Calibrating Police Activity Across Hot Spot and Non-Hot Spot Areas

Abstract

Maximizing crime prevention through large-scale implementation of hot spot policing requires a more refined understanding of how to calibrate police activity across high and low-risk areas. This study investigates these issues based on the experience of a large urban police agency that substantially reduced proactive activities across a large area due to resource cutbacks while also shifting a larger share of its declining proactive work into prioritized micro hot spots. Time series models were used to estimate the effects of these changes on crime-related calls in hot spots and non-hot spot areas. Hot spots required higher levels of proactivity (expressed as rates per day or per crime) to control crime, and serious crime rose in these locations following modest reductions in proactivity. In areas outside hot spots, minor and property crimes rose, but only after reductions of one-half to two-thirds in proactive work. Violence was unaffected in these areas, and they did not experience accelerated growth in crime relative to prioritized hot spots. These results help to illuminate minimum levels of police activity that may be necessary to control crime in places of varying risk. They also suggest that police can reduce proactive work by substantial amounts in lower risk areas to place more emphasis on hot spots. Better understanding of these issues is central to widespread, systematic operationalization of hot spot policing as a means to reduce crime across large areas.

Keywords

proactive policing hot spots urban crime time series

“Hot spot” policing—i.e., policing focused on small geographic places or areas where crime is concentrated—has been one of the most important policing innovations of recent decades (Weisburd & Braga, 2019). The use of crime mapping to identify hot spots is common among police agencies (Burch, 2012; Reaves, 2010; Weisburd & Lum, 2005), and police cite hot spot enforcement as a leading approach to the reduction of violence and other crime problems (Police Executive Research Forum, 2007, 2008; also see Koper, 2014). Numerous evaluation studies also show that police interventions of various sorts focused on hot spots reduce crime at these locations (for reviews, see Braga et al., 2019; Lum & Koper, 2017; National Academies of Sciences, Engineering, and Medicine (NAS), 2018; National Research Council (NRC), 2004; Telep & Weisburd, 2012).

However, adapting hot spot policing from small-scale tests to system-wide implementation across large areas or an entire jurisdiction (i.e., “scaling up” this practice as described by Sherman et al., 2014) requires further understanding of how to optimize police activity across different areas based on their risk levels. In particular, practitioners need clearer guidance on the varying levels of police dosage that may be necessary to manage both hot spots and lower risk areas across a jurisdiction. Having more direct evidence on this issue would inform assessments of how much policing is needed in hot spots and the degree to which policing can be diverted from low risk areas to accommodate needs in hot spots without facilitating crime displacement to low risk areas or otherwise undermining deterrence and prevention in those areas. Without such understanding, it is not clear whether implementing a more widespread, systematic, and sustained preventive emphasis on hot spots in everyday police operations can produce large-scale aggregate reductions in crime (Nagin & Sampson, 2019; Sherman et al., 2014; Weisburd et al., 2017; Weisburd & Telep, 2014). This study addresses selected aspects of this problem using data from a U.S. city that experienced substantial reductions in proactive policing activity across a large area. Although not a formal test of a scaled-up hot spot policing effort, the study examines key dosage issues raised above that are implicit in the scaling up idea. Specifically, we investigate differences in the minimum levels of proactive policing that are necessary to prevent crime in hot spots and lower risk locations. In addition, we examine the utility of shifting a larger share of proactive policing into hot spots under these conditions from a macro, system-wide perspective. We then consider the implications of the findings for efforts to design, implement, and evaluate hot spot policing operations on a larger scale.

Preventing Crime in Hot Spots and Lower Risk Areas

Hot spot policing is grounded in research showing that about half of crime in a jurisdiction occurs at 5% or less of its street blocks and that this concentration is stable over time, due in large measure to chronic problem locations (e.g., Sherman et al., 1989; Sherman & Weisburd, 1995; Weisburd et al., 2004). Environmental, routine activities, and place-based criminologists have shown that these locations are often nodes for everyday living, business, leisure, and/or travel activities, and they commonly have social and environmental features that create criminal opportunities and facilitate offending (Brantingham & Brantingham, 1993; Eck, 2002; Roncek, 1981). Examples include locations with bars, convenience stores, parks, bus stops or depots, apartment buildings, parking lots, shopping centers, motels or hotels, adult businesses, and the like (e.g., Block & Block, 1995; Braga et al., 1999, pp. 551–552; Eck & Weisburd, 1995; Groff & Lockwood, 2014; Koper et al., 2015; Sherman et al., 1989, 45).¹ In addition, hot spots are often characterized by high levels of social disorganization as measured by low property values and high levels of social and physical disorder (e.g., places with high levels of loitering and abandoned buildings or vehicles) (Weisburd et al., 2012).

Controlling crime at these locations should—in theory—reduce it across the entirety of a jurisdiction (Weisburd et al., 2004). To provide a simple illustration, reducing crime by 20% at hot spots that generate 50% of a jurisdiction’s crime should reduce the locality’s overall crime level by roughly 10%, assuming no substantial displacement to other locations. However, research on hot spot policing has not yet clearly or consistently demonstrated such “system-level” impacts (Nagin & Sampson, 2019). Further, the application of hot spot policing on a sufficiently large scale to realize such benefits raises practical and operational questions for police that remain unanswered. These questions concern the resources needed to manage hot spots and the degree to which police can or should divert resources away from low risk areas to prioritize hot spots.

The first issue concerns the dosage of policing, measured in presence and/or activity levels, that can prevent crime most efficiently in hot spots. A growing number of studies indicate that police can reduce crime in hot spots by visiting them for relatively brief periods (typically around 15 minutes per stop—see Koper, 1995) once or a few times per day (sometimes less regularly) depending on the types and severity of the locations’ problems (e.g., Ariel et al., 2016; Barnes et al., 2020; Koper et al., 2015, 2021; Rosenfeld et al., 2014; Telep et al., 2014; Williams & Coupe, 2017). Nonetheless, it remains unclear how much police presence or activity is needed at minimum to keep crime in check at hot spots (i.e., to prevent it from rising). This is because hot spot policing studies have typically focused on the effects of marginal increases in police dosage relative to baseline levels (which are not always specified). Researchers have rarely examined the effects of reducing police dosages in hot spots or explicitly tested for minimal dosages that can prevent crime in these locations (for exceptions, see Barnes et al., 2020; Gibson et al., 2017).² Yet, understanding minimum dosage levels needed for hot spots could be very important to agencies seeking to expand their coverage across large numbers of problem locations as well as agencies struggling with staffing reductions and/or negative community reactions to what citizens perceive as over-policing.

