Prevalence and Measurement of the Victim-Offender Overlap: A Systematic Review and Meta-Analysis

Research indicates that individuals may experience both victimisation and offending across the life course, known as the victim-offender overlap. We conducted a systematic review with meta-analytic synthesis to examine the prevalence and strength of the victim-offender overlap and assess whether reported estimates vary according to methodological choices. Articles published between 2012 and 2023 were identified through ProQuest, PsycINFO, Embase, and Criminal Justice Abstracts. Studies were included if they reported numerical overlap between direct victimisation (e.g., violent, sexual, and property) and offending (e.g., violent, property, and drug-related) behaviours, operationalised via self-, peer-, teacher-, or official records; gender and offence/victimisation subtypes were extracted where reported. Studies focusing solely on cybercrime, bullying, or indirect victimisation were excluded. Risk of bias was assessed using the Joanna Briggs Institute Checklist for Prevalence Studies. Sixty-eight studies including 549,377 analytic participants were included. Three random-effects meta-analyses synthesised prevalence (population-level, offender-conditional, and victim-conditional; k = 29–32) and two synthesised association strength (concurrent and lagged; k = 8–14). Population-level prevalence was 11.9% (95% CI [8.2%, 16.9%]; 95% PI [1.4%, 56.0%]), rising to 57% and 50% among offender and victim subsamples, respectively. Concurrent associations were moderate to large ( d = 0.87) and significantly stronger in justice-involved samples (Δ d = 0.66, p = .002); lagged associations were weaker ( d = 0.42) with no statistically significant directional asymmetry. Heterogeneity was extreme across syntheses ( I 2 = 98.3–99.8%), and was consistent with substantial variation in data sources, operationalisation, and sample characteristics. Findings underscore the need for clearer operational definitions, transparent reporting of analytic decisions, and multi-source longitudinal research designs.