Effects of Peer Influences and Life-History Strategy on Chinese Junior High School Students’ Prosocial and Antisocial Behaviors

Peer influence and life-history strategy have been shown in previous research as facilitators of adolescents’ social conduct. The current research uses the data from a two-wave, nationally representative survey of Chinese junior high school students to examine how different aspects of peer influence and life-history strategy in Grade 7 might contribute to prosocial and antisocial behaviors in Grade 8. We also considered differences between local and migrant students. The results showed that friend prestige predicted more prosocial behaviors and less antisocial behaviors, whereas friend deviancy predicted less prosocial behaviors and more antisocial behaviors. Moreover, the facilitating effect of friend deviancy on antisocial behaviors was amplified in migrant students more than for local students. Slow life-history strategy predicted more prosocial behaviors and, especially for migrant students, less antisocial behaviors. These findings indicate adolescents’ migrant backgrounds deserve extra attention when investigating peer influence and life-history strategy as distinct contributors to adolescents’ social conduct.


INTRODUCTION
The transition to adolescence (which coincides with the entrance of the junior high school in many societies) has critical implications for the development of social behaviors (Chang et al., 2007), including prosocial behaviors (e.g., helping others and adhering to public rules) and antisocial behaviors (e.g., aggression and substance use). Research on adolescents' social development generally follows the traditional risk-factor approach (Loeber, 1990) without theorizing the qualitative differences among different types of "risk factors" or "protective factors." The current research sought to fill this research gap by highlighting the difference between current social environments (e.g., peer influence) and life-history factors. The latter represents an interaction between genetic endowment and earlier developmental environments. Utilizing large, nationally representative samples of Chinese adolescents, we examined predictions derived from the peer socialization theories (Dishion et al., 1996;Brechwald and Prinstein, 2011;Dishion and Tipsord, 2011) and the life-history theory (Del Giudice et al., 2015). These samples also allowed us to examine hypotheses related to migrant students, who are common in China, and provides additional insights into how deviancy training and lifehistory adjustments work when teenagers try to adapt to a new environment.
Peer influence on social behaviors becomes especially prominent during adolescence (Dishion et al., 2004;Steinberg and Monahan, 2007). It is defined as a combination of two processes: selection (preferentially affiliating with like-minded friends with similar behavioral proclivities) and socialization (a tendency to become similar to one's friend over time). Recent research has mostly devoted to explicating the socialization effect of peer groups (Brechwald and Prinstein, 2011), which may work in two directions. In the positive direction, peers may serve as prestigious role models (Wentzel et al., 2004). Peer acceptance and group membership have been shown to motivate academic achievement and prosocial behaviors among adolescents (Wentzel and Caldwell, 1997). Friends might also serve as. Indeed, research has shown that friends' prosocial behavior predicted positive changes in 6th-graders' prosocial behavior via prosocial goals (Wentzel et al., 2004;Barry and Wentzel, 2006).
Nevertheless, the bulk of research on peer influence focused on negative processes that predict antisocial outcomes such as substance use, academic failure, high-risk sexual behavior, delinquency, and violence in adolescence (Dishion et al., 1995(Dishion et al., , 1996(Dishion et al., , 1997(Dishion et al., , 2004Ary et al., 1999;Fergusson et al., 2002;Tolan et al., 2003;Villanti et al., 2011). Specifically, the deviancy training theory (Dishion et al., 1996;Dishion and Tipsord, 2011) highlighted adolescents' endorsement of deviant behaviors through associating with deviant friends that encourage and amplify antisocial tendencies. Indeed, studies utilizing multilevel social network analysis that tracks students' friendship networks at different time points found that influence from deviant peers might be responsible for aggression and delinquency (Monahan et al., 2009;Sijtsema et al., 2010). Overall, it seems that the type of peers and friends an adolescent has is more influential than mere peer acceptance or friendship on adolescents' prosocial and antisocial behaviors. Therefore, based on peer socialization theories, we hypothesized that friend prestige (i.e., having friends who are well-adjusted at school, especially in academic performance) should exhibit increased prosociality, whereas friend deviancy (i.e., having friends with antisocial behavioral problems) should exhibit increased antisocial behaviors.
Another set of risk/protective factors for adolescents' social conduct can be traced to individual differences in life-history strategy (LHS; Chang et al., 2019), which refers to a suite of traits resulting from gene-environment interactions in the direction that maximize reproductive success in various environmental conditions . Such geneenvironment interactions lead to fundamental tradeoffs between traits facilitating immediate reproductive efforts and those facilitating future-oriented somatic efforts (Chisholm, 1993;Del Giudice et al., 2015;Chang and Lu, 2018). The traits serving similar functions in such tradeoffs are called LHS, which can be represented by a single fast-slow continuum Chen and Chang, 2016). Human LHS is sensitive to ones' developmental environment (especially early environment; Belsky et al., 1991;Chisholm, 1993) and exhibits plasticity in behavioral and social aspects such as attachment, sexuality, selfregulation, health behaviors, and relationship quality (Figueredo et al., 2006;Ellis et al., 2009;Del Giudice et al., 2015). Specifically, slow strategies are characterized by secure attachment, late reproduction, sexual restraint, stable relationships, high selfcontrol, and prosociality, whereas fast strategies are characterized by insecure attachment early reproduction, sexual promiscuity, unstable relationships, impulsivity, and exploitative tendencies (Del Giudice, 2009;Del Giudice et al., 2015).
