Frequency Versus Duration of Police Patrol Visits for Reducing Crime in Hot Spots: Non-Experimental Findings from the Sacramento Hot Spots Experiment

Abstract

Was the effect of police patrols reducing crime and disorder in the Sacramento Hot Spots Experiment (SHSE) more strongly associated with the average frequency of patrol visits or the total duration of all visits combined? Two independent measures of patrol were used to analyze initial and new data on the 42 hot spots that Telep et al. (2014) studied in 21 matched pairs for the SHSE: Computer-Assisted Dispatch (CAD) records on the number of and duration of patrol visits to each hot spot (initial data), and Automatic Vehicle Locator (AVL) (new) data on total patrol time within the “geo-fenced” area of each hot spot. The patrol measures were linked to two outcome measures at each hot spot. Part I crime reports and citizen-generated Calls for Service (CFS) records for events occurring within the hot spot street address boundaries. All SHSE pairs of hot spots were examined as a meta-analysis of 21 experiments for differences in frequency and total duration of patrols. Then all 42 hot spots were examined for correlations between outcome measures and CAD data on frequency and duration, plus AVL data on duration [AVL data could not reliably measure frequency of discrete patrol visits]. The SHSE created large mean differences between experimental and control conditions in both frequency and total duration of patrols, which were highly correlated (r = .9). Correlations of CAD measures of before-during increases in frequency, but not duration, of patrols across all 42 hot spots predicted somewhat fewer event counts of both outcome measures: crime reports (r = −.273, one-tailed p = .04) and CFS (r = −.234, one-tailed p = .068). In contrast, AVL data on before-during increases in duration showed somewhat less crime with more duration of total patrol time (r = −.231, one-tailed p = .07), but no effect of more duration on CFS. Based on the analysis of CAD data, police may be able to reduce crime and disorder more effectively by making more frequent patrols in each hot spot, while reducing the total time of all patrols. Contrary to the policy conclusion from the Koper (1995) Curve, the CAD findings may suggest there is greater benefit in having more visits of under 15 min rather than having fewer visits that all last from 10 to 15 min each. Yet, the evidence from this study is mixed. The CAD finding is contradicted by this study’s AVL data analysis, which show that greater increases in total patrol duration did predict somewhat fewer crimes (but not CFS). Since both the CAD and AVL findings are correlational rather than based on random assignment of frequency and duration, field experiments are now needed to provide more definitive answers to the research question.