, but in the present study only those nominated as friends were included. The youths were also asked to nominate a maximum of eight peers who they considered belonged to their peer group in school. Thus, every youth could nominate a maximum of 11 peers at each time point. These nominations were used to delineate peer networks at both measurements separately for each school.

Immigration status

Participants were asked if they were born in Sweden, in a Nordic country, or in another country. Youths born in other countries were asked to write down their countries of birth. The same procedure was done for the countries of birth of their mothers and fathers separately. Two immigration status groups were distinguished: minority youths who were not born in Nordic countries (Finland, Norway, and Sweden) or for whom both parents were born outside of the Nordic countries; and majority youths who were of Swedish and Nordic descent.

Delinquency

The scale consisted of 16 questions about whether the youths had engaged in delinquent behaviors during the past year. The questions were about: shoplifting; vandalizing public or private property; taking money from home; creating graffiti; breaking into a building; stealing from someone’s pocket or bag; buying or selling stolen goods; stealing a bike; being in a physical fight in public or carrying a weapon; and stealing a car, moped or motorcycle (Kerr, Stattin, & Burk, 2010). The responses ranged from 1 (never) to 5 (more than 10 times). The scale showed good internal consistency, with Cronbach’s alphas of .81 at T1 and .92 at T2. Individual mean level differences were examined using the average score of all items. Six categories were created from the average scores for the actor-based models. These six categories were constructed in an identical manner for T1 and T2 and designed to include a meaningful number of adolescents in each category. The first category included adolescents who did not report any delinquent activities during the previous year. The remaining categories reflect adolescents with delinquency scores ranging from: 1.01 to 1.24, 1.25 to 1.40, 1.41 to 1.60, 1.61 to 2.00, and 2.01 to 5.00, respectively.

Subjective socioeconomic status

The youths were asked five questions about their experiences of the economic situations in their families. The items and response options were: “How much money does your family have in comparison to other families?” (1 = We have a lot less money than other families to 5 = We have a lot more money than other families); “Has it happened this semester that you could not go with your friends because you could not afford it?” (1 = Yes, several times to 3 = No); “If you want to buy things that cost a lot of money (e.g., computer, skateboard, cell phone, etc.), do your parents have the money to buy them?” (1 = No, probably not to 3 = Yes, probably); “How is the economic situation in your family?” (1 = My parents often complain about not having enough money to 3 = My parents never complain about not having enough money); and “If you would compare with others in your class, do you have more or less money with which to buy things?” (1 = I have much less money than others in my class to 5 = I have a lot more money than others in my class). The items were all coded so that high values indicated high subjective socioeconomic status (SES) and the scale was created from the mean of the five items. The scale had a Cronbach’s alpha of .71 at both time points.

Plan of analyses

We initially compared majority and minority youths in immigrant-minority and immigrant-majority schools on numbers of peers, delinquency, and SES. We also compared the two types of schools on characteristics of the peer networks.

The primary analyses utilized actor-based models of network and behavioral dynamics which are implemented in the Simulation Investigation for Empirical Network Analyses (SIENA) software program (Snijders, Steglich, Schweinberger, & Huisman, 2007). These models estimate parameters describing changes in peer ties (network dynamics) and changes in individual delinquency (behavioral dynamics) using a continuous-time Markov Chain Monte Carlo (MCMC) approach (Snijders et al., 2007). Specifically, all changes in friendship ties and individual delinquent behaviors observed between the two measurements are modeled as the most probabilistic sequence of events explaining the total amount of observed changes. That is, these models include two dependent variables, one describing changes in peer ties and one describing changes in delinquency, which are modeled simultaneously with each depending on the other. The complexity of the resulting model did not allow for explicit calculation of parameter values, so estimated parameters and their standard errors were derived from iterative computer simulations. Readers interested in more detailed descriptions of these models and statistical formulations of parameters are referred to Snijders and colleagues (Snijders et al., 2007; Snijders, van de Bunt, & Steglich, 2010).