The transition from childhood to adolescence marks an opportunity to adjust one's LHS before attaining sexual and reproductive maturity (Del Giudice and Belsky, 2011). This indicates alternative pathways for adjustments in adolescents' social conduct. For instance, impulsivity and parental uninvolvement have been recognized as strong risk factors for antisocial behavior (Loeber, 1990). These two apparently irrelevant factors converge as part of fast life-history strategies (Chisholm, 1993;Griskevicius et al., 2011Griskevicius et al., , 2013. The life-history pathways toward social conduct are also evidenced by longitudinal studies showing that unpredictable environments that are conducive to fast LHS predicted deviancy in adolescence and young adulthood . Insecure attachment, which indicates fast strategies, was found to predict antisocial, coercive ways of resource-control in middle childhood (Chen and Chang, 2012). Moreover, both cognitive components (e.g., future-oriented planning, cognitive perspective-taking) and emotional components (e.g., stable relationships, empathic concerns) of slow LHS predicted prosocial moral reasoning and judgments (Zhu et al., 2018). Thus, slow (fast) LHS seems to be related to more (less) prosocial behaviors and less (more) antisocial behaviors.
The peer-influence and life-history factors are intertwined in development. Peer groups form among children of similar backgrounds (Cohen, 1977;Carrell et al., 2013), and common environmental conditions shape similar LHS Del Giudice et al., 2015). Accordingly, Dishion et al. (2012) argued that deviant peer clustering constitutes a core part of fast LHS. However, distinguishing between LHS as a product of geneenvironment interactions (Figueredo et al., 2006;Del Giudice et al., 2015) and behavioral conformity to peer influence might be necessary. Therefore, in the current study, we simultaneous examined slow LHS (self-regulation and relationship quality with parents) and peer influence (number of friends and friend characteristics) as independent predictors of adolescents' social behaviors.
The data we used include different subsamples of Chinese junior high school students from both local and migrant backgrounds. The social and economic changes in China over the past few decades led to a considerable proportion of migrant students in Chinese junior high schools (Jia and Liu, 2017) and changes in social environment for youths (Chang et al., 2011). Because family mobility might disrupt adolescents' social networks, they might become more reliant on new friends for social influence and acceptance in the new environment. This might expose them to more deviancy training (Dishion and Tipsord, 2011). Further, aggression and other antisocial behaviors might serve to exert resource control and to earn social status in peers, especially for disadvantageous individuals (Wilson and Daly, 1985;Hawley, 1999Hawley, , 2014. Indeed, research has shown that perceived peer exclusion and the need for peer acceptance are associated with increased antisocial behaviors (Strohmeier et al., 2012;Jia and Liu, 2017). Therefore, we predicted that migrant teenagers, compared with local teenagers in China would be more susceptible to antisocial peer influence, due to their eagerness to gain peer acceptance (Dishion et al., 1996;Strohmeier et al., 2012;Jia and Liu, 2017). This does not mean that migrant students necessarily perform worse than local students. In fact, studies in China showed that migrant children with better relationship qualities demonstrated higher hope, which in turn predicted better school performance (Fang et al., 2016). This is consistent with previous research showing that slow life-history traits serve as protective factors against antisocial behaviors in adolescence and young adulthood ). Hence, we hypothesized that slow LHS might serve as a protective factor against antisocial behaviors among migrant students.
Peer influence and LHS are unlikely to be the only factors contributing to adolescents' prosocial and antisocial behaviors. In the current study, in addition to students' migrant status, their number of best friends, friend characteristics, and LHS, we also included students' gender, current family financial conditions, and educational levels of the parents as control variables. There is ample evidence that girls are more prosocial and less involved in antisocial behaviors (except relational aggression) than boys (e.g., Lu and Chang, 2019). Gender might also moderate the effects of peer influence and socioeconomic status (SES) on adolescents' social conduct (Dishion et al., 1996;Veenstra et al., 2006). Students' SES, indicated by family financial condition and their parents' educational level, has also been shown to be related to prosocial and antisocial behaviors (Veenstra et al., 2006;Piff et al., 2010). Specifically, low SES might lead to stress within families and low parental engagement, which, in turn, contributes to antisocial behaviors due to a lack of parental attention or discipline (Loeber, 1990). Therefore, it is important to include it as a covariate. Overall, this study sought to provide insights into the relative importance of peer influence and LHS on the prosocial and antisocial behaviors of adolescents with different backgrounds, after controlling for gender and socioeconomic factors.

