Contact Name
Barbara Dietsch
562-985-9488

Contact Email
bdietsc@WestEd.org

Peter Bearman, Ph.D.
Hannah Bruckner, M.A.
Lazarsfeld Center for the Social Sciences
Columbia University

Considerable methodological challenges confront social scientists working with peer data. Drawing on nationally representative data from the National Longitudinal Study of Adolescent Health (Add Health; for more information, see http://www.cpc.unc.edu/addhealth/), this study addresses these problems and shows that peers do matter in the context of adolescent health behavior. Peer effects on sexual debut and pregnancy (given sexual debut) are analyzed for five different levels of peer contexts: single best friends, immediate friendship circle, larger peer group, leading crowd of the school, and the school itself. Add Health collected data for each student of a school as well as complete network data, allowing us to identify an adolescent's network position, her friends and their characteristics. We observe friendships, friends' characteristics, romantic partners, and individual risk factors before a girl experiences sexual debut or pregnancy. This helps in separating selection from influence effects. We start from a baseline model of sexual debut (including 5070 girls) and pregnancy risk (3015 sexually active girls), controlling for socio-demographic, family, and individual characteristics as well as adolescents' risk status and involvement in romantic relationships. For the analysis of pregnancy risk only, we also control for timing of sexual debut and contraceptive use. We then identify peers' characteristics in terms of the most important individual risk factors for sexual debut and pregnancy. We find that:

• Peer influence on sexual behavior is a significant factor. Subsequent research risks misunderstanding key parts of the puzzle of adolescent sexual behavior and pregnancy if network and peer effects are not considered carefully.

• Peer influence operates at multiple levels of peer context, and the level social scientists (and perhaps parents) have spent the most time on - best friends - is the least important. Much more important influences arise from the network of close friends and the larger peer group in which adolescents are embedded. Successful peer intervention programs may well be those that build from the natural affinity groups that organize the relational world of adolescents.

• The risk status of friends seems to matter less for sexual debut than for pregnancy. Programs designed to help girls delay intercourse may benefit from concentrating on the larger peer group, where these groups are disproportionally composed of high-risk teens. High-risk peer groups (and being member of a peer group) are associated with increased likelihood of sexual debut but not pregnancy (given intercourse). Gaining a better understanding of the nature and mechanisms of peer influence in these larger groups should be an important goal for further studies.

• Much of the peer influence we can observe is positive. At all levels of peer context, the strongest influence effects are those carried by low-risk friends. Girls who have low-risk friends in their immediate circle of friends or in their peer group are much less likely to transition to intercourse or experience a pregnancy than girls who have average or high-risk friends. Adolescent girls appear able to block or ignore the attitudes and behaviors of high-risk girls in their networks. The results presented in this paper suggest that adolescent girls are generally good judges of the others who are around and in relation to them, and are able to counteract potential negative influences.

• The effect of male and female friends' risk status may be different. The risk status of female best friends seems to be associated with sexual debut but not pregnancy, while the opposite obtains for male best friends. Best male friends may have a more direct effect than female friends.

• How close children are to their parents is as important as how close their friends are to their parents. Girls whose friends have bad relationships to their parents are at increased risk for an early transition to first intercourse. Parents should not only make sure that their daughters know that they care about them, but also know who their daughters' friends are, and they should try to assess how close these friends are to their parents.

Considerable methodological challenges remain which make it difficult to understand how peer influence operates. Our sense is that the direct influence conceived of as "peer pressure" is significantly less important than more diffuse dynamics associated with modeling. Pressure, if it is salient, is likely to appear in proximal relationships. But we observe peer influence operating among the more distal contexts. It is hard to see how a peer group (characterized by heterogeneity of behavior and attitudes) can pressure an individual. It is easier to see how modeling dynamics may operate to draw a girl closer to the risk-profile of the group she is a member of. Understanding mechanisms of influence is critically important if we are to harness the power of peers to externally designed intervention programs.

Room for error is large in peer programs. Our findings show absolutely no effect of leading crowds on individual behavior. Yet those involved in peer programs almost routinely think that the diffusion of new ideas and practices will not be successful unless first adopted by the adolescents in the leading crowd. If the leading crowd does not shape adolescent sexual behavior (it certainly shapes other adolescent behaviors), such efforts are misplaced. Likewise, our research shows few significant school-level effects on either sexual debut or pregnancy risk. The relevant sphere of influence is smaller than schools, just beyond the rim of best friends, but well below the whole school environment. Programs oriented towards schools may be less successful than those that build from and make use of the naturally occurring affinity groups that arise within schools when adolescents select and reject friends.

This article highlights some of the results of a study sponsored and published by the National Campaign to Prevent Teenage Pregnancy in Power In Numbers: Peer Effects on Adolescent Girls' Sexual Debut and Pregnancy