School of Information Management, Leeds Metropolitan University, Email: p.marchant@leedsmet.ac.uk

Abstract

This article discusses the important differences between evaluating area-based crime reduction interventions through comparison of areas, and evaluating treatment effectiveness by a classic randomised controlled trial (RCT). The differences must be respected in the analysis of data. Wrong results will be given for a comparative area- based crime reduction trial if it is thought of as an RCT. The problem with evaluations which compare areas, one with an intervention and one without, is that such studies are neither randomised nor controlled, and the crime events being counted are correlated rather than statistically independent. Wrongly using those methods which are appropriate for an RCT will give misleading results. Statistical uncertainties will tend to be large and poorly determined, and statistical biases problematic because of the lack of equivalence between areas at the outset. Crime studies need to be appropriately designed to avoid such problems if costly mistakes are to be avoided.