The Socialization of Adolescent Risk Behavior: Parent and Peer Influences in a Large, Longitudinal Sample

Abstract

Previous research has not been explicitly clear about which relational influences, parents or peers, affect individual risk behavior, and vice versa. This may be attributed to the use of various methodological designs. This study examined the reciprocal influences of parents and peers on individual risk behavior by explicitly testing two different sociological theories: the group socialization theory (Harris, 1995) and the stage-environment fit theory (Eccles et al., 1993). Longitudinal data were used from the National Institute of Child Health and Human Development (NICHD) to investigate peer risk behavior, individual risk behavior, and child-parent relationship quality influences in the early to middle adolescent stages of development. Longitudinal cross-lagged latent variable panel models were used to investigate the interrelatedness of these relationships using a general construct of risk behavior. Results supported full strong longitudinal measurement invariance for individual risk behavior and child-parent relationship quality, but only weak invariance for peer risk behavior. The latent parameters (factor variances, latent means) were mostly non-invariant across the stages of development, with increases in individual risk behavior with concomitant decreases in child-parent relationship quality over time. After controlling for previous levels of peer risk behavior, individual risk behavior explained changes in subsequent peer risk behavior and in the same magnitude (β = .33 to .61) across the stages of development. Peer risk behavior did not explain changes in subsequent individual risk behavior. Child-parent relationship quality did not explain changes beyond itself. Individual risk behavior and child-parent relationship quality were stable across the stages of development, whereas peer risk behavior was less stable. Those structural relations that were tested in multi-group longitudinal panel models showed that the latent regression pathways were invariant across gender groups. These findings provide limited support for the stage-environment fit theory (Eccles et al., 1993) and the group socialization theory (Harris, 1995).