Violence Reduction in Joliet, Illinois: An Evaluation of the Strategic Tactical Deployment Program

48 

 

observed for two groups for two time periods. One of the groups is exposed to a treatment in the 

second period but not in the first period. The second group is not exposed to the treatment during 

either period. In the case where the same units within a group are observed in each time period, 

the average gain in the second (control) group is subtracted from the average gain in the first 

(treatment) group. This removes biases in second period comparisons between the treatment and 

control group that could result from permanent differences between those groups, as well as 

biases from comparisons over time in the treatment group that could be the result of trends.  

The outcome Yi is modeled by the following equation: 

Yi = α + βTi + γti + δ (Ti · ti) + εi 

where the coefficients given by the Greek letters α, β, γ, δ , are all unknown parameters and εi is 

a random, unobserved "error" term which contains all determinants of Yi which our model omits. 

The equation coefficients have the following interpretation: 

  α = constant term 

  β = treatment group specific effect (to account for average permanent differences between    

        treatment and control) 

  γ = time trend common to control and treatment groups 

  δ = true effect of treatment 

In the current evaluation, we used a DiD panel regression design to assess whether the rate 

of change in shots fired in the STD target areas (Sectors 11, 16, 22) was significantly different 

than the non-target sectors between pre-intervention and post-intervention periods. Similar to the 

models presented above, the DiD regression model of shots fired counts was estimated by 

general linear modeling using the Poisson distribution. The analysis included a dummy variable 

for STD vs. non-STD sectors and a dummy variable for pre and post intervention time periods.  

---

49 

 

The model also included controls for percent minority population, unemployment rates and 

average monthly drug arrests.    

Table 8 contains the post-intervention by target area interaction estimate (δ = true effect of 

treatment) for 7.75 years (January 2005 – September 2012) of observations (n = 3,134). The 

interaction estimate (δ = -.004, p = .910) indicates that there was no significant difference in the 

relative change between pre-intervention and post-intervention periods in confirmed shots fired 

between targeted and non-targeted sectors.   

Table 8 

DiD Poisson regression Results of Confirmed Shots Fired 

Variable Coefficient 

SE 

IRR 

p-value 

Targeted Sectors 

-.004 .035  .996  .910 

Intercept 

4.08 .001 ---  .000 

 

Table 9 contains the post-intervention by target area interaction estimate of 1, 242 robbery 

observations. The interaction estimate (δ = -.013, p = .852) indicates that there was no significant 

difference in the relative change between pre-intervention and post-intervention periods in 

robberies between targeted and non-targeted sectors.   

Table 9 

DiD Poisson regression Results of Robberies 

Variable Coefficient 

SE 

IRR 

p-value 

Targeted  

Sectors 

-.013 .071  .987  .852 

Intercept 

4.06 .021 ---  .000 

 

---

50 

 

Conclusions 

A number of studies have examined the application of problem oriented policing to hot 

spots (Baker & Wolfer 

2003

; Braga et al. 

1999

; Braga & Bond 

2008

; Mazerolle et al. 

2000; 

Weisburd & Green 

1995

), and most of these have shown that the efforts reduced some forms of 

crime and disorder. Results, however, have been more mixed with respect to violent crime.  

Studies by 

Cohen and Ludwig (2003) and McGarrell, et al. (2001) found that directed police 

patrol aimed at violent-crime hotspots help to reduce gun-related crime. On the other hand, 

interventions studied by Mazerolle et al. (

2000

), Sherman, Gartin and Buerger (

1989)

 and 

Weisburd and Green (

1995

) did not reduce violence. Additionally, Sherman and Weisburd 

(1995) and Taylor et al. (2011) found that the deterrent effects of police patrol in crime hot spots 

for violent crimes were generally non-significant, though in the expected direction. Braga’s 

(2007

) meta-analysis of results from five randomized experiments suggests that the effects of 

hot-spot policing are most pronounced on disorderly behaviors; although violent and property 

crimes declined on average across the studies, these effects were not statistically significant 

overall.  The lack of significant effects could reflect the impulsive, expressive nature of many 

violent crimes (which may make them harder to prevent) and the rarity of violent crime in very 

small locations. Alternatively, the particularly high concentration of violence in a relatively small 

number of places would seem to weigh in favor of using hot spots strategies to curb violence.

 

With regard to the current evaluation, the results are consistent with those found in the 

studies cited above (

Braga, 

2007; 

 Mazerolle et al., 

2000; 

Sherman, Gartin & Buerger, 

1989; 

Sherman & Weisburd, 1995; Taylor, et al., 2011; Weisburd & Green, 

1995)

; that there were no 

statistically significant effects of the intervention on violent crime (i.e., shots fired and 

robberies).

 

---