Cascading peer dynamics underlying the progression from problem behavior to violence in early to late adolescence

Thomas J. Dishion, Marie Hélène Véronneau, Michael W. Myers

Research output: Contribution to journalArticle

90 Scopus citations

Abstract

This study examined the peer dynamics linking early adolescent problem behavior, school marginalization, and low academic performance to multiple indices of late adolescent violence (arrests, parent report, and youth report) in an ethnically diverse sample of 998 males and females. A cascade model was proposed in which early adolescent risk factors assessed at age 11 to 12 predict gang involvement at age 13 to 14, which in turn, predicts deviancy training with friends at age 16 to 17, which then predicts violence by age 18 to 19. Each construct in the model was assessed with multiple measures and methods. Structural equation modeling revealed that the cascade model fit the data well, with problem behavior, school marginalization, and low academic performance significantly predicting gang involvement 2 years later. Gang involvement, in turn, predicted deviancy training with a friend, which predicted violence. The best fitting model included an indirect and direct path between early adolescent gang involvement and later violence. These findings suggest the need to carefully consider peer clustering into gangs in efforts to prevent individual and aggregate levels of violence, especially in youths who may be disengaged, marginalized, or academically unsuccessful in the public school context.

Original languageEnglish (US)
Pages (from-to)603-619
Number of pages17
JournalDevelopment and psychopathology
Volume22
Issue number3
DOIs
StatePublished - Aug 2010

ASJC Scopus subject areas

  • Developmental and Educational Psychology
  • Psychiatry and Mental health

Fingerprint Dive into the research topics of 'Cascading peer dynamics underlying the progression from problem behavior to violence in early to late adolescence'. Together they form a unique fingerprint.

  • Cite this