Predicting Cannabis Use across Waves: Marginal Calibration with Additional Longitudinal Data

• In Event: New Advances in the Prediction of Crime and Recidivism

Abstract

This study seeks to compare models in terms of predictive validity for cannabis use among Reservation-dwelling American Indian adolescents based on identified risk and protective factors, including availability, perceived harm, various peer influence measures, school and familial factors, self-esteem, depression, risky behavior, ethnic identity and perceived discrimination. Specifically, we compare the ability of models to predict cannabis use at successively later stages in the future compared to models which incorporate additional longitudinal information prior to the prediction wave. We develop and test models using longitudinal data on 280 adolescents from two reservation-based schools in the United States who were surveyed three times six months apart.

Authors

• Kimberly L. Henry, Colorado State University

• Thomas A. Loughran, Pennsylvania State University

• Linda R. Stanley, Colorado State University

• Randall C. Swaim, Colorado State University