Impact of a Crime Prevention Ordinance for Small Retail Establishments

Abstract

Objective:

We evaluated the impact of an ordinance requiring crime prevention measures on crime rates.

Methods:

Crime reports for robbery and aggravated assault from January 2006 through December 2015 were linked to randomly selected convenience stores and small retail grocers in Houston (n=293). Store characteristics and compliance with a list of safety measures were collected by surveyors in 2011. Generalized estimating equations were used to compare rates of crime before and after the implementation of the ordinance.

Results:

Robberies decreased significantly after the ordinance went into effect (RR=0.38; 0.29,0.51). No individual safety measure was associated with decreased robbery rates. No similar decrease was observed for aggravated assault.

Conclusions:

City ordinances mandating crime prevention measures can be effective. We could not parse out the effectiveness of individual elements, suggesting a comprehensive approach may be more effective.

INTRODUCTION

Workplace violence remains a leading cause of injury and death for workers in the United States.¹ The combination of working with the public late at night or early morning and with cash can put workers in businesses like convenience stores and liquor stores at high risk for robbery-related injuries.^{2 3} Homicides are the leading cause of worker deaths in the US retail trade sector. In 2016 there were roughly double the number of worker deaths from homicide (n=120) than the number of worker deaths related to transportation (n=65).⁴ In 2016, every one of the 29 convenience store worker fatalities was a homicide, with homicides accounting for almost all of convenience store worker fatalities for years 2012 through 2015.^4–8

Occupational injury research efforts for convenience stores have focused on designing and evaluating interventions for reducing well-known risks for robbery-related crime.^9–14 This research is based on Crime Prevention Through Environmental Design (CPTED), a theory that changes to the business environment can deter robbery by making the business an unappealing target.⁹ A set of safety measures identified as most effective in reducing robbery-related violence in convenience stores is the recommended foundation for promising interventions: cash control policies including use of a drop safe—a place to deposit cash and reduce amount of cash in register, good lighting, employee training and good visibility.^{13 15} However, the implementation of these strategies by convenience stores has not been widely successful with few stores adopting all of the recommended strategies.¹⁵ In recent years, two separate approaches have been employed to effectively reach convenience store owners and management – voluntary robbery and violence prevention programs and legislation mandating the implementation of city or statewide safety measures. The goal of both approaches is to increase compliance with effective safety measures either through targeted and individualized consultation (voluntary robbery and violence prevention programs) or a legally enforceable route requiring every store within a jurisdiction to adopt all specified safety measures (city ordinances/state laws). Previous research has shown that robbery and violence prevention programs increase compliance with crime prevention safety measures, and violent crime and robberies decreased in stores with high compliance.¹⁶

We evaluated how compliance with the individual requirements of an ordinance in Houston impacted crime rates after the ordinance was enacted in 2010. In a previous study, full compliance with the ordinance was determined to be low in the year following enactment.¹⁷ To evaluate the effectiveness of the Houston city ordinance in reducing crime and related violence among convenience store personnel, we compared robbery and aggravated assault rates before and after implementation of the ordinance.

METHODS

Study Design

In March 2008, Houston passed a city ordinance effective January 1, 2010, with the primary objective of reducing robbery-related crime, preventing its escalation, and increasing the successful prosecution of crime that is more prevalent in convenience stores. The ordinance required the following: ¹⁸

1. Registering each store with the city police department and affixing the registration decal to the main ingress

2. Viewing the violence de-escalation training video produced by the police department, available in six languages and accessible via a commonly used app for online viewing.

3. Authorizing the police department to enforce trespassing laws on their behalf (trespass affidavit) and posting “No Trespassing” signs

4. Posting security signs, height markers (for identification of perpetrator height), and indication of an alarm system

5. A drop safe with a cash accessibility policy for how cash is handled and how much cash is available on hand

6. Surveillance camera system

7. Unobstructed visibility of and from the cash register and sales area.

To facilitate compliance, the Houston Police Department (HPD) publicized the new ordinance through industry trade associations, and police officers personally visited all of the eligible convenience stores to deliver informational packets that included the registration sticker, training CDs, trespass affidavit, and a list of vendors with cost information for the drop safe, surveillance cameras and alarm systems.