The second issue is how a scaled up hot spot strategy might affect areas outside designated hot spots (i.e., non-hot spot areas), particularly if the strategy significantly reduces policing in these areas. One consideration is the possibility that crime will be displaced from hot spots to other areas, thus offsetting some or all of the crime prevention gains from a hot spot strategy. Studies demonstrate that police presence and activities in targeted hot spots usually do not displace crime to areas very nearby (Bowers et al., 2011; Braga et al., 2019). Whether crime displaces to locations farther from targeted hot spots and how extensively is debatable. Offenders operating at a hot spot cannot easily move their criminal activities elsewhere unless they find other locations that present similar criminal opportunity structures. Offenders also have to feel comfortable moving to these other locations, as this may increase their risks of detection and/or victimization, particularly if the locations are less familiar (Weisburd et al., 2006).³ Displacement opportunities should also become more limited as police expand their hot spot interventions throughout a locality (many hot spot policing studies are based on test programs that were limited in geographic scope). Nevertheless, this issue has received little systematic study (for exceptions, see Blattman et al., 2019; Koper et al., 2021).

Beyond displacement, another consideration is that systematically focusing operations on hot spots (e.g., having patrol officers spend all or most of their non-committed time at hot spots) will require police to reduce their presence and activities in non-hot spot areas. Shifting police resources and activities from low to high risk areas is logical with respect to crime prevention and organizational efficiency. Nonetheless, there could be critical levels of police presence and activity that are needed to prevent crime, as well as public demand for police service, from rising in low risk areas and countering the benefits of hot spot operations (also see Nagin & Sampson, 2019).⁴ Hence, operationalizing hot spot policing on a widespread basis may require further efforts to determine the optimal balance in resource and activity allocation across hot spot and non-hot spot locations.

Extant research provides little guidance on this issue. For starters, existing knowledge is not sufficient to specify minimum police dosage levels needed to control crime and satisfy public demand, calibrated for areas of varying risk. Findings from the landmark Kansas City Preventive Patrol Experiment (Kelling et al., 1974) suggest that proactive patrol can be reduced substantially, or perhaps eliminated, without causing crime to rise in lower risk areas (but see critiques of the Kansas City study as summarized by Sherman & Weisburd, 1995). However, research since then has emphasized focusing police attention on more precisely defined high-risk locations. There has been little to no follow-up work on minimum patrol levels needed to manage lower risk areas. Further, only a few studies have attempted to measure the effects of hot spot initiatives measured over larger areas like police beats, administrative districts, or entire jurisdictions. These studies have produced varying positive and null results (Blattman et al., 2019; Caeti, 1999; Jang et al., 2012; Koper et al., 2021; Lawton et al., 2005; Mohler et al., 2015; Smith & Purtell, 2007; Uchida & Swatt, 2013; Weisburd et al., 2015). However, some examined hot spot initiatives that were carried out by special units or supported by overtime funding or other additional resources; as a result, there was likely little change in resource allocation to non-hot spot areas as would likely be needed for more extensive implementation of hot spot policing. Moreover, none of these studies completely measured changes in policing and crime outside hot spot locations.

Another way of examining optimal resource allocation across areas is illustrated by Weisburd et al. (2017), who used agent-based modeling to simulate the effects of hot spot policing strategies on robbery measured at the borough level, which they define as an area encompassing about 3.2 square miles. Using various assumptions to simulate the levels, temporal changes, sensitivity to police risk, and geographic distribution of robbery in very large cities, Weisburd et al. estimated that a deployment scheme directing one-third of officers to spend half of their time at the top five hot spots in their respective beats would reduce robbery by 2.4 percent at the borough level in comparison to the use of standard random (i.e., undirected) patrol. In contrast, directing half of officers to spend all of their time at their top five hot spots was estimated to reduce robbery at the borough level by 11.7 percent over random patrol. These results suggest that the benefits of hot spot policing outweigh any counteracting effects from displacement or the reduction of police resources and activities in lower risk areas. Nonetheless, it remains to be seen whether these simulations will be borne out by real-world experience as applied to robbery as well as other forms of crime, disorder, and citizen demands for police service. For instance, Nagin and Sampson (2019, p. 137) note that Weisburd et al.’s projections show a questionable result of lower crime in non-hot spot areas despite the diversion of resources from those places.

The study presented here attempts to address some of the questions raised in the preceding discussion by examining the experience of a large urban police agency that substantially reduced proactive police activity in a large part of its jurisdiction during a time of staffing cutbacks. These reductions occurred in both hot spot and non-hot spot locations but were more accelerated in the latter as the agency shifted a larger share of its activity into selected hot spots. This study estimates and compares the effects of these reductions in police activity on crime and disorder in the hot spot and non-hot spot locations to help illuminate minimum levels of policing that may be needed to control crime in these different types of locations. The study also assesses whether the shift of more proactive work into hot spots improved the allocation of police resources across the study area as a whole based on observed trends in the hot spots and non-hot spot areas.

Study Site and Context

This investigation focuses on one administrative division of a large metropolitan police agency in the United States (hereafter, “Division A”). The agency overall serves a population of more than 1 million with higher than average rates of serious crime for jurisdictions of similar size.⁵ At the time of this study, Division A covered approximately 100 square miles of the jurisdiction and had a population of more than 400,000 (more recently, the division’s boundaries have been changed). Located outside the core city’s downtown area, Division A was a densely populated but largely residential and lower poverty area, with some sections that were more heavily commercial or industrial.⁶ The division accounted for the largest share of the agency’s calls for service in comparison to other divisions. However, it was not the agency’s highest risk area on a per capita basis, as it accounted for 30% of the service area population but only 18% its calls and roughly 15% of its homicides.

The study examines a recent four-year timeframe. As described below, proactive police activities began declining notably in Division A during year 3 of the study period. This was linked to a 7% decline in sworn officer staffing that occurred across the agency (a loss of about 200 officers) at a time when calls for service in Division A were rising by 4 percent (hence, workloads per officer were increasing, which likely reduced officers’ time for proactive work and thus added to the decline in proactivity). Proactivity patterns in the division were further altered during the early months of year 4 by an initiative that focused more of the division’s proactive efforts (particularly directed patrols, community policing efforts, and business/premise checks) on nine micro hot spots with concentrations of commercial, entertainment, and temporary lodging establishments that contributed disproportionately to crime problems in the division. These locations, most of which were less than 0.1 square miles, together constituted about 1% of the division’s land area but accounted for about 13% of its serious crime.⁷ During years 1 and 2 (the baseline period), the hot spots generated about ten crimes per week on average, including about two serious property crime calls and one violent crime call per week. Although proactive work declined both in the hot spots and outside the hot spots during years 3 and 4, the decline was more accelerated outside the hot spots due in part to the hot spot initiative.

Table 1.

Weekly Average Proactive Policing and Crime Calls in HS Locations for Baseline Period (Years 1 and 2) and Semi-Annual Periods of Years 3 and 4.