We focused on parameters estimating the effects of immigration status and delinquency on changes in peer formations and the effects of immigration status and friends’ delinquency on changes in adolescent delinquent behaviors. Specifically, three parameters describe the effects of immigration status on peer dynamics: immigration status ego describes the effect of the nominator’s immigration status on selection, immigration status alter describes the effect of nominees’ immigration status on selection, and same immigration status describes the tendency for adolescents to nominate peers with the same immigration status. Three similar parameters describe the effects of delinquency on peer dynamics. An interaction between the immigration status of the nominator and delinquency of nominee (immigration status ego × delinquency alter) was also included to test whether immigrant youths were more inclined than nonimmigrant youths to choose delinquent peers. Another three parameters described the effects of immigration status and peers’ delinquency on individual delinquent behavior dynamics. Effect from immigration status describes differences between immigrants and nonimmigrants on changes in delinquency, delinquent influence (average similarity parameter) describes peers’ delinquent behaviors as predictors of change in delinquency, and the interaction between immigration status and delinquent influence tests whether immigrants are more susceptible to peer influence than nonimmigrants.

Identically-specified models were performed for each school-based peer network. To simplify presentation of these results, we aggregated the results of the seven models using the multi-group option within SIENA (see Snijders et al., 2007). Parameters were estimated separately for the three immigrant-majority schools and the four immigrant-minority schools, and each parameter was tested to determine differences between the two school contexts (Ripley, Snijders, & Preciado, 2011).

Results

Descriptive analyses

The first research question concerned differences in delinquency between immigrants and nonimmigrants in the two types of schools and between immigrants in majority-immigrant schools and immigrants in minority-immigrant schools. The lower part of Table 1 presents means and standard deviations of delinquency across waves, immigration status groups, and school context, as well as the number of adolescents in each of the delinquency categories created for the actor-based models. The results of a 2 (immigration status) × 2 (school context) × 2 (time) repeated measures analysis of variance (ANOVA) revealed a within-subjects effect of time, F(1, 1037) = 19.14, p < .001, the average mean level of adolescent delinquency increasing from T1 to T2. All between-subjects effects were nonsignificant.

Table 1.

Descriptive statistics of peer networks and delinquent behaviors

	School
	Minority-immigrant		Majority-immigrant		Total
	T1	T2	T1	T2	T1	T2
Network characteristics
Participants	817	817	352	352	1169	1169
Number of ties	4,074	4,619	2,311	2,854	6,385	7,473
Nonimmig → nonimmig	2,871	3,351	584	545	3,455	3,896
Nonimmig → immig	455	513	348	392	803	905
Immig → immig	238	232	1,029	1,516	1,267	1,748
Immig → nonimmig	510	523	350	401	860	924
Average degree	4.04	4.64	3.08	4.24	3.63	4.47
Reciprocity index	.66	.70	.62	.67	.64	.69
Transitivity index	.53	.49	.49	.47	.51	.48
Delinquent behavior
M total (SD)	1.15 (.25)	1.22 (.54)	1.10 (.20)	1.20 (.45)	1.14 (.24)	1.22 (.51)
M nonimmigrants (SD)	1.15 (.24)	1.22 (.50)	1.11 (.21)	1.21 (.49)	1.14 (.24)	1.22 (.50)
M immigrants (SD)	1.19 (.32)	1.23 (.54)	1.10 (.20)	1.19 (.41)	1.13 (.25)	1.20 (.46)
Frequency
None	358	364	164	179	522	543
Very low	281	243	98	96	379	339
Low	83	82	27	25	110	107
Moderate	30	45	6	17	36	62
High	45	38	8	14	53	52
Very high	20	39	3	19	23	58
Missing	19	6	27	2	46	8

Note. The average degree describes the average number of adolescents’ friendship nominations. The reciprocity index describes the proportion of ties that are reciprocated; the transitivity index describes the proportion of triadic relationships exhibiting transitive network closure. Delinquency ranges from 1 (none) to 5 (more than 10 times).

Table 1 also displays descriptive statistics of the structural characteristics of the network. As can be seen, the number of ties and the average number of ties for participants (average degree) increased from T1 to T2. About two-thirds of all nominations were reciprocated, and roughly half of the youths became peers with their peers’ peers in triadic relationships (transitivity). At both time points approximately 75% of all ties were between individuals in the same immigration status group (about 20% between immigrants and 55% between nonimmigrants), with the remaining 25% of ties including roughly equal numbers of nonimmigrants nominating immigrants and of immigrants nominating nonimmigrants. While the number of intra- and interethnic ties somewhat differed as a function of school context, the percentage of intraethnic ties was similar in the minority-immigrant and majority-immigrant schools (77% and 71%, respectively).