A commercial list of Houston convenience stores (including gas stations) and small retail grocers (<10,000 sq ft) (n=1,729) provided the sampling frame for a randomized selection. In 2011, store managers were interviewed in person by trained study surveyors to document store type (corporate, franchise, single owner), knowledge and implementation of store safety ordinance requirements, potential costs associated with compliance with the safety ordinance requirements, store location characteristics, indicators of crime activity, and demographic characteristics of management and employees. Corporate stores are under the management of the larger corporation. Franchise stores purchase the right to use the parent company’s name and products but can have varying management structures. Corporate stores must follow corporate policies and procedures, whereas franchise stores can use the corporate policies and procedures but aren’t necessarily bound by them. Current implementation of ordinance requirements, as well as the manager’s recall of implementation of any ordinance requirements occurring before the ordinance, was collected. After interviewing the store managers, trained surveyors completed a store safety checklist to ascertain whether or not the ordinance requirements visible to a visitor, such as registration sticker, signage, clerk visibility, alarms, and surveillance camera systems, were implemented.

The HPD maintains a database of crimes with a designation made for crimes occurring on the property of convenience stores and small retail grocers. All crimes occurring at convenience stores and small retail grocers January 2006 through December 2015 were provided. Due to the heterogeneity of the crimes, only robbery and aggravated assault were analyzed pre ordinance (2006–2009) and post ordinance (2010–2015). Crime data were linked by address to convenience store and small retail grocer survey data that was collected in 2011. The NIOSH Institutional Review Board approved the study protocol. Surveyors representing the demographic makeup of the store managers and clerks administered the questionnaires and conducted the store safety checklists.

Statistical Analyses

Differences in crime rates were assessed by store characteristics and adherence to ordinance requirements using negative-binomial regression. The outcomes of robbery and aggravated assault were analyzed separately with two negative-binomial models. Store characteristics evaluated were the following: type of store (corporate, franchise, single owner), perceived background crime, open 24 hours, located in a strip mall, located within a half a mile of a highway, store rurality, and store appearance. A store was identified as being in an area of perceived background crime if the store manager indicated that the store was within one mile of known drug trafficking activity; loitering youth, gangs, panhandlers or prostitutes; or areas with broken windows, graffiti, or abandoned cars on empty lots.

The following individual elements from the manager questionnaire of the ordinance were evaluated with the two negative binomial models for robbery and aggravated assault: store provided employee training in robbery prevention, signage, visibility of the cash/counter/sales area, presence of a cash drop box and cash accessibility policy, and the use of an alarm system and camera surveillance systems. The required signs indicated the following: no loitering or trespassing, surveillance cameras are in use, minimum cash is on hand, employee cannot open drop safe, and security alarm system in use. Thresholds of compliance were determined based on tertiles of the number of robbery prevention measures implemented by the stores (0–3, 4, and 5–6).

The impact of robbery prevention measures can be confounded by environmental and store-level characteristics. To account for this, an adjusted negative-binomial model was built for each of the crime outcomes by including all store level characteristics that were found to be statistically significant in the univariate models. The evaluated characteristics included the following: location in an area of background crime (Yes, No), store location (Urban, Rural, Suburban), being open 24 hours (Yes, No), store type (Corporate, Franchise, Single Owner), and store appearance (New, Old – Well Kept, Old). Collinearity was checked using condition indices by comparing the square root of the eigenvalues of the individual covariates. A limit of 10 was used for the collinearity indices as an indicator that the collinearity was impacting the regression estimates.¹⁹ Univariate and adjusted negative-binomial models and 95% confidence intervals were calculated using SAS PROC GENMOD.

To assess the overall impact of the ordinance on crime rates, generalized estimating equations with a negative-binomial distribution were used to compare rates of crime before (2006–2009) and after (2010–2015) the implementation of the ordinance for the individual stores. ²⁰ Rate ratios and 95% confidence intervals were calculated using SAS PROC GLIMMIX stratifying on the individual store to account for correlated crime rates of an individual business measured over the two time periods. All analyses were performed using SAS 9.4 (SAS Institute, Inc., Cary, NC, USA).

RESULTS

Three-hundred and fifty-eight convenience and small grocery stores in Houston completed the manager survey about adherence to the robbery prevention ordinance, and site evaluation was completed for 324 stores by a trained surveyor. Of the stores with completed manager surveys and site evaluations, 293 convenience stores (81.8%) reported being in business before the January 2010 ordinance became effective were included in this study. During the six-year period after the Houston ordinance went into effect (January 1, 2010 to December 31, 2015) 2,232 total crimes were identified in the HPD crime data at the sites of the 293 convenience and small grocery stores included in this study. There were 398 robberies and 293 aggravated assaults.