Data
We used student data obtained from the China Education Panel Survey (CEPS; National Survey Research Center, 2015). CEPS is a large-scale, longitudinal survey project conducted in the Chinese mainland by the National Survey Research Center (NSRC) at the Renmin University of China. The survey covered a range of topics such as family contexts, school processes, communities and social structure, and educational outcomes. The CEPS sample is nationally representative, consisting of approximately 20,000 junior high school students (Grade 7 and Grade 9) in 438 classrooms of 112 schools across 28 county-level units. The baseline (Wave-1) survey was completed in the 2013-2014 academic year, and a Wave-2 survey in 2014-2015 tracked 10,279 8th-Graders (Grade 7 in the baseline survey) with a 91.9% followup rate (see Table 1 for a summary of demographic information of the participants). The current study used a combined dataset of Wave-1 and Wave-2 data, which had an overall sample size of 8984 after excluding respondents with missing responses to important items. This overall sample was further divided into three subsamples. For the purpose of this study, we conducted analyses separately using a core national sample (Subsample 1) of 4,861 students (2,536 males), which represent all 2,870 countylevel units in the Chinese Mainland, and a second subsample with a high proportion of migrant students (Subsample 2) of 3,165 students (1,658 males) representing 120 counties with a high proportion of migrant students.

Predictors: Peer Influence
We examined three aspects of peer influence using items in the Wave-1 questionnaire: Number of friends, friend prestige, and friend deviancy. Students indicated how many best friends they had and rated their best friends in several aspects. Specifically, they were asked, "how many of your best friends mentioned above fit in the following descriptions" (1 = none of them, 2 = one or two of them, 3 = most of them). Friend prestige was represented by the average of three items: (1) having good academic performance, (2) studying hard, (3) expecting to go to college (Cronbach's α coefficients were 0.79, 0.78, and 0.80 within Subsamples 1, 2, and 3, respectively). Friend deviancy was represented by the average of five items: (1) skipping classes, (2) criticized or punished for violating school rules, (3) fighting with others, (4) smoking or drinking alcohol, and (5) dropped out of school (Cronbach's α coefficients were 0.83, 0.89, and 0.84 within Subsamples 1, 2, and 3, respectively).

Predictors: Life-History Strategy
We created a composite score of slow LHS using 10 items in the Wave-1 questionnaire. Specifically, students rated their agreement with the following items given their experiences in Grade 6: (1) I would try my best to go to school even if I had reasons to stay at home, (2) I would try my best to finish even the homework I dislike, (3) I would try my best to finish my homework even if it would take me quite a long time, (4) I was able to express myself clearly, (5) I was able to give quick responses, (6) I was a fast learner, and (7) I was curious about new stuff. We also included an item asking the student to rate their confidence in their own future. These items reflect the self-regulation component of slow LHS. Additionally, two items asked the student to rate how well their relationship with their mother and father were, respectively, reflecting the relationship component of slow LHS. These two components were given the same weight in the composite score, which ranged from 1 to 6 with higher scores reflecting slower LHS. Cronbach's α coefficients for the 10-item LHS measure were 0.74, 0.79, and 0.76 within Subsamples 1, 2, and 3, respectively.

Other Predictors
We also gathered the following information from the CEPS dataset: gender of the student (1 = male, 0 = female), migrant status of the student (1 = migrant student, 0 = local student), self-reported current financial condition (1 = very poor, 2 = poor, 3 = average, 4 = rich, 5 = very rich), and the highest educational level attained by either parent of the student ranging from 1 (none) to 9 (Master's degree or higher).

RESULTS
Descriptive statistics are separately presented for the two subsamples in Table 2. In the core subsample (Subsample 1), an independent samples t-test revealed that non-migrant students reported poorer family financial conditions than did migrant students, t(4849) = −4.27, p < 0.001, CI95 [−0.20, −0.08]. Other variables (number of friends, friend prestige, friend deviancy, LHS, highest educational level of parents, prosocial behaviors and antisocial behaviors) did not differ between these two groups of students, ps > 0.08. In the subsample with a high proportion of migrant students (Subsample 2), however, several significant differences were found between the two groups. Compared with migrant students, non-migrant students had more prestigious best friends, t (3163) Table 3). Number of friends, friend prestige, friend deviancy, LHS, gender, migrant status, highest educational level of parents, and financial condition were entered as predictors. To examine the hypothesis that migrant students were particularly susceptible to peer influence, we also entered the interactions between friend prestige and migrant status,

Variable
Subsample 1 Subsample 2 Non-migrant a M (SD) Wave-1 variables (predictors) and between friend deviancy and migrant status. To examine the hypothesis that slow LHS serves as a protective factor against antisocial behaviors particularly for migrant students, we also entered the interaction between LHS and migrant status.

