Meta-Analysis of Longitudinal Studies Comparing Risk Factors for Offending Measured Using Self-Reported versus Official Records

• In Event: Evidence-Based Crime Policy III

Abstract

One of the central issues in criminology is the measurement of our key outcome variable of offending. Two of the most popular measures are official records, derived from police contacts, arrests, or convictions, and self-reported offending. Each of these data sources has unique strengths, weaknesses, and biases that impact the validity and reliability of the measure and of the conclusions. While only a few longitudinal datasets include both official and self-reported offending measures, several scholars have used this data to examine the extent to which risk factors are similar for self-reported versus official records of offending. This paper presents the first systematic review and meta-analysis of research that examines the key predictors of criminal behavior when measured using official records and self-reported offending. Using data from over 50 studies, based on 15 major longitudinal datasets in criminology, we will evaluate the strongest predictors of offending in both self-reported and official records, as well as the level of agreement (or disagreement) in results when using similar predictors but different outcome measures of criminal activity. Implications for criminal career research, as well as potential biases in self-reports, arrests, convictions, and police contacts, will also be discussed.

Authors

• Bryanna Hahn Fox, University of South Florida

• David Farrington, Cambridge University