Peer network and delinquent behavior dynamics: Total sample

We performed a series of actor-based models to test selection effects based on immigrant status (same immigration status) and delinquency (delinquency similarity selection), while controlling for the network structure and homophily of gender, age, and SES. The results, presented in Table 2, indicate that both selection effects were statistically significant across the seven models. That is, similarity between relationship partners on immigration status and delinquency uniquely predicted changes in friendship. In order to test the second research question, whether there was a difference in how immigrant and nonimmigrant youths selected their peers, we included an interaction between delinquency and the immigration status (delinquency alter × immigration status ego). This interaction was nonsignificant in our model, indicating that immigrant youths were not more or less likely to select a delinquent peer than were nonimmigrant youths. Thus, youths selected peers on the basis of same immigration status and similar levels of delinquency, but immigrant youths were not more likely to select delinquent peers than nonimmigrant youths.

Table 2.

Parameter estimates for the actor-based model estimating delinquent selection and influence across all schools (N = 1,169)

	Est.	SE	t-ratio	p
Network dynamics
Immigration status ego	.110	.044	2.50	.012
Immigration status alter	.070	.039	1.79	.112
Same immigration status (selection)	.184	.034	5.41	<.001
Delinquency ego	.037	.024	1.54	.123
Delinquency alter	.028	.022	1.27	.203
Delinquency similartiy (selection)	.427	.164	2.60	.009
Delinq. alter × immigration status	−.346	.230	1.50	.133
Control variables
Outdegree	−3.916	.053	−73.89	<.001
Reciprocity	2.379	.053	44.89	<.001
Transitive triplets	.512	.017	30.12	<.001
3-cycles	−.589	.031	−19.00	<.001
Same classroom	1.141	.034	33.56	<.001
Same gender	.754	.037	20.38	<.001
Same age	.583	.096	6.07	<.001
Same SES	.068	.048	1.42	.157
Delinquency dynamics
Effect from immigration status	.044	.119	0.37	.712
Delinquency similarity (influence)	1.812	.634	2.86	.004
Delinq. influence × immigration status	.411	1.420	0.29	.772
Control variables
Linear shape	−.465	.060	−7.75	<.001
Quadratic shape	.199	.023	8.65	<.001
Effect from gender	−.176	.062	−2.84	.005
Effect from age	−.064	.050	−1.28	.201
Effect from SES	−.046	.040	−1.15	.250

Note. Statistical significance of parameter estimates is determined with a t-ratio, which is obtained by dividing the unstandardized estimate (Est.) by the standard error (SE).

In the third research question we asked whether immigrant adolescents were more likely than nonimmigrant adolescents to adopt the delinquent behaviors of their peers. We tested the influence effect of delinquency (average similarity: influence), which was statistically significant, indicating that peers’ delinquency predicted changes in youth delinquency. In order to test whether there was a difference between immigrants and nonimmigrants in susceptibility of peer influence, we included an interaction between immigration status and friends’ delinquent behaviors (delinquency similarity × immigration status). That interaction was nonsignificant, indicating that immigrant youths were not more likely than Swedish youths to be influenced by their peers’ delinquent behaviors. Thus, youths adopted the behaviors of their peers, but immigrant youths were not more susceptible to delinquent peer influence than the nonimmigrant youths.

Peer network and delinquent behavior dynamics: Comparing school contexts

The fourth research question was whether peer selection and social influence processes related to delinquency differently between the two school contexts. The results are presented in Table 3. As can be seen, the two groups of schools were more similar than dissimilar. None of the homophilic processes of primary interest differed significantly between the two types of schools. That is, selection effects of immigration status and of delinquency in minority-immigrant and majority-immigrant schools did not differ significantly in magnitude. The interaction effects were not significantly different in the two types of schools, indicating that immigrant youths were not more likely to select a delinquent peer in majority-immigrant schools than in minority-immigrant schools. Looking at the influence effect of delinquency, we could see that, even though there was a significant influence effect only in the minority-immigrant schools, the difference between the two types of schools was not significant. Thus, selection processes related to immigration status and delinquency and deviant peer influence did not significantly differ between schools with a majority of immigrant youths and in schools with a majority of nonimmigrant youths. The only differences between the two types of schools that were related to our research questions were the ways immigration status was related to activity and popularity. Immigrant youths in the minority-immigrant schools received significantly more nominations (immigration status alter) and nominated significantly more peers (immigration status ego) than immigrants in the majority-immigrant schools.