Characteristics of the convenience and small grocery stores along with corresponding crime rate ratios are presented in Table 1. In the univariate analysis of crime rates ratios, businesses in areas of perceived high background crime and rural locality were at a significantly increased risk of robbery. Being a franchise store or the store having a new appearance decreased the risk of robbery in the univariate analysis. Being open 24 hours was the only significant risk factor for aggravated assault in the univariate analysis. These variables were included in the adjusted multivariable models examining rates of robbery or aggravated assault and their relationship to store characteristics (Table 1) and measures of robbery prevention (Table 2). The possibility of collinearity was evaluated using the condition index¹⁹ and it was determined that collinearity was unlikely to have affected the rate ratio estimates.

Table 1:

Crime Rate Ratios by Store Characteristics, 2010–2015

			Robbery		Aggravated Assault
Covariate	Level	N	RR (95% CI)	Adj. RR* (95% CI)	RR (95% CI)	Adj. RR* (95% CI)
Type	Corporate	63	0.94 (0.62–1.43)	1.24 (0.78–1.95)	1.05 (0.61–1.80)	1.62 (0.92–2.84)
	Franchise	61	0.48 (0.30–0.77)	0.63 (0.38–1.05)	0.59 (0.33–1.05)	0.81 (0.42–1.56)
	Single Owner	167	-	-	-	-
Background Crime	Yes	120	1.55 (1.10–2.18)	1.41 (1.01–1.97)	1.51 (0.97–2.35)	1.38 (0.91–2.09)
	No	173	-	-	-	-
Open 24-Hours	Yes	64	1.13 (0.74–1.70)	1.38 (0.91–2.08)	1.91 (1.16–3.15)	2.72 (1.65–4.49)
	No	226	-	-	-	-
Located in a Strip Mall	Yes	77	0.89 (0.60–1.33)	0.92 (0.62–1.36)	0.97 (0.59–1.59)	0.94 (0.58–1.52)
	No	216	-	-	-	-
Located Near a Highway	Yes	154	0.88 (0.62–1.24)	0.95 (0.66–1.36)	1.12 (0.72–1.73)	1.22 (0.78–1.92)
	No	139	-	-	-	-
Store Location	Urban	117	1.19 (0.82–1.74)	1.07 (0.71–1.60)	1.20 (0.74–1.94)	1.25 (0.74–2.13)
	Rural	47	1.82 (1.12–2.94)	1.80 (1.08–3.00)	1.45 (0.77–2.71)	2.30 (1.17–4.51)
	Residential	129	-	-	-	-
Store Appearance	New	80	0.51 (0.29–0.91)	0.43 (0.23–0.81)	0.35 (0.17–0.74)	0.21 (0.09–0.48)
	Old Well Kept	176	0.80 (0.48–1.33)	0.79 (0.47–1.33)	0.79 (0.42–1.49)	0.66 (0.34–1.30)
	Old	36	-	-	-	-

^*Model includes: location in an area of background crime (Yes, No), store location (Urban, Rural; Residential), being open 24 hours (Yes, No), store type (Corporate, Franchise; Single Owner), and store appearance (New, Old – Well Kept; Old)

Table 2:

