Police Foot Patrol and Crime Displacement

Abstract

Police patrol, motorized and foot, has a long history of being used as a crime prevention method. Scientific evaluations of this crime prevention technique have been undertaken for at least 40 years, with mixed results. One of the important questions to be answered regarding the implementation of a police patrol is the presence of crime displacement: criminal activity simply moving around the corner, away from the primary patrol area. Previous investigations of this phenomenon have found that, most often, crime displacement is nonexistent or less than crime reductions in the primary area of interest. In this article, we investigate local crime displacement. We use a spatial point pattern test that can identify changes in the spatial patterns/distribution of crime even if crime in all areas has decreased. We find moderate evidence for the presence of this spatial shift and discuss the implications.

Keywords

policing crime hot spots police patrol crime displacement spatial point pattern test

Introduction

Police foot patrol is a method used as an attempt to reduce crime and disorder. The theoretical expectation of an increased police presence would be crime reductions because of an increase in perceived risk by offenders. However, as reviewed below, though recent research has found a positive effect from police foot patrol (reductions in criminal activity), the history of this literature has been far from consistent. In an era of crime reductions and fiscal restraint, the efficacy of policing needs to be scrutinized to best allocate resources and be able to justify the corresponding expenditures.

Part of the efficacy of police foot patrols is the issue of crime displacement. If the criminals who are discouraged by the presence of a police foot patrol simply move to a neighboring area and commit their crimes there (displacement), such a crime prevention initiative would not be a good allocation of policing resources. Despite the empirical support for a lack of crime displacement from police foot patrols, displacement areas may at times be very large compared with the primary police foot patrol area. This makes the identification of crime displacement difficult in some circumstances.

These issues are critically important to the future of hot spots policing. If the allocation of police resources to hot spots is going to be scrutinized to justify expenditures, these decisions need to be based on as much crime science as possible. In the current context, there are two important topics that need to be addressed for the future of hot spots policing. First, we need to know if displacement is present and how to measure it. As stated above, the empirical literature finds little support for displacement, but there is a measurement issue, discussed below. And second, we must continue to assess the utility of new spatial tools to investigate these issues. It may be true that there is no displacement of crime, but that does not mean that the spatial patterns of crime do not change favoring other areas.

In this article, we consider these issues, evaluating the impact of increased police foot patrol in the community of Lower Lonsdale, City of North Vancouver, British Columbia, Canada. Specifically, we investigate the presence of crime displacement from this police initiative. Previous research has identified a crime reduction with no evidence of crime displacement (Andresen & Lau, 2013). However, this analysis suffered from this identification problem for crime displacement, noted above. We address this issue through the use of a local analysis of crime displacement. We are able to show that considering the many areas in which crime displacement may occur significantly affects interpretations.

The Impact of Police Foot Patrol on Crime and Displacement Expectations

Kelling, Pate, Dieckman, and Brown (1974) is the first known systematic study of (motorized) police patrol. In this year-long study, the authors analyzed three different forms of police patrol in different police beats: reactive, proactive, and control. Reactive beats were only initiated when a call for service was made; proactive beats increased the police presence by as much as two to three times compared with before the study began, and control beats received the same amount of police patrols as before the study began. In terms of the level of crime, citizens’ attitudes toward police services, citizens’ fear of crime, police response time, and citizens’ satisfaction with police response time, Kelling et al. (1974) found no statistically significant differences for the different forms of police patrol. In a subsequent analysis of police foot patrols on crime and disorder, the Newark Foot Patrol Experiment only revealed a short-lived impact on some property crimes and community disturbances (Bowers & Hirsch, 1987; Kelling, Pate, Ferrara, Utne, & Brown, 1981; Pate, 1986). Without any notable impacts of police motorized or foot patrol, the conventional wisdom for many years was that police patrol was for the reduction in the fear of crime, not actual crime (National Research Council, 2004)—Esbensen (1987), analyzing police foot patrols in a medium-sized southeastern city, found similar results.