Table 3.

Parameter estimates for the multigroup actor-based models of the three majority-immigrant and four minority-immigrant schools

	Schools
	Minority immigrant				Majority immigrant				School type differences
	Est.	SE	t-ratio	p	Est.	SE	t-ratio	p	Δ	p
Network dynamics
Immigration status ego	.19	.05	3.80	<.001	.01	.08	0.13	.901	−1.98	.048
Immigration status alter	.17	.05	3.40	.001	−.07	.07	−1.00	.317	−2.72	.007
Same immigration status (selection)	.28	.05	5.60	<.001	.41	.07	5.86	<.001	1.58	.114
Delinquency ego	.05	.03	1.67	.096	.16	.06	2.67	.008	1.73	.084
Delinquency alter	.05	.02	2.50	.012	.07	.05	1.40	.162	0.22	.826
Delinquency similarity (selection)	.53	.18	2.94	.003	.90	.42	2.14	.032	0.81	.418
Delinq. alter × immigration status	−.01	.29	−0.03	.973	−1.13	.60	1.88	.060	−1.67	.095
Control variables
Outdegree	−4.06	.07	−58.00	<.001	−4.32	.12	−36.00	<.001	−1.96	.050
Reciprocity	2.39	.06	39.83	<.001	2.70	.11	24.55	<.001	2.41	.016
Transitive triplets	.51	.02	25.50	<.001	.56	.04	14.00	<.001	1.20	.230
3-cycles	−.60	.03	−20.00	<.001	−1.58	.11	−14.36	<.001	−8.59	<.001
Same classroom	1.14	.04	28.50	<.001	1.49	.07	21.28	<.001	4.27	<.001
Same gender	.72	.04	18.00	<.001	1.13	.09	12.56	<.001	4.21	<.001
Same age	.76	.12	6.33	<.001	.52	.16	3.25	.001	1.19	.234
Same SES	−.01	.05	−0.02	.842	.27	.10	2.70	.007	2.34	.019
Behavior dynamics
Effect from immigration status	.05	.14	0.36	.797	.06	.20	0.30	.764	0.04	.968
Delinquency similarity (influence)	1.99	.69	2.88	.004	1.39	1.13	1.23	.219	−0.45	.653
Delinquency infl. × immigr. stat.	−.34	1.81	−0.19	.826	1.42	2.42	0.59	.557	0.58	.562
Control variables
Linear shape	−.43	.06	−7.17	<.001	−.55	.10	−5.50	<.001	−1.09	.276
Quadratic shape	.18	.02	9.00	<.001	.26	.05	5.20	<.001	1.37	.171
Effect from gender	−.10	.07	−1.43	.153	−.37	.13	−2.85	.004	−1.88	.060
Effect from age	−.07	.06	−1.17	.243	−.03	.08	−0.38	.708	0.48	.631
Effect from SES	−.02	.05	−0.40	.689	−.11	.08	−1.38	.169	0.08	.936

Note. Statistical significance is determined with a t-ratio, which is obtained by dividing the unstandardized estimate (Est.) by the standard error (SE). Δ indicates the differences between the unstandardized parameter estimates in the two school contexts.

Additional actor-based models were also performed to examine differences between first and second generation immigrants. Specifically, we included parameters that examined whether first and second generation immigrants received more peer nominations, nominated more peers, and were more likely to select peers with the same immigration status as nonimmigrants (alter, ego, and similarity parameters, respectively). Interactions also tested whether first and second generation immigrants were more likely than nonimmigrants to select delinquent peers (immigration status ego × delinquency alter) and whether first and second generation immigrants were more likely to adopt their friends’ delinquent behaviors (average similarity × immigrant status). While some differences between first and second generation immigrants emerged in terms of the number of peer nominations sent and received (second generation immigrants nominated more friends and received more nominations than nonimmigrants in the immigrant-majority schools), all four interactions testing differences between first and second generation immigrants on selection and influence of delinquent peers were nonsignificant in models including all seven schools and those examining the immigrant-minority and immigrant-majority schools.

