NCJ Number
206770
Journal
British Journal of Criminology Volume: 44 Issue: 3 Dated: May 2004 Pages: 448-467
Date Published
May 2004
Length
20 pages
Annotation
This paper counters P.R. Marchant's criticisms of the methodology used in the British Home Office Research Study 251 (HORS 251), a systematic review and meta-analysis of the effects of improved street lighting on crime.
Abstract
Marchant's criticism of HORS 251 focused on the odds ratio measure of effect size and the estimate of variance in comparing crime rates in an area before and after the installation of improved lighting with a control area where no changes in lighting were made. Marchant argued that the true confidence interval should have been much wider than that used in HORS 251, which concluded that improved lighting did reduce crime. According to Marchant, the wider confidence interval would have made it impossible to determine whether or not lighting may reduce or even increase crime. The authors of the current article note that in the 13 evaluations of the effects of improved street lighting on crime that were included in the HORS 251 review, they empirically estimated the ratio of the variance of the number of crimes to the mean number of crimes, based on before-and-after data on crime in control areas. They indicate that in focusing on overdispersion, Marchant is arguing that the variance of natural logarithm of the odds ratio (LOR) is underestimated by the common formula used in HORS 251; the authors argue, however, that the fact that "before" (lighting improvement) crimes were positively related to "after" crimes means that this variance is overestimated, because paired data are being analyzed as though they were unpaired. Generally, the "before" crimes and the "after" crimes will be committed by the same offenders against the same victims. Further, although Marchant criticizes HORS 251 for excluding small studies, the authors argue that it seemed reasonable to them to analyze only those studies that were methodologically sound. Other criticisms by Marchant are also answered in detail. The authors do acknowledge the need for further research on this issue with better measures of effect size and more high-quality evaluations of the effects of area-based crime-reduction programs; however, they contend that three of the evaluations reviewed in HORS 251 did provide evidence that improved lighting caused a decrease in crime. 4 tables and 31 references