Trojanowicz (1986) analyzed the impact of a police foot patrol in Flint, Michigan, with more promising results. Trojanowicz (1986) found that the volume of reported crime fell by almost 9% in evaluation areas, whereas control areas experienced a 10% increase. Moreover, police foot patrol was found to be far more effective than police motor patrol. Nasar and Fisher (1993) sought to explain the differences in the results of studies on police foot patrol through an emphasis on the importance of the physical environment and situational factors as well as where and when police foot patrols took place. On this note, Sherman and Weisburd (1995) showed that there was a “hot time” for calls for police service between 7 p.m. and 3 a.m.

Koper (1995) showed the importance of different police patrol techniques in the context of police motorized patrol. In this research, Koper (1995) found that police officers must patrol proactively and unpredictably, stopping their vehicles for a minimum of 10 min. He found that the optimal time for police patrols to stop was between 14 and 15 min. The returns to these patrol stops diminished after this amount of time.

In a randomized control trial in Philadelphia, Ratcliffe, Taniguchi, Groff, and Wood (2011) undertook a 3-month before and after analysis of police foot patrol on violent crime. The researchers identified 120 hot spots and randomly assigned them to intervention and control areas. Ratcliffe et al. (2011) found that the intervention areas exhibited a 23% decrease in violent crime, relative to the control areas. Moreover, and curiously, these effects were only present in areas with a minimum threshold of violence.

And most recently, Andresen and Lau (2013) assessed the impact of a police foot patrol initiative in a relatively low-crime area in North Vancouver, British Columbia. They found that there was an overall decrease in calls for police service of 17% with most of this impact being from decreases in property crimes. Specifically, these authors found that the greatest impact of police foot patrol was on the property crime of mischief,¹ most often representing some form of property damage, with decreases in both the primary patrol area and the potential crime displacement area; only commercial burglary was also found to have a statistically significant decrease. Relevant for the current study, Andresen and Lau (2013) did not find any evidence of crime displacement in the context of assault, robbery, commercial burglary, theft, and mischief.

An important consideration for this literature is the possibility of crime displacement, a commonly measured phenomenon in the crime prevention literature (Cornish & Clarke, 1986, 1989). During the 1990s, there were three extensive reviews of the empirical evidence for crime displacement, or lack thereof (Barr & Pease, 1990; Eck, 1993; Hesseling, 1994). These reviews included analyses in Canada, the United States, the United Kingdom, and Continental Europe, and they all had the same general conclusion: If crime displacement was present, it was less than crime reductions in police patrol target areas. And most recently, Weisburd et al. (2006) investigated the phenomenon of crime displacement addressing the limitations of previous research. Overall, they found no evidence for crime displacement. Rather, areas surrounding the police patrol target areas (catchment areas) also experienced crime reductions—a diffusion of crime control benefits.

Despite these positive results for both police foot patrols and (the lack of) crime displacement, there is a methodological difficulty measuring crime displacement. Consider the situation presented in Figure 1. In this situation, the police patrol target area is the center block of a nine-block area. Consequently, the catchment area for potential displacement is eight times the size of the police patrol target area. Even if there were a 50% decrease in the police patrol target area, if displaced, those crimes would be spread across an area that is relatively large. This may result in not much of a change in the volume of crime in the catchment area even if there was perfect displacement.

Figure 1.

Measuring crime displacement.

In this article, we address this concern with a local analysis of crime displacement from a police foot patrol. Although the situation discussed above relative to Figure 1 is not as extreme in our analysis, the catchment area for displacement is three times the size of the police patrol target area. Rather than treating the catchment area as a whole, we analyze changes in criminal activity at the level of the police patrol atom²/beat, 32 units of analysis rather than 1 unit of analysis.

Study Area, Data, and Method

The Study Area: Lower Lonsdale, North Vancouver

The City of North Vancouver is a small municipality that is part of the Metro Vancouver area in British Columbia and patrolled by the Royal Canadian Mounted Police (RCMP). North Vancouver is located on the northern side of Burrard Inlet, north of the City of Vancouver; the City of North Vancouver is surrounded by the District of North Vancouver. North Vancouver has a population slightly less than 50,000 residents and has been growing at a moderate rate in recent years. Within North Vancouver itself, Lower Lonsdale is the most populated neighborhood, containing approximately 12,000 residents, and is almost four square kilometers in size—the City of North Vancouver is approximately 12 square kilometers. The average family income in Lower Lonsdale was just over CAN$64,000 (US$62,000) in 2010, essentially the same as City of North Vancouver as a whole. Within Lower Lonsdale, more than 90% of the residences are classified as apartments of some form.