Discussion

The overall aim of this study was to investigate whether the delinquency of immigrant youths could be explained by processes in their social networks and whether these processes were moderated by the context in which the peer formations occurred. Our results show that the social processes behind the delinquency of immigrant youths are similar to the processes of nonimmigrant youths. Our results also suggest that the ethnic composition of the school does not moderate the homophilic processes, because the selection and influence effects were not significantly different between majority-immigrant schools and minority-immigrant schools. Furthermore, we did not detect any significant differences between first and second generation immigrants on selection and influence related to delinquency. These findings both support and extend previous literature.

Our findings support theories that suggest similarity between peers’ delinquency can be attributed to both selection and influence effects (Erickson et al., 2000; Oetting & Beauvais, 1987). Youths in our study, both immigrant and nonimmigrant, selected similarly-delinquent peers and were influenced by their peers’ delinquent behavior. The novel features of this study were that immigrant youths did not seem to be more likely to select delinquent peers, nor were they more influenced by delinquent friends than were nonimmigrant youths. These results confirm Moffitt’s (2006) proposition about there being no differences in the social processes behind delinquency between immigrants and nonimmigrants. However, we also extended this idea, showing that the social processes behind delinquency did not differ as a function of the ethnic composition of the school. In the overall model we found no differences in selection and influence effects between schools where immigrants were in a majority and schools where they were in a minority. That is, immigrant youths growing up in disadvantaged contexts are not more inclined to choose delinquent peers or to be influenced by their delinquent peers. In other words, neither the levels nor the processes of delinquency can be attributed to the contexts in our study.

How can these results be explained? One possible explanation could be connected to the social status of the immigrants. Previous results have shown that immigrants and minority youths perceive themselves to have lower relative positions in the social hierarchy (Reitzel et al., 2010). However, our results suggest that this depends on the composition of the context, as immigrants in schools with few other immigrants were found to be both more active and more popular than immigrants in schools with many other immigrants. Thus immigrant youths seem to be more active and central in peer networks of schools in which they are the minority.

There are important practical implications of the results showing no differences between immigrant and nonimmigrant youths on the social processes of the development of delinquency. Knowing the processes behind a behavior is important for planning prevention and interventions programs. The programs to fight delinquency will be differently designed depending on the underlying processes. If selection is behind delinquency, interventions must focus on preventing the establishment of antisocial peer formations, but if influence is the process behind delinquency, the focus must be on breaking already-existing antisocial peer formations. In the current study both homophilic processes were found for delinquency and, thus, both types of interventions are needed. More importantly, the social processes were found to be the same for immigrants and nonimmigrants, indicating that intervention programs could be the same for all youths, regardless of immigration status.

Some limitations of the present study should be mentioned. First, the average levels of delinquency are low in the current sample. This might be due to the fact that, by only including youths with nomination data from both time points, we might have missed out on some of the most delinquent youths with the highest school absence. However, the sample is community based as the aim of the study was to explore the social processes related to delinquency in a representative sample. Also, we utilized an analytic technique that models delinquency as a categorical dependent variable to accommodate the distributional properties of this measure. Second, the possible nominations were restricted to within the schools. This was done because not all schools in the town were included in the study. As most delinquent acts take place after school and with peers outside of the school context, it is possible that, by restricting the nominations to the schools, we missed important information about networks and delinquency (Kerr, Stattin, & Kiesner, 2007; Kiesner et al., 2004). However, most peers are captured by in-school nominations (Ennett & Bauman, 1994), and our results are similar to those of studies using similar analytic techniques on Swedish adolescent peer networks that included both in-school and out-of-school peers (Burk et al., 2007, 2008).