	Proactive work	Total crime (proactive to crime ratio)	Violent and property crime (proactive to crime ratio)	Violent crime (proactive to crime ratio)
Years 1–2	217	87 (2.49)	20 (10.85)	9 (24.11)
Year 3, first half	217	99 (2.19)	20 (10.85)	10 (21.7)
Year 3, second half	158	100 (1.58)	23 (6.87)	10 (15.8)
Year 4, first half	157	123 (1.28)	32 (4.91)	13 (12.08)
Year 4, second half	110	132 (0.83)	39 (2.82)	15 (7.33)

Key aspects of these trends are illustrated in Figures 1 and 2, which present weekly proactive policing activities and crime-related calls for service in the hot spot (hereafter, HS) locations and non-hot spot (hereafter, NHS) areas, respectively, during the study period. (The composition of the proactivity and crime series are described below.) Proactive work began declining considerably in the NHS areas in the early part of year 3 and in the HS locations by the late part of that year. Notably, the juxtaposition of the proactive policing and citizen call trends shows that the decline in proactivity, which by the latter part of year 4 was roughly half in the HS locations and nearly two-thirds in the NHS areas, was not caused by a decline in crime.

Figure 1.

Proactive Policing and Crime-Related Calls for Service in Hot Spot (HS) Locations by Week, Years 1–4.

Figure 2.

Proactive Policing and Crime-Related Calls for Service in Non-Hot Spot (NHS) Areas by Week, Years 1–4.

Further, proactive work became more concentrated in HS locations during this period. As shown in Figure 3, the weekly ratio of proactive activities in HS locations to those in NHS areas increased 39%, from an average of 0.18 during the baseline period of years 1 and 2 to an average of 0.25 during year 4. Stated differently, 15% of the division’s proactive work was focused in the HS locations during the baseline period. By year 4, this percentage had increased one-third, rising to 20%.

Figure 3.

Ratio of Proactive Activities in Hot Spots (HS) to Proactive Activities in Non-Hot Spot (NHS) Areas by Week, Years 1–4.

Data and Methods

Changes in proactive policing and the effects thereof in the designated HS locations and the NHS areas of Division A were investigated using calls for service data from the agency’s computer-assisted dispatch (CAD) system. The CAD system captures both citizens’ calls for police service and officer-initiated calls to dispatchers to record various citizen interactions, investigative activities, and administrative functions. Proactive police activities were measured using officer-initiated calls for a variety of investigative, enforcement, and preventive actions (e.g., see Wu & Lum, 2017). The vast majority of these activities (roughly 97%) consisted of vehicle stops, investigation of suspicious persons and vehicles, directed patrols, premise checks, and other investigative follow-ups. Though limited in some respects (e.g., few community policing or problem-solving activities were recorded),⁸ these activities are typical of the types of proactive work commonly employed in patrol and tracked by American police agencies (Koper et al., 2020). Ample evidence also suggests these police actions can affect crime and disorder, particularly when focused on hot spots (e.g., Boydstun, 1975; Braga et al., 2019; Bryant et al., 2015; Jang et al., 2012; Josi et al., 2000; Koper & Mayo-Wilson, 2012; McGarrell et al., 2001; NAS, 2018; NRC, 2004; Rosenfeld et al., 2014; Sherman & Rogan, 1995; Telep et al., 2014; Wilson & Boland, 1978). To measure outcomes, citizen calls for service were grouped into three broad categories: total calls about crime and disorder,⁹ combined calls for violent crime (e.g., assault, robbery) and property crime (e.g., burglary, larceny), and calls for violent crime only.¹⁰ The call records were geocoded (with a 91% success rate) to determine whether they corresponded to events occurring in one of the targeted HS locations or elsewhere in the division (final sample n = 542,450).

Descriptive and time series analyses were used to assess changes in the outcome measures in response to changes in police proactivity, which are treated as exogenous based on the context of these policing changes as discussed above. For this purpose, the call data were aggregated into two time series, one for the combined HS locations and one for the combined NHS areas. Creating a single data series for the NHS areas required aggregating data over a large area that undoubtedly contained locations with heterogenous characteristics, crime risks, and proactive police work. However, the data were analyzed in this way for illustrative purposes to examine the overall balance of proactive work and crime across the HS and NHS areas as operationally defined by the agency during this period. The data for each set of areas were also aggregated temporally into weekly measures, providing a time series of 208 weeks across the four study years. Time series regression models were then estimated for each crime type grouping within each type of area.

The first set of models estimated for the HS locations and NHS areas test whether there were statistically significant shifts in the level of each crime series during years 3 and 4, controlling for seasonal patterns and other forms of trending and autocorrelation as described below. The effects of major changes in police proactivity were estimated with a series of semi-annual indicator variables corresponding to the early and late periods of years 3 and 4 (years 1 and 2 serve as the baseline reference period). The decision to model the effects of changes in proactivity in this exploratory manner was based on notable shifts in the levels and geographical distribution of proactivity that corresponded roughly to these time frames (see Figures 1 through 3, and Tables 1 and 2). The semi-annual indicators are thus intended to capture the cumulative effects of changes in policing that occurred over long periods. A measure of weekly proactive activities was also incorporated into the model to control for the possible effects of week to week variations in proactivity.¹¹ However, as shown, this measure was rarely statistically significant, which suggests that crime and disorder levels are more strongly influenced by longer-term patterns of police activity than by weekly fluctuations.

Table 2.

Weekly Average Proactive Policing and Crime Calls in NHS Areas for Baseline Period (Years 1 and 2) and Semi-Annual Periods of Years 3 and 4.

	Proactive work	Total crime (proactive to crime ratio)	Violent and property crime (proactive to crime ratio)	Violent crime (proactive to crime ratio)
Years 1–2	1,236	1,293 (0.96)	210 (5.86)	82 (15.07)
Year 3, first half	977	1,332 (0.73)	205 (4.77)	81 (12.06)
Year 3, second half	930	1,342 (0.69)	206 (4.51)	78 (11.92)
Year 4, first half	622	1,369 (0.45)	221 (2.81)	83 (7.49)
Year 4, second half	445	1,418 (0.31)	240 (1.85)	86 (5.17)

An additional set of models was also estimated to assess whether crime trends in the NHS areas significantly diverged from those in the HS locations as proactivity patterns changed during years 3 and 4. For this purpose, a ratio variable was constructed for each crime series based on the number of those crimes that occurred in the HS locations each week divided by the number that occurred in the NHS areas. Controlling for any prior trends, significant shifts in these ratios during years 3 and 4 provide insight into whether the changing distribution of proactive work across HS locations and NHS areas was optimal, despite the declining levels of proactivity in both sets of places. Of most interest, a significant reduction in this ratio would suggest that the reallocation of more proactive work to the HS locations caused crime to rise in the NHS areas while remaining steady in the HS locations or to rise faster in the NHS areas than in the HS locations. This could potentially reflect crime displacement from HS locations and/or excessively low levels of proactivity in NHS areas.¹² In turn, such a pattern would suggest that perhaps too much proactive work had been shifted away from the NHS areas. The semi-annual indicators for years 3 and 4 were used to test for shifts in each ratio series during those years, and the weekly proactivity variable was converted to a ratio reflecting the level of proactivity in HS locations to that in NHS areas each week.

