Measuring recurrent victimization: Evaluating operationalization strategies and predictors using the Crime Survey for England and Wales.

Abstract

Victimization is concentrated among a small group of individuals, commonly referred to as recurrent victims. However, there is no consensus on the operationalization of recurrent victimization. This study investigates optimal measurement strategies and identifies predictors of recurrent victimization through a meta-analytic synthesis of multiple analytic approaches estimated on the 2019/20 Crime Survey for England and Wales. The results suggest that defining recurrent victimization using a Top 10% binary categorization and estimating logistic regression models can lead to biased conclusions. In contrast, operationalizations based on experiencing two or more victimization types or incidents performed substantially better when paired with bivariate probit models. Count-based operationalizations, particularly total victimization counts across crime types, also performed well when analysed using negative binomial or zero-inflated negative binomial models. Taken together, the findings indicate that researchers wishing to categorise recurrent victims should employ theoretically informed category- or incident-based measures analysed with bivariate probit models, whereas those seeking to identify individuals who experience higher volumes of victimization should use count-based measures estimated with negative binomial frameworks. Across all approaches, mental health conditions consistently emerged as the strongest correlate of recurrent victimization.