Study 1
Piza and colleagues (2020) conducted a quasi-experimental evaluation with a micro-synthetic control group to evaluate the effect of a newly established Newark (N.J.) Police Department (NPD) substation on aggravated assault, burglary, murder, theft from auto, motor vehicle theft, and robbery at 6 years postintervention. The substation was located within the downtown business improvement district (BID), operated by the Newark Downtown District.
The analysis estimated effects over a 6-year postintervention time frame with two distinct time periods: 1) the initial 3-year program period (Sept. 1, 2012, when the substation opened, through Aug. 31, 2015) and 2) the period subsequent to the expiration of the memorandum of understanding between the NPD and Newark Downtown District that established the roles and responsibilities of each entity (Sept. 1, 2015, through July 23, 2018). The second 3-year period was assessed to determine whether any changes in the nature of the substation operations between the two periods resulted in variable outcomes. The second 3-year period was slightly truncated, owing to the expansion of the substation target area on July 24, 2018. Restricting the postintervention period to July 23, 2018, ensured that only the effect of the original substation and related crime control efforts were measured. The CrimeSolutions review of this study focused on results at the end of the entire 6-year postintervention period.
Both street segments and street intersections were the units of analysis. Street segments represented “the two block faces on both sides of a street between two intersections” (Weisburd et al. 2004, 290). Street intersections were created through computer geoprocessing techniques, which generated points at every location in Newark where two or more streets intersected. Jointly, street segments and street intersections were considered street units. This process resulted in 4,081 street intersections and 6,166 street segments (n = 10,247 street units) in Newark.
A micro-synthetic control method was used to identify a control area that mimicked pre-intervention crime trends and that was similar in demographics to the treated area. This procedure incorporated the following nine covariates to match control street units to the treated street units: 1) number of overall crime incidents experienced during the 3-year pre-intervention period; 2) Newark’s urban enterprise zone (which mirrors a BID’s focus on urban revitalization without the supervision of a nonprofit organization and revenue raised from self-imposed taxation), to compare street units from the treated area—predominately comprising commercial properties—with street units from similarly constituted areas of Newark; 3) proportion of pre-intervention, high-crime street units in the surrounding Census block group; 4) street unit type: street segment (coded as “1”) or as an intersection (coded as “0”); 5) principal roadways (as more-frequently used streets may present more crime opportunities than less-busy streets, each street segment was coded either as a principal roadway [coded as “1”] or as part of another roadway classification [coded as “0”]); 6) length of street unit (as longer street segments may offer more crime opportunities relative to shorter street segments); 7) number of enforcement actions conducted during the 3-year pre-intervention period (pre-enforcement actions were included to reflect the level of law enforcement activity within each street unit during the 3-year pre-intervention period); 8) percentage of new construction in the Census block group; and 9) percentage of units occupied (nonvacant) in the Census block group.
This process resulted in a matched analysis of 248 street units in the substation treatment area, and 248 street units in the weighted control area. There were no statistically significant differences between the treated area and control area on crime trends or demographics. The synthetic control procedure was able to generate exact matches for all of the time-invariant characteristics, and for the time-varying crime control characteristics.
The Newark Police Department provided crime data. The micro-synthetic crime analysis provided measures of crime changes (relative to controls) in both the treated area and the surrounding catchment area following the opening of the substation. Linear difference-in-difference estimates of aggregated crime counts in the treated area compared with the control area were conducted for individual years and for cumulative subtotals of 3-year subsets to assess the effect of the substation on property and violent crime outcomes over the entire 6-year postintervention period. Subgroup analysis was conducted to assess whether spatial displacement or diffusion of benefits occurred following the opening of the substation.