Study 1
Weisburd and Green (1995) identified 56 “hot spots” of drug activity in Jersey City, N.J., using computer mapping techniques with narcotics-related arrest and emergency calls data. Randomization into “statistical blocks” was used to assign treatment and control conditions. The randomization by blocks was used to account for the significant variation in the quantity and nature of the drug trade at the different hot spots. Randomization took place within “blocks,” which are pairs or groups of similar hot spots, to ensure that comparison between intervention and control sites was viable. By analyzing the distribution of arrest and call activity of the 56 hot spots, the authors classified the sample into four categories of drug and call activity: very high (n = 10), high (n = 8), medium (n = 26), and low activity (n = 12). After block randomization, groups were equivalent by number of narcotics arrests and calls, mean age of narcotics suspects arrested, type of hot spot, proportion of African American residents, and proportion of minors living within the hot spot boundaries. The groups differed, however, on numerous measureable variables. For example, there were 17 cocaine hot spots in the treatment group but only 12 in the control group, and there were 7 West District hot spots in the treatment group and 14 in the control group.
Before the intervention, Jersey City operated a narcotics unit consisting of three teams of officers each with two squads. The squads were randomized to treatment and control conditions (one for each team), thus minimally affecting shift rosters. The two groups of officers were separated physically to avoid contamination through mimicking behavior, and were monitored daily through several sources. The officers assigned to the treatment condition implemented the Drug Market Analysis Program, while the officers assigned to the control hot spots maintained activities in line with the tactics used during the years leading up to the intervention. While this control condition maintained previous enforcement levels (arrest-based, unsystematic enforcement) officers in the three control squads were expected to concentrate on the activities of the 28 control hot spots specifically. Though the treatment strategy was supposed to last 12 months, this was extended to 15 months, because by the 9th month implementation in a majority of hot spots was shown to be progressing too slowly. Individual hot spot implementation plans were drawn up with the narcotics squad to ensure that the treatment strategy was extended to all target sites.
The effectiveness of the intervention was assessed by comparing the emergency calls for service of treatment and control hot spots in the 7 months before and the 7 months following the intervention, using two-tailed ANOVA tests. Outcomes of interest included emergency calls for disorder-related offenses (such as suspicious persons or nuisance), violent offenses, and property offenses. Emergency calls for narcotics-related offenses were also examined; however, due to high skew in the data, the statistical tests were unstable and statistical significance could not be established. Therefore, the results for narcotics-related offenses is not reported here. The study authors conducted an additional analysis to examine the diffusion effects on surrounding areas.