Similarities and Differences in Risk and Protection Across Ethnic Groups: Tests of Invariance in Predicting Problem Behaviors in Adolescence

Introduction: Risk and protective factors across individual, school, family and community domains are important for prevention scientists seeking to predict a range of adolescent problem behaviors including delinquency, conduct problems and school suspension.  Previous studies have found that the measurement of risk and protective factors is consistent across ethnicities and gender (Glaser et al. 2005).  Building on this research we investigated differences in the relationship of risk and protective factors to delinquency, conduct problems and school suspension during middle and high school across three ethnic groups.

Methods: This study employed cross-sectional data (N=3777) gathered with the Evidence2Success Youth Experience Survey on multiple risk and protective factors from a mid-sized Northeastern city.  Equivalence of the covariance structure of risk and protective factors with problem behaviors was tested across Latinos, African Americans and Caucasians using multiple group modeling.  Risk and protective factors were modeled separately during middle school and high school by testing two-group comparisons (Latino and Caucasian, Latino and African American, Caucasian and African American).  Twelve models were fitted with equality constraints across ethnic groups placed on relationships of risk and protective factors with delinquency, conduct problems and school suspension.

Results: Results indicated that risk and protective factors operated similarly in their relationship to conduct problems, delinquency and school suspension during both middle and high school, and for protective factors during high school.  Tests of goodness of fit for nine out of twelve models signaled invariance across ethnic groups.  Results of three high school models of risk factors correlated with delinquency, conduct problems and school suspension, respectively, during high school indicated some cross-ethnic differences (partial invariance).  These differences were primarily attributed to differential magnitude of relationships between reported number of close friends involved in gangs and self-reported delinquency.  African Americans showed the strongest relationships between friends in gangs and delinquency followed by Latinos and Caucasians respectively.

Conclusions: Results of this study increase confidence for prevention scientists seeking to assess risk and protective factors with surveys among diverse communities.  As research has shown that measurement of risk and protection is consistent across ethnic groups, the current study demonstrates that there are also similar relationships of risk and protective factors with three important problem behaviors across ethnic groups during both middle and high school.  Future research should address the generalizability of these relationships to nationally representative samples.