Crime Rate Ratios by Compliance with Ordinance, 2010–2015

			Robbery		Aggravated Assault
Covariate	Level	N	RR (95% CI)	Adj. RR* (95% CI)	RR (95% CI)	Adj. RR* (95% CI)
Full Compliance	Yes	20	0.83 (0.41–1.68)	0.58 (0.28–1.19)	0.83 (0.34–2.01)	0.64 (0.25–1.61)
	No	265	-	-	-	-
Compliance Level	5–6	108	1.02 (0.68–1.55)	0.96 (0.62–1.47)	1.05 (0.63–1.74)	1.06 (0.62–1.79)
	4	80	0.91 (0.58–1.42)	0.99 (0.63–1.56)	0.61 (0.34–1.08)	0.75 (0.42–1.33)
	0–3	97	-	-	-	-
Employee Training	Yes	221	1.15 (0.76–1.73)	1.17 (0.77–1.77)	1.18 (0.70–1.98)	1.08 (0.65–1.82)
	No	70	-	-	-	-
Signage	Yes	57	1.15 (0.75–1.77)	1.04 (0.67–1.64)	0.95 (0.54–1.65)	0.88 (0.49–1.57)
	No	234	-	-	-	-
Visibility of Cash Counter/Sales Area	Yes	188	0.89 (0.62–1.28)	1.00 (0.69–1.43)	0.90 (0.57–1.42)	0.98 (0.63–1.54)
	No	105	-	-	-	-
Cash Drop Box with Cash Accessibility Policy	Yes	164	1.04 (0.74–1.48)	0.86 (0.60–1.24)	1.13 (0.73–1.77)	0.95 (0.60–1.50)
	No	129	-	-	-	-
Alarm System	Yes	234	1.12 (0.71–1.76)	1.05 (0.65–1.71)	0.94 (0.53–1.64)	1.16 (0.64–2.10)
	No	54	-	-	-	-
Cameras	Yes	287	1.02 (0.30–3.47)	1.26 (0.39–4.04)	0.74 (0.17–3.32)	0.77 (0.19–3.14)
	No	6	-	-	-	-

^*Model includes: location in an area of background crime (Yes, No), store location (Urban, Rural; Residential), being open 24 hours (Yes, No), store type (Corporate, Franchise; Single Owner), and store appearance (New, Old – Well Kept; Old)

The majority of the stores were single owner (57.4%), and the remainder were roughly equal numbers of franchise and corporate stores. While not statistically significant, robbery at corporate stores trended higher when compared to single owner stores (adj. RR = 1.24, 95% CI: 0.78–1.95), and robbery at franchise stores trended lower when compared to single owner stores (adj. RR = 0.63, 95% CI: 0.38–1.05). The results were similar for the analysis examining aggravated assault. None of the adjusted rate ratios comparing corporate and franchise stores to single owner was statistically significant (Table 1). When comparing franchise and corporate stores, franchise stores had a significantly lower rate of robbery (RR = 0.52, 95% CI: 0.30–0.89). The rate of aggravated assault was also lower for franchise stores compared to corporate stores, but this difference was not statistically significant.

Slightly less than half of all convenience and small grocery stores were in a perceived high crime area (41.0%). These stores had statistically significantly higher rates of robbery (adj. RR = 1.41, 95% CI: 1.01–1.97), and although not statistically significant, there was also an increased risk of aggravated assault (adj. RR = 1.38, 95% CI: 0.91–2.09). Stores who reported being open twenty-four hours (22.1%) experienced higher adjusted rates of aggravated assaults (adj. RR=2.72, 95% CI: 1.65–4.49) than stores not open 24 hours (Table 1). Compared to stores in suburban locations, convenience stores in rural locations experienced an 80% higher rate of robbery (adj. RR = 1.80, 95% CI: 1.08–3.00) and more than double the rate of aggravated assault (adj. RR = 2.30, 95% CI: 1.17–4.51)after the ordinance was in effect. No significant differences in crime rates between convenience and small grocery stores in suburban and urban areas were found. Finally, convenience stores with a new appearance had statistically significantly lower rates for both measurements of crime when compared to businesses with an older appearance. This effect was most pronounced for aggravated assault (adj. RR = 0.21 95% CI: 0.09–0.48).

Crime prevention elements associated with the crime ordinance and their effect on the various measures of crime are presented in Table 2. Only 7.0% (n=20) of the stores had all six of the crime prevention elements. Convenience stores having all six of these elements in place (i.e., in full compliance) had decreased rates of robbery and aggravated assault when compared to stores who had less than six of these elements, but this decrease was not statistically significant. The rarity of full compliance limited the power of this comparison. Having a camera had the strongest increase in experiencing robbery compared to businesses with no camera (adj. RR = 1.26, 95% CI: 0.39–4.04). However, only six businesses did not have cameras and the results were not statistically significant. Overall, stores that adhered to individual elements of the ordinance did not appear to have a reduced risk of robbery or aggravated assault when compared to stores that did not adhere to these requirements.

