Study 1
To assess the impact of hot spots policing in Louisville (Ky.) on crime and disorder, Schaefer and colleagues (2019) used block randomization to ensure an equal balance of urban and suburban hot spots across the treatment and control groups. The experiment was designed to examine the effectiveness of directed patrol as a crime-reduction strategy in both urban and suburban locations in Louisville.
When the experiment was implemented, the Louisville Metropolitan Police Department (LMPD) oversaw eight geographically based patrol divisions. Out of the eight patrol divisions, seven participated in the experiment. To identify hot spots, crime data and crime-related calls for service were collected for each of the seven participating divisions for a 2-year period prior to the implementation of the experiment (January 2012–December 2013). Data from this period were geocoded to street segments and categorized as part 1 crimes, part 2 crimes, and crime-related calls for service. Part 1 crimes correspond to those included in Part 1 of the Uniform Crime Report (UCR) data; these are often referred to as “index” crimes. Part 2 crimes are less serious and comprise other forms of criminal activity; the present study refers to these as “soft” crimes. Once all the crime data were geocoded and categorized, 80 street segments with the greatest volume of crime and crime-related calls for service were selected from each division for possible inclusion in the study.
After identifying an initial pool of potential street segments, the study authors used Sherman and Weisburd’s (1995) inclusion criteria to determine eligible hot spots. To be eligible for inclusion in the experiment:
- No hot spot [could be] larger than one standard linear street block.
- No hot spot [could] extend for more than one-half block form either side of an intersection.
- No hot spot [could be] within one standard linear block of another hot spot.
After the removal of ineligible hot spots, each division commander selected 10–24 hot spots from their division for inclusion in the experiment. These selected hot spots were then randomly assigned to either the treatment or control group. In total, across the seven patrol divisions, 94 hot spots were selected, including 47 in the treatment group and 47 in the control group.
To determine whether the hot spot was in an urban or suburban location, the study authors relied on classifications established by the National Center for Education Statistics (NCES) Education Demographic and Geographic Estimates (EDGE) program. This process resulted in 24 urban hot spots and 23 suburban hot spots in each group. To ensure that the treatment and control hot spot pairs were equivalent on the three outcome variables of interest (part 1 crimes, part 2 crimes, and crime-related calls for service), an independent samples t-test was conducted using crime data from January 2012–December 2013. The results revealed that there were no statistically significant differences between the treatment and control groups on any of the outcome measures at baseline.
The experiment lasted 90 days, beginning on June 2, 2014, and ending on August 30, 2014. To determine the impact of the hot spots strategy, the study author used ordinary least squares (OLS) regression models with a difference-in-differences (DID) approach. DID measures were used to estimate the effects of the intervention on the three outcomes between the treatment and control conditions during the 90-day intervention period.