Affiliates tend to be similar on key attributes. This phenomenon, known as homophily, is of particular interest during adolescence because of concerns that friends and romantic partners promote maladaptive behaviors and interfere with academic achievement at a time when youth are especially susceptible to peer influence. Such claims have been sharply criticized, however, for overstating peer socialization effects and ignoring the fact that youth tend to select peers who resemble themselves as friends and romantic partners. Progress reconciling these dissonant views has been slowed by the significant methodological challenges posed by the nonindependent data characteristic of participants in close relationships (i.e., mutual influence between partners that makes them more similar to one another than to random others but that also biases statistical analyses). As a result, scientific understanding of similarity between close peers is incomplete and there is almost no empirical evidence describing which individuals are most apt to change to increase their similarity with friends and romantic partners. The project utilizes archival data from two longitudinal projects: (a) the 10 to 18 Project, a 5-year community study of all students enrolled in the 4th through the 12th grade in a small city in central Sweden; and (b) Project STAR, a 5-year study of friend and romantic relationships in Denver youth ages 14 to 19. These studies contain complementary information about homophily that encompasses a variety of educational outcomes and factors known to constrain educational achievement. The project has two objectives: (a) specify selection and socialization effects in domains that reflect learning and in contexts that constrain learning, and (b) identify characteristics of youth who are susceptible to influence from friends and romantic partners. To meet these objectives, new analytic techniques that measure homophily will be devised and refined. Analytic strategies designed for nonindependent data will estimate selection and socialization over time, disentangling partner influence from individual stability.
The project will have a profound impact on the scientific study of peer relationships and, more generally, on assumptions about the influence of peers over academic achievement and adjustment. In terms of impact on the scientific community, the dyadic data analytic techniques devised to assess homophily will set a new methodological standard for research on the topic, rendering obsolete past empirical practices and revising the conventional wisdom about peer influence over development. In terms of impact on adolescent learning and contexts that constrain it, the project will lay the groundwork for advances in education and public health. Current efforts to promote academic achievement and curb deviant peer influences assume a model in which relationship participants are mutually influential. School curricula, instructional activities, and peer interventions will be more effective if influence agents are distinguished from influence recipients, with programs specifically crafted for each.