Toward a Causal Estimate of Friends’ Influence on Delinquency

• In Event: Peer Influence: Effects and Developmental Processes

Abstract

Objective: A fundamental question in criminology is whether individuals are influenced by their peers’ delinquency. The literature has primarily focused on two peer groups, friends and school peers. While friends are theoretically more influential for delinquent outcomes, causal identification is difficult in the presence of friendship selection. Studies of school peers exploit random variations in peer assignment for causal identification, but the theoretical importance of school peers is weaker. The resulting tension is between peers that are theoretically more influential, friends, and peers that are methodologically more tractable, school peers.

Data/Methods: To resolve this tension, I utilize an instrumental variable approach to estimate a causal effect of friends’ delinquency. Specifically, friends’ health and tiredness are exploited as instruments for friends’ delinquency. These models are estimated on a sample of 9,492 adolescents from the National Longitudinal Study of Adolescent to Adult Health.

Results: This approach demonstrates a significant positive causal effect (ES:0.104, P-Value:0.007) of friends’ delinquency on individual delinquency, which is robust to several alternative specifications.

Conclusions: This research moves the literature toward peer effect estimates that are both causally identified and theoretically supported, which has substantial implications for both theories and policies regarding the spread of delinquent behaviors among youth.

Author

• Jason Robey, University of Wisconsin-Madison