All models included monthly indicator variables to control for seasonal patterns. Autoregressive and/or moving average terms were also added as needed to control for serial correlation in the residual terms based on inspection of autocorrelation and partial autocorrelation functions (McCleary & Hay, 1980) and evaluation of the Portmanteau test (Box & Pierce, 1970; Ljung & Box, 1978) for serial correlation. Examination of residual correlation in the estimated models and preliminary testing of all outcome series with Augmented Dickey-Fuller tests for mean stationarity (Davidson & MacKinnon, 1993; Hamilton, 1994) suggested that it was not necessary to de-trend the outcome variables through the use of time trends or first differencing. Accordingly, all models presented below were estimated with outcomes measured in levels.

Results

Tables 1 and 2 present average weekly counts of proactive policing and crime-related calls in the HS locations and NHS areas, respectively, for the baseline period (years 1 and 2) and the semi-annual periods of years 3 and 4. Each table also shows the ratio of proactive to crime calls, calculated separately for each crime category, to illustrate the changes in proactivity relative to crime trends in the study areas (we refer to these as proactivity ratios). Tables 3 and 4 present time series models for changes in crime in the HS locations and NHS areas, respectively.

Table 3.

Time Series Regression Models for Weekly Measures of Crime in HS Locations.

	Total crime	Violent and property crime	Violent crime
Weekly proactivity	.07 (.03)*	.02 (.01)	.00 (.01)
Year 3, first half	15.7 (4.3)***	.4 (1.6)	.8 (.9)
Year 3, second half	14.5 (3.8)***	4.7 (1.8)**	1.6 (1.0)
Year 4, first half	45.0 (4.5)***	13.5 (2.0)***	4.2 (1.1)***
Year 4, second half	49.0 (4.9)***	21.5 (1.8)***	6.2 (.9)***
Autoregressive term (lag 1)	.18 (.08)*	–	–
Portmanteau test for white noise	B = .613p = .846	B = 1.084p = .190	B = 1.127p = .157

Note. N = 208. Standard errors of coefficients in parentheses. Monthly indicators (not shown) included in all models. Portmanteau test value shown for the 40th lag.

*statistically significant at p<=.05; **statistically significant at p<=.01; *** statistically significant at p<=.001.

Table 4.

Time Series Regression Models for Weekly Measures of Crime in NHS Areas.

	Total crime	Violent and property crime	Violent crime
Weekly proactivity	–.01 (.05)	–.00 (.01)	–.00 (.01)
Year 3, first half	48.6 (33.2)	–7.0 (9.5)	–3.2 (3)
Year 3, second half	30.1 (25.1)	–5.1 (7.4)	–2.4 (2.8)
Year 4, first half	85.9 (39.4)**	10.4 (11.2)	–2.1 (5.5)
Year 4, second half	102.5 (46.6)**	27.7 (11.7)*	5.4 (5)
Autogressive term (lag 1)	.21 (.07)***	.89 (.16)***	–
Moving average term (lag 1)	–	–.83 (.20)***	–
Portmanteau test for white noise	B = .899 $p$ = .39	B = .660 $p$ = .78	B = .518 $p$ = .95

Note. N = 208. Standard errors of coefficients in parentheses. Monthly indicators (not shown) included in all models. Portmanteau test value shown for the 40th lag.

*statistically significant at p<=.05; **statistically significant at p<=.01; ***statistically significant at p<=.001.

Proactive policing calls in the HS locations declined by 27% in the latter half of year 3 and held steady in the early part of year 4 before dropping to 49% below baseline levels by the late part of year 4 (Table 1). During the baseline period, officers recorded three to four proactive activities per day on average in each HS location; by the latter part of year 4, this figure had declined to less than two. In the NHS areas, proactive calls declined steadily during years 3 and 4, falling to 25% below their baseline levels by the latter half of year 3 and to nearly two-thirds below baseline by the late part of year 4 (Table 2). Police conducted about 4 proactive activities per week per square mile in the NHS areas in the latter part of year 4, down from about 12 per week and square mile during the baseline period. Expressed relative to crime trends, the ratio of proactive calls to crime calls in the HS locations dropped 67% to 74% across the three crime call categories from the baseline period to the latter half of year 4 (Table 1). Similarly, these proactivity ratios had dropped 66% to 68% by the late part of year 4 in the NHS areas (Table 2).

The decline in proactivity in both sets of areas was driven largely by a reduction in vehicle stops, which accounted for the majority of proactive work during the baseline period (56% in HS locations and 65% in NHS areas).¹³ Traffic stops declined by two-thirds in both sets of areas by the latter part of year 4 and accounted for 70%–79% of the overall drop in proactivity.¹⁴ Trends in other forms of proactivity varied by location and activity type, with NHS areas experiencing greater reductions in a wider range of activities. In NHS areas, other common proactive activities (as defined above) declined from 43% (for directed patrol) to 64% (for premise checks). In HS locations, premise checks, investigation of suspicious persons/vehicles, and other investigative follow-ups declined between roughly one quarter and one-third overall, though premise checks peaked during the early half of year 4 (at the start of the HS initiative). In contrast, directed patrols in HS locations remained close to or above baseline levels throughout years 3 and 4, and reached a high point in the first half of year 4.

Turning to crime trends, crime-related calls in the HS locations increased about 15% during year 3 for the overall crime and combined violent and property crime categories (Table 1). Larger increases then occurred in all categories during year 4. By the end of that year, crime levels in the HS locations were roughly 52% to 95% above their baseline levels. Time series models for the HS locations further illustrate these patterns, as shown by the semi-annual indicators for years 3 and 4 (Table 3). Total crime calls began increasing significantly early in year 3 even before proactive work in these places began declining. By the late half of year 3, combined calls for property and violent crimes were also rising significantly, driven mostly by growth in property crimes. These trends accelerated considerably in the early half of year 4, at which point calls for violence also began to rise significantly. This coincided with the division’s HS initiative and may reflect reporting effects associated with greater police presence and community-related efforts during those months (e.g., see Weisburd et al., 2020). Following the initiative, as proactivity declined further, crime continued to increase in the latter part of year 4.

In the NHS areas, crime-related calls showed little if any change during year 3 (Table 2). By the latter half of year 4, total crime-related calls had increased by nearly 10% over baseline, while combined property and violent crime calls had risen by 14%, due almost entirely to an increase in property crime calls. Time series models confirm that there were no statistically significant changes in crime-related calls in the NHS areas until year 4, when total crime and disorder calls rose throughout the year and combined violent and property crime calls increased during the late part of the year (Table 4)¹⁵ The increase in serious offenses appears to have been due to changes in property crime, as there were no statistically significant changes in violence calls.¹⁶

Models examining trends in the ratio of weekly HS crimes to weekly NHS crimes indicate that crime was rising at a faster rate in the HS locations during years 3 and 4 (Table 5). Shifts in the HS to NHS crime ratios were positive and statistically significant throughout much of year 3 and year 4 for all crime categories despite the shift of a higher share of proactive work to the HS locations. Conversely, this suggests that shifting more proactivity away from the NHS locations did not cause crime to rise disproportionately in those places.