Despite these limitations, the study has several strengths. The first strength is the unique sample. It includes schools of various ethnic compositions from a single community followed for two consecutive years, and we used nominations of all schoolmates, not only within the same grade or class. A further strength is the use of peers’ self-reports of their own delinquency. Commonly, respondents’ perceptions of their peers’ behaviors are used. However, perceptions and actual behaviors do not always match (Berndt & Keefe, 1995), as adolescents often overestimate the degree of similarity between themselves and their peers (Erickson et al., 2000). This could lead to overestimations of the strength of peer influence (Aseltine, 1995). By using the nominated peers’ own reports of their delinquency we could be sure we were not overestimating the effects in our results. Finally, when examining social networks it is important to estimate selection and influence simultaneously, and with social network analyses this is possible without excluding a majority of relationships in order to ensure independence of observations. Also, SIENA makes it possible to control for various selection effects. Thus, effects found in this study are over and above the selection effects of age, gender, and socioeconomic status. Hence this study offers new and unique insights into the social processes behind the development of delinquency.

In conclusion, the development of delinquency among immigrants does not seem to be different from that of nonimmigrants, or different in contexts with high proportions of immigrants, than in contexts with few immigrants. These results have implications for interventions to reduce delinquency. Delinquency is not only a problem of immigrants in segregated, low-SES areas, and interventions cannot be limited to that group. As the social processes behind the development of youth delinquency seem to be the same regardless of immigration status and context composition, the interventions can be universal. In fighting youth delinquency, we need to broaden the view of immigrant youths and youths in deprived areas as the main target groups.

Footnotes

Acknowledgments

We acknowledge the support of the teachers, principals, and pupils who participated in the study.

This study was supported by grants from the Swedish Council for Working Life and Social Research.

References

Aseltine

R. H.

(1995). A reconsideration of parental and peer influences on adolescent deviance. Journal of Health and Social Behavior, 36, 103–121.

Baerveldt

Van Duijn

M. A. J.

Vermeij

Van Hemert

D. A.

(2004). Ethnic boundaries and personal choice: Assessing the influence of individual inclinations to choose intra-ethnic relationships on pupils’ networks. Social Networks, 26, 55–74.

Berndt

T. J.

Keefe

. (1995). Friends’ influence on adolescents’ adjustment to school. Child Development, 66, 1312–1329.

Brottsförebyggande rådet (The Swedish National Council for Crime Prevention). (2005). Brottslighet bland personer födda i Sverige och i utlandet [Crime among people born in Sweden and abroad]. Report 2005, no. 17. Stockholm, Sweden: Brottsförebyggande rådet.

Burk

W. J.

Kerr

Stattin

(2008). The co-evolution of early adolescent friendship networks, school involvement and delinquent behaviors. Revue Française de Sociologie, 49, 499–522.

Burk

W. J.

Steglich

C. E. G.

Snijders

T. A. B.

(2007). Beyond dyadic interdependence: Actor-oriented models for co-evolving social networks and individual behaviors. International Journal of Behavioral Development, 31, 397–404.

Byrne

(1971). The attraction paradigm. New York, NY: Academic.

Cairns

R. B.

Cairns

B. D

. (1994). Lifelines and risks: Pathways of youth in our time. New York, NY: Cambridge University Press.

Dishion

T. J.

Spracklen

K. M.

Andrews

D. W.

Patterson

G. R.

(1996). Deviancy training in male adolescents friendships. Behavior Therapy, 27, 373–390.

10.

Ennett

S. T.

Bauman

K. E.

(1994). The contribution of influence and selection to adolescent peer group homogeneity: The case of adolescent cigarette smoking. Journal of Personality and Social Psychology, 67, 653–663.

11.

Erickson

K, G.

Crosnoe

Dornbusch

S. M

. (2000). A social process model of adolescent deviance: Combining social control and differential association perspectives. Journal of Youth and Adolescence, 29, 395–425.

12.

Kandel

D. B.

(1978). Homophily, selection, and socialization in adolescent friendships. The American Journal of Sociology, 84, 427–436.

13.

Kerr

Stattin

Burk

W. J.

(2010). A reinterpretation of parental monitoring in longitudinal perspective. Journal of Research on Adolescence, 20, 39–64.

14.

Kerr

Stattin

Kiesner

(2007). Peers and problem behavior: Have we missed something? In Engels

R. C. M. E.

Kerr

Stattin

(Eds.), Friends, lovers and groups: Key relationships in adolescence. Hot topics in developmental research (pp. 125–153). New York, NY: John Wiley & Sons Ltd.