Figure 1 shows the changes in robbery and aggravated assault over the study period. In 2010, the year the ordinance was enacted, there appeared to be lower rates of robbery than 2009 for convenience and small grocery stores. The comparison of the rates pre- and post-ordinance using mixed negative binomial regression are presented in Table 3 for the two measures of crime by selected business characteristics. The robbery rate for convenience and small grocery stores was substantially lower (RR = 0.38, 95% CI: 0.29–0.51) in the years following the ordinance (2010–2015) compared to the businesses’ robbery rate during the period before the ordinance (2006–2009). This decreased robbery rate was consistently found across business types, urban and suburban locations, and overall compliance with the ordinance.

Figure 1:

Number of robbery and aggravated assaults per year for stores included in study

Note: Ordinance effective January 1, 2010

Table 3:

Crime Rate Ratios Comparing Rates After Ordinance (2010–2015) to Before Ordinance (2006–2009)

		Robbery	Aggravated Assault
All		0.38 (0.29–0.51)	NC
Type	Corporate	0.38 (0.21–0.69)	0.87 (0.40–1.92)
	Franchise	0.29 (0.13–0.65)	1.04 (0.34–3.19)
	Single Owner	0.41 (0.29–0.58)	0.80 (0.49–1.29)
Location	Urban	0.36 (0.23–0.56)	0.86 (0.47–1.56)
	Rural	0.57 (0.30–1.07)	0.94 (0.38–2.29)
	Residential	0.34 (0.21–0.54)	NC
Compliance	5–6	0.41 (0.20–0.83)	0.87 (0.48–1.56)
	4	0.42 (0.24–0.76)	NC
	0–3	0.35 (0.22–0.58)	0.88 (0.46–1.67)

NC=Non-convergence of model

Figure 1 shows that there was no meaningful change in aggravated assaults after the ordinance was enacted. Among the models that converged, the mixed negative binomial regressions showed no change pre- to post-ordinance enaction.

DISCUSSION

Robbery risk was reduced in the years following implementation of the ordinance requiring robbery prevention controls. Convenience and small grocery store robbery rates decreased substantially post-ordinance, regardless of ownership type. Robbery rates also decreased for stores in both urban and suburban locations, with a smaller non-statistically significant decrease observed for stores in rural locations. We saw a similar reduction in the pre- and post-ordinance crime rates regardless of level of compliance with the ordinance. We were not able to link any one element of the ordinance’s prevention strategies to the observed decrease in robberies.

Aggravated assaults were not meaningfully impacted by the ordinance. These types of crimes were not the primary target of the ordinance, so it is expected that rates would remain relatively constant over the time periods. Aggravated assaults can occur on the business property but outside of the actual business. Several of the ordinance provisions focus on the interior of the business, so they will be less effective in reducing crimes for the property as a whole. Future crime prevention efforts should also consider ways to reduce aggravated assault to further protect convenience store workers.

A possible explanation for the reduction in robbery is temporal changes in various city characteristics (i.e. population, environment, or social) or enforcement strategies that may have resulted in changes to overall reported crime. However, if general changes in crime determinants were to explain the results, the change in overall crime might be expected to exhibit a similar decrease. Instead, published crime statistics for Houston indicated that background crime rates for overall crime increased from 2006 to 2015. Therefore, the decreases in robbery rates cannot be explained by temporal changes in factors that influence crime rate alone.

While not statistically significant, we found a general decrease in robbery and aggravated assault for stores with full compliance with the ordinance when compared to stores not in full compliance. There were very few stores in full compliance, so additional evaluation of full compliance to the ordinance is needed. A previous study found businesses with low implementation of a Crime Prevention Through Environmental Design (CPTED) program had similar rates of robbery to businesses without the programs, while those with high implementation had a non-statistically significant decreased risk of robbery.^{14 16} In the current study it is difficult to fully assess implementation. Only the presence of these elements in 2011 can be evaluated while the degree of implementation of each element may vary across businesses and time. This makes it challenging to identify which stores contribute more in reducing robberies using CPTED programs.

