Comment on Estimating the True Rate of Repeat Victimization from Police-Recorded Crime Data

Abstract

The Recorded Repeats Adjustment Calculator (RRAC) methodology introduced by Frank, Brantingham, and Farrell (2012) doesn’t consider the entire probability mass function associated with the binomial distribution. Although the authors recognized and documented this as a limitation, the implications are significant because some of their key findings are not internally consistent with their explicit underlying assumptions or the available empirical data. This article proposes revised estimates, based on a Bayesian treatment of the empirical data and underlying assumptions stated by the authors themselves. According to the initial RRAC estimates they reported, repeat burglary victims would have represented 47.1% of all burglaries or 22.0% of all burglary victims. The proposed revisions suggest instead that repeat burglary victims represent approximately 30.8% of all burglaries or 15.9% of all burglary victims. By comparison, based strictly on the police-recorded data, repeat victims account for 19.8% of all reported burglaries or 10.0% of all burglary victims. While these findings tend to temper the empirical findings and original conclusions of Frank, Brantingham, and Farrell (2012), they provide renewed support for their insight that raw police-recorded crime data are likely to underestimate the true rate of repeat victimization.