Table 5.

Time Series Regression Models for the Weekly Ratio of Crime in HS Locations to Crime in NHS Areas.

	Total crime ratio	Violent and property crime ratio	Violent crime ratio
Weekly proactivity ratio	.04 (.02)^#	–.01 (.05)	.08 (.09)
Year 3, first half	.01 (.00)**	.00 (.01)	.01 (.01)
Year 3, second half	.01 (.00)**	.02 (.01)**	.02 (.01)^#
Year 4, first half	.02 (.00)***	.05 (.01)***	.04 (.02)**
Year 4, second half	.02 (.00)***	.07 (.01)***	.06 (.01)***
Portmanteau test for white noise	B = .744 $p$ = .64	B = .415 $p$ = 1	B = .705 $p$ = .70

Note. N = 208. Standard errors of coefficients in parentheses. Monthly indicators (not shown) included in all models. Portmanteau test value shown for the 40th lag.

#statistically significant at p<=.1; *statistically significant at p<=.05; **statistically significant at p<=.01; ***statistically significant at p<=.001.

Comparing coefficients across the HS and NHS models in Tables 3 and 4 also reveals that the HS locations accounted for much of the increase in crime across the entire division during years 3 and 4. By the latter half of year 4, the nine HS locations accounted for approximately one-third of the division’s weekly growth in total crime calls (49/(49 + 102.5)), almost half of its growth in combined property and violence calls (21.5/(21.5 + 27.7)), and all of its statistically significant increase in violence.

Discussion

This case study of one large urban jurisdiction has investigated the sensitivity of crime in high and lower risk areas to reductions in police proactivity. Minor crime and disorder (e.g., disturbances and suspicion calls) showed the greatest responsiveness to police activities across both sets of areas, followed by property crime. Violent crimes showed the least sensitivity, particularly in NHS areas. As expected, crime of all types showed more sensitivity to changes in proactivity in HS locations.

Patterns in the HS locations must be interpreted cautiously. Total crime calls began rising in the first half of year 3 before there was a decline in proactivity. As noted, patterns in the early part of year 4 may have been confounded by reporting effects stemming from the division’s brief HS initiative. Nevertheless, it appears that dosage levels of three to four proactive activities per day (per spot) were insufficient to prevent some forms of crime from rising in the HS locations, and further reductions in dosage were associated with increases in more serious forms of crime. Expressing dosage levels relative to crime suggests that proactivity to total crime ratios of lower than 2.5 either contributed to increases in minor crime or were insufficient to stop increases caused by other factors. Combined property and violent crimes rose significantly when the proactivity to crime ratio for these offenses dropped to about seven, and violence increased significantly when the proactivity to violent crime ratio fell to about twelve.

Crime levels in lower risk, NHS areas showed much less sensitivity to reductions in proactive policing. Proactive work declined by 50% to nearly two-thirds in these areas before there was evidence of increases in crime (or crime reporting) for minor and property offenses. Violence levels, in contrast, were unaffected by changes in police activity within the range of activity levels observed.

The proactivity ratios needed to control crime in NHS areas were also much lower than in HS locations. The proactivity to total crime ratio dropped below 0.7 in NHS areas before total crime rose significantly. For combined property and violent crime, the ratio dropped to as low as 2.8 without triggering significant increases in these crimes. Finally, a proactivity to violent crime ratio of roughly five to one was sufficient to prevent violence from rising.

The findings also illustrate that reallocating a higher share of proactive work into HS locations can provide a more optimal allocation of resources. In this study, the share of proactive work conducted in HS locations increased by one-third without producing a disproportionate or accelerated increase in crime in NHS areas. Indeed, the results suggest that an even larger shift of proactive work to the HS locations would have been desirable to maximize crime prevention. To illustrate, proactive activities declined by at least 306 per week across the NHS areas (by the late half of year 3) without causing any change in crime (this represents a 25% reduction from the baseline level of 1,236 proactive policing calls per week during years 1 and 2). Shifting these efforts to the targeted HS locations would have been more than enough to offset the entire drop in proactivity observed across those places through the end of year 4, which amounted to 107 fewer activities per week. In theory, this might have prevented much or all of the increase in crime observed in the HS locations and across the division, if proactive work could have been held at those levels.

The significance of these findings is twofold. First, the study provides tentative insight into minimum proactivity levels necessary to control crime in HS locations, expressed as activity rates per day or per crime. The figures presented here will need replication in other sites and types of HS locations, but they could prove useful as a starting point for police agencies seeking to optimize resource allocation and trying to prevent crime in HS locations from rising with minimal intrusiveness (e.g., see Barnes et al., 2020). The latter concern is likely to be particularly important in the United States as police adjust to strong public backlash in recent years (especially during 2020) from unnecessary use of force and other aggressive forms of policing in minority communities.

The second issue of significance is that this study directly supports, up to a point, an assumption that is central to “scaling up” hot spot policing (Sherman et al., 2014) into an operational model for patrol activity throughout a jurisdiction—specifically, the implicit notion that police can reduce their activities in low risk areas, perhaps by substantial amounts, to place more emphasis on HS locations without causing crime or citizens’ concern about crime to rise in these lower risk areas. Indeed, these findings imply that lower risk areas are over-policed in some jurisdictions (also see Wu & Lum, 2017). At the same time, they also suggest that there are minimum, threshold levels of proactivity needed to keep crime from rising even in lower risk areas. This issue has not been explicitly evaluated in prior hot spot research, but studies of it seem necessary for determining the optimal balance of police activities across HS and NHS areas.

This study provides only a starting point for this line of inquiry, and several caveats should be noted. For one, this study has not evaluated whether hot spot policing reduces crime across larger areas. The changes in policing that occurred in the study jurisdiction did not constitute a planned intervention that could be evaluated prospectively under controlled conditions. In this particular context, shifting resources to HS locations during a time of declining resources may have prevented crime from rising further than it did across the division studied, but the initiative was limited in scope, and there was not a sufficient increase in proactivity in the HS locations to test for additional crime prevention effects. On the contrary, proactivity levels were below desirable levels in both HS and NHS areas, which further complicated assessment of trends. Accordingly, the analyses presented here should be viewed as an exploratory effort to illuminate issues for future study that can inform efforts to scale up hot spot policing.

Another general limitation to the study is that other factors which may have affected crime trends in the study location were not explicitly measured. Hence, shifts in crime during the early and/or late periods of years 3 and 4 may have been due in some measure to factors other than reductions in policing. Even so, a conservative reading of these results (from a policy standpoint) is that proactive policing levels were insufficient to prevent crime from rising during these periods, even if that rise was caused by other social forces.