DISCUSSION
Adolescents' prosocial and antisocial behaviors are likely shaped by both peer influence (e.g., Wentzel et al., 2004;Barry and Wentzel, 2006;Dishion and Tipsord, 2011), and their LHS (Chisholm, 1993;Ellis et al., 2009;Del Giudice and Belsky, 2011). This view is supported in this study, which examined nationally representative samples of Chinese junior high school students. Consistent with peer socialization theories, we found that friend prestige measured by academic achievement and aspiration predicted more prosocial behaviors and less antisocial behaviors in both subsamples. Conversely, friend deviancy in terms of aggression, problematic behaviors, and poor school adjustments predicted less prosocial behaviors in the core subsample and more antisocial behaviors in both subsamples. By contrast, number of best friends reported by the student, which might reflect friend acceptance or popularity, only predicted more antisocial behaviors in Subsample 2, which include a high proportion of migrant students. Combined with the finding that migrant students exhibited a stronger link between friend deviancy and increased antisocial behavior than non-migrant students did, this seems to support the view that antisocial behaviors serve as a means to gain acceptance or popularity, especially in migrant youths who crave peer acceptance (Strohmeier et al., 2012;Jia and Liu, 2017). Indeed, migrant students also indicated less prosocial behaviors and more antisocial behaviors than non-migrant students did in Subsample 2, but not Subsample 1. In Subsample 2, although migrant students, on average, had similar numbers of best friends as local students, their friends were less prestigious than those of local students. Thus, both environmental circumstances and migrant status might contribute to adolescents' desire to impress peers with deviancy. However, the link between peer prestige and more prosocial behaviors did not differ between non-migrant and migrant groups. This is understandable, as the prosocial behaviors we examined (e.g., helping others, following rules, and being honest) generally reflect good social adjustment but are unlikely to impress other teenagers. Previous research has shown that the positive peer influence that promotes adolescents' prosociality mainly works through reciprocated friendships, which reinforce prosocial goals (Wentzel et al., 2004).
In both subsamples, slow LHS measured by better selfregulation and warmer relationship with parents predicted more prosocial behaviors and less antisocial behaviors. This is consistent with the results of the existing research (e.g., Chen and Chang, 2012;Zhu et al., 2018) that link prosocial behaviors to future development (Chisholm, 1993), and antisocial behaviors to more immediate goals of resource control (Hawley, 1999;Chen and Chang, 2012). Moreover, slow LHS seems to serve as a protective factor against antisocial tendencies for migrant students. This is consistent with recent evidence that stable relationships with family, which indicate slow LHS, predicted higher levels of hope and academic performance among Chinese migrant students (Fang et al., 2016).
Consistent with existing research (e.g., Lu and Chang, 2019), we found that girls are more prosocial and less antisocial than boys. The highest education levels of the parents also predicted more prosocial behaviors and less antisocial behaviors, which might reflect the effects of both parental socialization and SES. Financial difficulties did not appear to hinder prosociality for Chinese students, but in Subsample 2 we did observe that poor family financial condition was associated with more antisocial behaviors. This supports the view that antisocial behaviors serve as a resource-control strategy mainly for disadvantaged individuals (Wilson and Daly, 1985). However, the inconsistent finding between the two subsamples implicates that environmental circumstance is also important.
In conclusion, a major contribution of this study is to distinguish between the effects of peer influence and LHS on adolescents' social behaviors and examine these predictors simultaneously using the CEPS data. Although the findings are inevitably limited by the format of responses (i.e., mostly self-reported scales) and some measures with less ideal internal consistency, they offer valuable insights into the multifaceted phenomena of adolescent social adjustment. Importantly, adolescents are not all the same. Chinese junior high school students with different backgrounds differed in their susceptibility to antisocial peer contagion and the manifestation of LHS. Building on these findings, future research should examine in more detail distinct peer processes that might be responsible for such differences, and whether these findings can be extended to other cultures.

DATA AVAILABILITY STATEMENT
Publicly available datasets were analyzed in this study. This data can be found here: Chinese National Survey Data Archive, https: //ceps.ruc.edu.cn.

AUTHOR CONTRIBUTIONS
NZ and LC were responsible for the conceptualization of the research. NZ conducted formal analyses and wrote the original draft. HL and LC were responsible for writing and editing the final draft. All authors contributed to the article and approved the submitted version.