Previous studies of CPTED programs have found that businesses that use these programs have a reduced risk of robbery, but have similarly not been able to consistently identify the most effective elements of these programs.^{10 12–14 16 21} Of the CPTED elements we investigated, only cash handling policies were found to be significantly related to crime in previous studies.^{12 21} Hendricks et al ²¹ used a matched case-control study design to investigate how various CPTED elements and store characteristics were related to robbery. Hendricks et al ²¹ found that cash handling policies and bullet-resistant shielding were the only CPTED elements that reduced the risk of robbery after controlling for various store characteristics. In an evaluation of CPTED elements of Florida convenience stores Amandus et al ¹² found that cash handling policies were associated with an increased rate of robbery. This is contrary to what is expected for cash handling policies and may be the result of unmeasured confounding in the previous study. We did not find cash handling polices to be related to decreased robbery rate as part of adherence to the cash handling and cash drop box requirements of the ordinance.

We did find several environmental characteristics, unrelated to the ordinance, that were associated with an increase in robbery or aggravated assault risk. Stores located in neighborhoods with higher perceived background crime rates were associated with robbery, 24-hour stores were associated with aggravated assault, and rurally located stores were associated with both robbery and aggravated assault. Stores self-described as new in appearance were significantly associated with reduced robbery and aggravated assault. Similarly, both Hendricks et al ²¹ and Amandus et al ¹² found several store characteristics associated with robbery rate and concluded that environmental and social factors of the surrounding community needed to be considered in the analysis of robbery prevention measures. Hendricks et al ²¹ found location in a shopping center, older building construction, proximity to graffiti, and years of operation as various store characteristics related to robbery rates. Amandus et al ¹² found high county crime rate and high county population size were related to an increased rate of robbery. We did not look at social characteristics as risk factors. We did find rural located stores to have a higher risk of both of our crime measures compared to suburban and urban locations conflicting with these previous results. In the pre and post analysis of the effectiveness of the ordinance robbery rates were reduced for all locations, but a statistically significant reduction in robbery rate was only observed for stores in urban and suburban stores.

The timing of the manager surveys which were used to assess compliance with the ordinance was a limitation of this study. The investigators were not able to conduct store manager surveys and perform safety measure checklists until about one year after implementation of the ordinance. This limited the study design for measuring store compliance with a post-only assessment where pre-ordinance implementation and timing of safety measures relied on recall of the store manager. Changes to compliance is also likely to change over time and we only assessed compliance at one point. This misclassification could partially explain our inability to link individual prevention measures to robbery prevention. Additionally, a lack of heterogeneity in the various stores’ overall approach to robbery prevention may further explain our inability to find a difference in effectiveness between the individual elements. While these limitations make it difficult to draw conclusions about individual elements of the robbery prevention ordinance, we did find the ordinance overall to be effective in reducing robbery supporting that a comprehensive approach to robbery prevention is more effective. This reduction may be tempered by the slow adoption of a comprehensive approach. Conversely, awareness of robbery prevention surrounding the implementation of the ordinance may have itself been effective in reducing robbery. Additional analysis of robbery rates is necessary to determine if a sustained impact resulted from the ordinance.

This study provides evidence supporting the effectiveness of city ordinances requiring adherence to crime prevention measures, especially for robbery. Additionally, we determined that post ordinance rural locations had a higher rate of robbery compared to stores in suburban areas. Continued efforts to understand the differences between urban, suburban, and rural stores are needed to more effectively develop and implement crime prevention ordinances. The ordinance did appear to contribute to a reduction in robbery, with rates trending down for stores regardless of location, but there was only a statistically significant reduction for stores located in suburban and urban areas. The inability to detect an effect for rural stores is in part due to the small number of stores located in the city studied, although it had rural parts within its city limits. Additional studies focusing on rural locations would further inform the effectiveness of robbery prevention ordinances for rural stores. Future endeavors to adopt similar ordinances in urban and suburban areas could lower the burden of injury and crime to convenience store employees and managers.

1. What is already known about this subject?

Compliance with ordinances designed to reduce convenience store crime that mandate a comprehensive set of safety measures has been found to be low.

2. What are the new findings?

Crime reports corresponding to before and after a city ordinance reveal a post-implementation reduction in robbery rates. There is insufficient evidence to identify a particular safety measure or threshold for compliance that is tied to reduced robbery rates.

3. How might this impact policy or clinical practice in the foreseeable future?

An ordinance using Crime Prevention Through Environmental Design (CPTED) principles was effective in reducing post-implementation robbery rates in small retail businesses. Ordinances requiring CPTED elements have been shown to reduce robbery, but when developing these ordinances policymakers should consider including measures to reduce other types of crimes, such as aggravated assault, for small retail businesses.