Conclusions here are also contingent on the types of proactive work under investigation. Some beneficial forms of proactive work, including problem-solving, situational crime prevention assessments, and positive community engagement (e.g., see NAS, 2018), were rarely used, or at least rarely recorded in CAD (a problem that is common in police agencies—see Koper et al., 2020). Hence, conclusions here may not generalize well to all forms of proactive work.

With regard to the recorded activities, it is notable that most of the decline in proactivity was specific to traffic stops in both sets of areas. The utility of traffic stops for reducing crime is a matter of debate, particularly when they are not used in carefully targeted ways in crime hot spots (Lum & Nagin, 2017; also see, e.g., Bryant et al., 2015; Jang et al., 2012; Josi et al., 2000; Koper & Mayo-Wilson, 2012; McGarrell et al., 2001; Weiss & Freels, 1996). Traffic stops can also contribute disproportionately to negative police-citizen interactions and racial disparities in police actions (Epp et al., 2014; NAS, 2018). Reducing unnecessary levels of traffic stops in lower risk areas might thus be beneficial in some regards. Further study should be directed to this issue but with additional consideration of how changes in traffic stops might affect vehicular accidents, which was not examined here (e.g., Bryant et al., 2015; DeAngelo & Hansen, 2014).

However, the reduction in traffic stops in HS locations, coupled with more modest reductions in other forms of proactive investigative work, had stronger deleterious impacts despite the fact that directed patrols in these locations were maintained at steady or even higher levels during years 3 and 4. This is consistent with some research suggesting that patrol visibility alone may not be sufficient to control crime in serious urban hot spots without more active enforcement and investigative work (Rosenfeld et al., 2014). On the other hand, the dosage of directed patrols, which was approximately 2 to 4 per week in the HS locations on average, may have been insufficient to generate strong preventive effects; in these types of urban hot spots, daily or multiple daily dosages may be needed (e.g., see Telep et al., 2014).

A related limitation is that police time and visibility in the study areas were not directly measured (e.g., using automated vehicle locator data (AVL) or survey measures of citizens’ perceptions) and may not have changed to the same degree as did proactive activities. The trend in directed patrol calls suggests a decrease in police visibility in the NHS areas but not in the HS locations. Nevertheless, more explicit measurement of police presence, perhaps using some combination of CAD, AVL, and survey data, would be needed for more precise assessments of police visibility vis-à-vis police activities. In future work, more specific measurements of both police presence and activities could help to provide more refined assessments of how police resources and activities should be allocated across different types of areas. Additional work, likely involving citizen survey data, will also be needed to understand how citizens in HS and NHS areas perceive changes in police presence and activity and whether these changes affect their reporting of crime and disorder.

Changes in police presence and activities may also have impacts on citizens’ attitudes towards police that were not examined here. Reductions in police presence and enforcement-oriented proactivity could conceivably have positive or negative effects on citizens’ confidence and trust in police depending on the type of location, type of activity (see discussion of traffic stops above), and citizen demographics. Further, if the general decline in proactivity in the study area also reduced positive community engagement, this may have had important ramifications for police-community relations and citizens’ views of police legitimacy and effectiveness.

Finally, caution should be exercised in generalizing these results to other jurisdictions or specific types of areas. The effects of reducing proactivity by a given amount likely vary based on social context as well as baseline levels of both proactivity and crime. This study has examined the experience of just one jurisdiction. Further, the study examined changes over a large NHS area that likely had much heterogeneity within (hence, specific locations within the NHS area may have required more or less proactive activity compared to the average across the NHS area). Nonetheless, the findings presented here may provide a useful reference point for future studies and comparisons. As further results of this sort accumulate across a variety of tests and settings, researchers and police may be able to specify the optimal calibration of police deployment and activity across high and low-risk locations with more precision.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the federal Bureau of Justice Assistance, U.S. Department of Justice (cooperative agreement 2011-DB-BX-K012). The views expressed are those of the authors.

ORCID iDs

Christopher S. Koper

Cynthia Lum

Notes

Author Biographies

Dr. Christopher S. Koper is an Associate Professor in the Department of Criminology, Law and Society at George Mason University and the principal fellow of George Mason’s Center for Evidence-Based Crime Policy. Dr. Koper holds a Ph.D. in Criminology and Criminal Justice from the University of Maryland and has worked previously for the Police Executive Research Forum, the University of Pennsylvania, the Urban Institute, the RAND Corporation, the Police Foundation, and other organizations. He specializes in issues related to policing, firearms policy, and program evaluation.

Dr. Xiaoyun Wu is a Research Associate in the Department of Criminology, Law and Society at George Mason University. Her research interests include policing, deterrence, and crime prevention. Her current work focuses on understanding everyday police proactive activities and deployment and their dynamic relationships with crime, disorder, and accidents, using various data sources, measurements, and statistical methods. She holds a PhD in Criminology, Law and Society from George Mason University.

Dr. Cynthia Lum is a Professor of Criminology, Law and Society at George Mason University and the director of its Center for Evidence-Based Crime Policy. She researches in the areas of policing, technology, evidence-based crime policy, translational criminology, and crime prevention. She holds a Ph.D. in Criminology and Criminal Justice from the University of Maryland.

References

Ackerman

J. M.

Rossmo

D. K.

(2015). How far to travel? A multilevel analysis of the residence-to-crime distance. Journal of Quantitative Criminology, 31(2), 237–262.

Ariel

Weinborn

Sherman

L. W.

(2016). Soft” policing at hot spots—do police community support officers work? A randomized controlled trial. Journal of Experimental Criminology, 12(3), 277–317.

Barnes

G. C.

Williams

Sherman

L. W.

Parmar

House

Brown

S. A.

(2020). Sweet spots of residual deterrence: A randomized crossover experiment in minimalist police patrol. SocArXiv Papers. https://doi.org/10.31235/osf.io/kwf98

Bernasco

(2010). A sentimental journey to crime: Effects of residential history on crime location choice. Criminology, 48(2), 389–416.

Bishaw

Posey

K. G.

(2016, December 8). A comparison of rural and urban America: Household income and poverty. United States Census Bureau, Census Blogs. https://www.census.gov/newsroom/blogs/random-samplings/2016/12/a_comparison_of_rura.html

Blattman

Green

Ortega

Tobon

(2019). Place-based interventions at scale: The direct and spillover effects of policing and city services on crime (SSRN working paper 3050823). National Bureau of Economic Research.

Block

R. L.

Block

C. R.

(1995). Space, place, and crime: Hot spot areas and hot places of liquor-related crime. In J. E. Eck & D. Weisburd (Eds.), Crime and place, crime prevention studies (Vol. 4, pp. 145–184). Criminal Justice Press and Police Executive Research Forum.

