How valid are self-reports of delinquent and criminal behavior? A meta-analysis comparing self-report and official record measures

Despite decades of research on the psychometric properties of self-reports of offending behavior, there remains no consensus on whether this method measures what it intends to measure—namely, delinquent and criminal behavior—or on how robust the method is to variations in sample and measurement characteristics. We conducted a systematic review with meta-analysis to estimate the convergent validity of self-reports relative to official records, while also examining its differential validity across sex, age, race/ethnicity, recall period, and the source of official records. Following PRISMA guidelines, a comprehensive literature search was conducted across ProQuest, Web of Science, Scopus, and PubMed. Studies were included if they evaluated the association between self-reported delinquent behavior and official records. Three-level meta-analyses were conducted to synthesize findings and evaluate moderators. The meta-analytic correlation between self-reports of delinquent acts and official records was 0.192 (95% CI [0.136, 0.247]; 87 effect sizes, 16 samples, N = 11,888), compared to 0.464 for self-reports of official action (95% CI [0.377, 0.544]; 51 effect sizes, 21 samples, N = 35,024). Together, these findings indicate reasonable convergent validity. However, convergent validity of self-reports of delinquent acts varied by recall period and the source of official records, and convergent validity of both self-report measures was contingent on race and ethnicity, raising concerns about generalizability. We hope that the present study helps to reorient scholarly attention toward the systematic evaluation of the psychometric properties of self-reports of delinquent acts.