15.

Kiesner

Cadinu

Poulin

Bucci

(2002). Group identification in early adolescence: Its relation with peer adjustment and its moderator effect on peer influence. Child Development, 73, 196–208.

16.

Kiesner

Kerr

Stattin

(2004). “Very important persons” in adolescence: Going beyond in-school, single friendships in the study of peer homophily. Journal of Adolescence, 27, 545–560.

17.

Kirke

D. M.

(2004). Chain reactions in adolescents’ cigarette, alcohol and drug use: Similarity through peer influence or the patterning of ties in peer networks? Social Networks, 26, 3–28.

18.

McPherson

Smith-Lovin

Cook

J. M.

(2001). Birds of a feather: Homophily in social networks. Annual Review of Sociology, 27, 415–444.

19.

Moffitt

T. E.

(2006). Life-course-persistent versus adolescence-limited antisocial behavior. In Cicchetti

Cohen

D. J.

(Eds.), Developmental psychopathology. Vol. 3: Risk, disorder, and adaptation (pp. 570–598). Hoboken, NJ: John Wiley & Sons.

20.

Oetting

E. R.

Beauvais

(1987). Peer cluster theory, socialization characteristics, and adolescent drug use: A Path analysis. Journal of Counseling Psychology, 34, 205–213.

21.

Reitzel

L. R.

Mazas

C. A.

Cofta-Woerpel

Vidrine

J. I.

Businelle

M. S.

Kendzor

D., … Wetter, D. W

. (2010). Acculturative and neighborhood influences on subjective social status among Spanish-speaking Latino immigrant smokers. Social Science & Medicine, 70, 677–683.

22.

Ripley

R. M.

Snijders

T. A. B.

Preciado

. (2011). Manual for SIENA version 4.0 (version December 11 2011). Oxford, UK: University of Oxford, Department of Statistics, Nuffield College. Retrieved from http://www.stats.ox.ac.uk/siena/

23.

Sam

D. L.

Vedder

Ward

Horenczyk

(2006). Psychological and sociocultural adaptation of immigrant youths. In Berry

J. W.

Phinney

J. S.

Sam

D. L.

Vedder

(Eds.), Immigrant youths in cultural transitions: Acculturation, identity and adaptation across national contexts (pp. 117–141). Mahwah, NJ: Lawrence Erlbaum Associates.

24.

Sampson

R. J.

Groves

W. B.

(1989). Community structure and crime: Testing social-disorganization theory. The American Journal of Sociology, 4, 774–802.

25.

Statistiska centralbyrån (SCB). (2010). Retrieved from http://www.ssd.scb.se/databaser/makro/MainTable.asp?yp=qmnggr&xu=91506001&omradekod=BE&omradetext=Befolkning&lang=1

26.

Snijders

T. A. B.

Baerveldt

(2003). A multi-level study of the effects of delinquent behavior on friendship evolution. Journal of Mathematical Sociology, 27, 123–151.

27.

Snijders

T. A. B.

Steglich

C. E. G.

Schweinberger

Huisman

(2007). Manual for SIENA version 3.1. Groningen, the Netherlands: University of Groningen.

28.

Snijders

T. A. B.

van de Bunt

G. G.

Steglich

C. E. G.

(2010). Introduction to actor-based models for network dynamics. Social Networks, 32, 44–60.

29.

Titzmann

Silbereisen

Schmitt-Rodermund

(2007). Friendship homophily among diaspora migrant adolescents in Germany and Israel. European Psychologist, 12, 181–195.

30.

van Geel

Vedder

(2010). The adaptation of non-western and Muslim immigrant adolescents in the Netherlands: An immigrant paradox? Scandinavian Journal of Psychology, 5, 398–402.

31.

Vermeij

van Duijn

M. A. J.

Baerveldt

(2009). Ethnic segregation in context: Social discrimination among native Dutch pupils and their ethnic minority classmates. Social Networks, 31, 230–239.

32.

Wilson

Karimpour

Rodkin

P. C.

(2011). African American and European American students’ peer groups during early adolescence: Structure, status, and academic achievement. Journal of Early Adolescence, 31, 74–98.