Bowers

Johnson

Guerette

R. T.

Summers

Poynton

(2011). Spatial displacement and diffusion of benefits among geographically focused policing initiatives. Campbell Systematic Reviews, 7(1), 1–144. https://doi.org/10.403/csr.2011.3

Box

G. E.

Pierce

D. A.

(1970). Distribution of residual autocorrelations in autoregressive-integrated moving average time series models. Journal of the American Statistical Association, 65(332), 1509–1526.

10.

Boydstun

J. E.

(1975). San Diego field interrogation final report. Police Foundation.

11.

Braga

A. A.

Bond

B. J.

(2008). Policing crime and disorder hot spots: A randomized controlled trial. Criminology, 46(3), 577–607.

12.

Braga

A. A.

Turchan

B. S.

Papachristos

A. V.

Hureau

D. M.

(2019). Hot spots policing and crime reduction: An update of an ongoing systematic review and meta-analysis. Journal of Experimental Criminology, 15(3), 289–311.

13.

Braga

A. A.

Weisburd

D. L.

Waring

E. J.

Mazerolle

L. G.

Spelman

Gajewski

(1999). Problem-oriented policing in violent crime places: A randomized controlled experiment. Criminology, 37(3), 541–580.

14.

Brantingham

P. J.

Brantingham

P.L.

(Eds.). (1991). Environmental criminology (2nd ed.). Waveland Press.

15.

Brantingham

P. L.

Brantingham

P. J.

(1993). Environment, routine and situation: Toward a pattern theory of crime. In R. Clarke & M. Felson (Eds.), Routine activity and rational choice, advances in criminological theory (Volume 5, pp. 259–294). Transaction Press.

16.

Bryant

K. M.

Collins

G. M.

White

(2015). Shawnee, Kansas, smart policing initiative: Reducing crime and automobile collisions through Data-Driven approaches to crime and traffic safety (DDACTS). Bureau of Justice Assistance, U.S. Department of Justice.

17.

Burch

A. M.

(2012). Sheriffs’ offices, 2007—statistical tables. Bureau of Justice Statistics, U.S. Department of Justice.

18.

Caeti

T. J.

(1999). Houston’s targeted beat program: A quasi-experimental test of police patrol strategies [Doctoral dissertation]. Sam Houston State University.

19.

Clarke

R. V. G.

Felson

(Eds.). (1993). Routine activity and rational choice ( Vol. 5). Transaction Publishers.

20.

Cohen

L. E.

Felson

(1979). Social change and crime rates: A routine activity approach. American Sociological Review, 44(4), 588–608.

21.

Davidson

MacKinnon

J. G.

(1993). Estimation and inference in econometrics. Oxford University Press.

22.

DeAngelo

Hansen

(2014). Life and death in the fast lane: Police enforcement and traffic fatalities. American Economic Journal: Economic Policy, 6(2), 231–257.

23.

Eck

J. E.

(2002). Preventing crime at places. In L. W. Sherman, D. P. Farrington, B. C. Welsh, & D. L. MacKenzie (Eds.), Evidence-based crime prevention (pp. 241–294). Routledge.

24.

Eck

J. E.

Weisburd

(1995). Crime places in crime theory. In J. E. Eck & D. Weisburd (Eds.), Crime and place: Crime prevention studies (vol. 4, pp. 1–33). Criminal Justice Press and the Police Executive Research Forum.

25.

Epp

C. R.

Maynard-Moody

Haider-Markel

D. P.

(2014). Pulled over: How police stops define race and citizenship. University of Chicago Press.

26.

Felson

(1987). Routine activities and crime prevention in the developing metropolis. Criminology, 25(4), 911–932.

27.

Gabor

Gottheil

(1984). Offender characteristics and spatial mobility: An empirical study and some policy implications. Canadian Journal of Criminology, 26(3), 267–281.

28.

Gibson

Slothower

Sherman

L. W.

(2017). Sweet spots for hot spots? A cost-effectiveness comparison of two patrol strategies. Cambridge Journal of Evidence-Based Policing, 1(4), 225–243.

29.

Groff

E. R.

Lockwood

(2014). Criminogenic facilities and crime across street segments in Philadelphia: Uncovering evidence about the spatial extent of facility influence. Journal of Research in Crime and Delinquency, 51(3), 277–314.

30.

Hamilton

J. D.

(1994). Time series analysis ( Vol. 2). Princeton University Press.

31.

Jang

Lee

C. B.

Hoover

L. T.

(2012). Dallas’ disruption unit: Efficacy of hot spots deployment. Policing: An International Journal of Police Strategies & Management, 35(3), 593–614.

32.

Josi

D. A.

Donahue

M. E.

Magnus

(2000). Conducting blue light specials or drilling holes in the sky: Are increased traffic stops better than routine patrol in taking a bite out of crime? Police Practice and Research, 1, 477–507.

33.

Kelling

G. L.

Pate

Dieckman

Brown

C. E.

(1974). The Kansas City preventive patrol experiment: A summary report. Police Foundation.

34.

Klinger

D. A.

Bridges

G. S.

(1997). Measurement error in calls‐for‐service as an indicator of crime. Criminology, 35(4), 705–726.

35.

Koper

C. S.

(1995). Just enough police presence: Reducing crime and disorderly behavior by optimizing patrol time in crime hot spots. Justice Quarterly, 12(4), 649–672.

36.

Koper

C. S.

(2014). Assessing the practice of hot spots policing: Survey results from a national convenience sample of local police agencies. Journal of Contemporary Criminal Justice, 30(2), 123–146.

37.

Koper

C. S.

Egge

S. J.

Lum

(2015). Institutionalizing place-based approaches: Opening ‘cases’ on gun crime hot spots. Policing, 9(3), 242–254.

38.

Koper

C. S.

Lum

Hibdon

(2015). The uses and impacts of mobile computing technology in hot spots policing. Evaluation Review, 39(6), 587–624.

39.

Koper

C. S.

Lum

Fritz

(2020). Proactive policing in the United States: A national survey. Policing: An International Journal, 43(5), 861–876.

40.

Koper

C. S.

Lum

Hegarty

(2021). The long-term and system-level impacts of institutionalizing hot spot policing in a small city. Policing: A Journal of Policy and Practice. Advance online publication. https://doi.org/10.1093/police/paaa096

41.

Koper

C. S.

Mayo-Wilson

(2012). Police strategies to reduce illegal possession and carrying of firearms: Effects on gun crime. Campbell Systematic Reviews, 8(1), 1–53. https://doi.org/10.4073/csr.2012.11

42.

Koper

C. S.

Taylor

B. G.

Woods

D. J.

(2013). A randomized test of initial and residual deterrence from directed patrol and use of license plate readers at crime hot spots. Journal of Experimental Criminology, 9(2), 213–244.

43.

Lawton

B. A.

Taylor

R. B.

Luongo

A. J.

(2005). Police officers on drug corners in Philadelphia, drug crime, and violent crime: Intended, diffusion, and displacement impacts. Justice Quarterly, 22(4), 427–451.

44.

Ljung

G. M.

Box

G. E.

(1978). On a measure of lack of fit in time series models. Biometrika, 65(2), 297–303.

45.

Lum

Koper

C. S.

(2017). Evidence-based policing: Translating research into practice. Oxford University Press.

46.

Lum

Koper

C. S.

Johnson

Stoltz

(2020). Examining the empirical realities of proactive policing through systematic observations and computer-aided dispatch data. Police Quarterly, 23(3), 283–310.

47.

Lum

Nagin

D. S.

(2017). Reinventing American policing. Crime and Justice, 46(1), 339–393.

48.

Mazerolle

L. G.

Ready

Terrill

Waring

(2000). Problem-oriented policing in public housing: The Jersey City evaluation. Justice Quarterly, 17(1), 129–158.

49.

McCleary

Hay

R. A.

(1980). Applied time series analysis for the social sciences. Sage Publications.

50.

McGarrell

E. F.

Chermak

Weiss

Wilson

(2001). Reducing firearms violence through directed police patrol. Criminology and Public Policy, 1(1), 119–148.

51.

Mohler

G. O.

Short

M. B.

Malinowski

Johnson

Tita

G. E.

Bertozzi

A. L.

Brantingham

P. J.

(2015). Randomized controlled field trials of predictive policing. Journal of the American Statistical Association, 110(512), 1399–1411.

52.

Nagin

D. S.

Sampson

R. J.

(2019). The real gold standard: Measuring counterfactual worlds that matter most to social science and policy. Annual Review of Criminology, 2(1), 123–145.

53.

National Academies of Sciences, Engineering, and Medicine., Weisburd

Majmundar

M. K.

(Eds.). (2018). Proactive policing: Effects on crime and communities. The National Academies Press.

54.

National Research Council., Skogan

Frydl

(Eds.). (2004). Fairness and effectiveness in policing: The evidence. The National Academies Press.

55.

Police Executive Research Forum. (2007). Violent crime in America: “A tale of two cities”. Author.

56.

Police Executive Research Forum. (2008). Violent crime in America: What we know about hot spots enforcement. Police Executive Research Forum.

57.

Reaves

B. A.

(2010). Local police departments, 2007. Bureau of Justice Statistics, U.S. Department of Justice.

58.

Roncek

D. W.

(1981). Dangerous places: Crime and residential environment. Social Forces, 60(1), 74–96.

59.

Rosenfeld

Deckard

M. J.

Blackburn

(2014). The effects of directed patrol and self‐initiated enforcement on firearm violence: A randomized controlled study of hot spot policing. Criminology, 52(3), 428–449.

60.

Sherman

L. W.

Gartin

P. R.

Buerger

M. E.

(1989). Hot spots of predatory crime: Routine activities and the criminology of place. Criminology, 27(1), 27–55.

61.

Sherman

L. W.

Rogan

D. P.

(1995). Effects of gun seizures on gun violence: “Hot spots” patrol in Kansas City. Justice Quarterly, 12(4), 673–694.

62.

Sherman

L. W.

Weisburd

(1995). General deterrent effects of police patrol in crime “hot spots”: A randomized, controlled trial. Justice Quarterly, 12(4), 625–648.

63.

Sherman

L. W.

Williams

Ariel

Strang

L. R.

Wain

Slothower

Norton

(2014). An integrated theory of hot spots patrol strategy: Implementing prevention by scaling up and feeding back. Journal of Contemporary Criminal Justice, 30(2), 95–122.

64.

Smith

D. C.

Purtell

(2007). An empirical assessment of NYPD's “operation impact”: A targeted zone crime reduction strategy. NYU Wagner Graduate School of Public Service.

65.

Telep

C. W.

Mitchell

R. J.

Weisburd

(2014). How much time should the police spend at crime hot spots? Answers from a police agency directed randomized field trial in Sacramento, California. Justice Quarterly, 31(5), 905–933.

66.

Telep

C. W.

Weisburd

(2012). What is known about the effectiveness of police practices in reducing crime and disorder? Police Quarterly, 15(4), 331–357.

67.

Uchida

Swatt

M. L.

(2013). Operation LASER and the effectiveness of hotspot patrol: A panel analysis. Police Quarterly, 16(3), 287–304.

68.

Warner

B. D.

Pierce

G. L.

(1993). Reexamining social disorganization theory using calls to the police as a measure of crime. Criminology, 31(4), 493–517.

69.

Weisburd

Braga

A. A.

(Eds.). (2019). Police innovation: Contrasting perspectives (2nd ed.). Cambridge University Press.

70.

Weisburd

Braga

A. A.

Groff

E. R.

Wooditch

(2017). Can hot spots policing reduce crime in urban areas? An agent-based simulation. Criminology, 55(1), 137–173.

71.

Weisburd

Bushway

Lum

Yang

S. M.

(2004). Trajectories of crime at places: A longitudinal study of street segments in the city of Seattle. Criminology, 42(2), 283–321.

72.

Weisburd

Gill

Wooditch

Barritt

Murphy

(2020). Building collective action at crime hot spots: Findings from a randomized field experiment. Journal of Experimental Criminology. Advance online publication. https://doi.org/10.1007/s11292-019-09401-1.

73.

Weisburd

Green

(1995). Policing drug hot spots: The Jersey City drug market analysis experiment. Justice Quarterly, 12(4), 711–735.

74.

Weisburd

Groff

E. R.

Jones

Cave

Amendola

K. L.

Yang

S. M.

Emison

R. F.

(2015). The Dallas patrol management experiment: Can AVL technologies be used to harness unallocated patrol time for crime prevention? Journal of Experimental Criminology, 11(3), 367–391.

75.

Weisburd

Groff

E. R.

Yang

S. M.

(2012). The criminology of place: Street segments and our understanding of the crime problem. Oxford University Press.

76.

Weisburd

Lum

(2005). The diffusion of computerized crime mapping in policing: Linking research and practice. Police Practice and Research, 6(5), 419–434.

77.

Weisburd

Telep

C. W.

(2014). Hot spots policing: What we know and what we need to know. Journal of Contemporary Criminal Justice, 30(2), 200–220.

78.

Weisburd

Wyckoff

L. A.

Ready

Eck

J. E.

Hinkle

J. C.

Gajewski

(2006). Does crime just move around the corner? A controlled study of spatial displacement and diffusion of crime control benefits. Criminology, 44(3), 549–592.

79.

Weiss

Freels

(1996). The effects of aggressive policing: The Dayton traffic enforcement experiment. American Journal of Police, 15(3), 45–64.

80.

Williams

Coupe

(2017). Frequency vs. length of hot spots patrols: A randomised controlled trial. Cambridge Journal of Evidence-Based Policing, 1(1), 5–21.

81.

Wilson

J. Q.

Boland

(1978). The effect of the police on crime. Law & Society Review, 12(3), 367–390.

82.

Lum

(2017). Measuring the spatial and temporal patterns of police proactivity. Journal of Quantitative Criminology, 33(4), 915–934.