Lower Lonsdale has a ferry terminal that is part of its transit system and departs every 15 min for Vancouver’s central business district. This ferry terminal is also connected to a large bus loop servicing the entire North Shore (North Vancouver and West Vancouver) and operates as a transportation transfer point for residents of the North Shore who work in Metro Vancouver. Lower Lonsdale consists of both residential and commercial developments that provide many amenities and recreational activities while also hosting numerous annual community events. Lower Lonsdale is the social capital of North Vancouver and attracts residents and nonresidents because of its varied land uses. These aspects of Lower Lonsdale create many opportunities for a variety of crime types, making it the highest volume crime area in this relatively low-crime municipality. Consequently, the police decided to implement a patrol presence in the area. These police foot patrols occurred during the summer months of 2010, June to September, and are continuing through to 2014.

The entire Lower Lonsdale area (primary patrol area and potential displacement area in Figure 2) is almost 4 square kilometers in area. The primary patrol area within Lower Lonsdale (the commercial strip) is approximately 0.90 square kilometers in area, 41 patrol atoms; the potential displacement area is approximately 3 square kilometers, 32 patrol atoms. One of the patrol atoms in the potential displacement area (the patrol atom that is furthest south in Figure 2) is just over 1.8 square kilometers on its own.

Figure 2.

Lower Lonsdale, North Vancouver.

The structure of the police foot patrols was to use 10-hr shifts to patrol the Lower Lonsdale area. From Monday to Friday, there was always one sworn member patrolling Lower Lonsdale at varying hours. This sworn member most often patrolled alone. Wednesday to Saturday, there were an additional two sworn members patrolling from 2 p.m. until midnight. And on Sunday, there were two sworn members patrolling from 12 p.m. until 10 p.m. With this schedule, there was at least one sworn member patrolling Lower Lonsdale each day of the week, particularly during the highest commercial and social activity hours for the area. The primary objective of this police foot patrol was to provide a high visible police presence to build stronger relationships within the community and businesses with the hope of decreasing crime as well.

Data

The units of analysis are police patrol atoms in Lower Lonsdale. There are a total of 73 police patrol atoms in the Lower Lonsdale area with 41 of those police patrol atoms being in the primary patrol area—see Figure 2. However, the majority of the police foot patrol in the primary patrol area was concentrated along the primary arterial road that cuts through the middle of the primary patrol area, Lonsdale Avenue, that is the border for 19 police patrol atoms.

The data source for the assessment of the impact of increased police foot patrol in Lower Lonsdale is the Police Records Information Management Environment (PRIME). These data contain the classification, location, date, and time of each police incident. The temporal coverage of these data is January 1, 2007, to September 30, 2010. Because the police foot patrol occurred from 9th June to 30th September in 2010, we restrict the analysis to this same time period for all available years to obtain the most appropriate comparisons. This provides 3 full years of data prior to the police foot patrol to provide a baseline for calls for service levels in the Lower Lonsdale area. Including 3 full years of data before the police foot patrol allows for analyses to compare the differences in spatial patterns of crime across 4 years as well as aggregating the 3 years of data prior to the police foot patrol in an effort to control for any aberrant years of data. The following crime types were made available for analysis: assault, commercial burglary, residential burglary, drug possession, mischief, robbery, shoplifting, theft, theft from vehicle, theft of bicycle, theft of vehicle, and an aggregate of all these crime types. However, because of very low-crime counts, robbery, shoplifting, theft of bicycle, and theft of vehicle are not included in the analysis below aside from within the aggregate crime type.

Spatial Point Pattern Test

To investigate the local impact of police foot patrol on crime displacement, the spatial point pattern test developed by Andresen (2009), and corresponding index, is instructive. This test is used to independently identify changes or differences in the spatial patterns of crime at the local level. This spatial point pattern test has been applied to test the stability in crime patterns (Andresen & Malleson, 2011), the appropriateness of aggregating crime types (Andresen & Linning, 2012), the spatial dimension of seasonal crime patterns (Andresen, 2013a; Andresen & Malleson, 2013a), and the degree of spatial heterogeneity within commonly used spatial units of analysis in spatial criminology (Andresen & Malleson, 2013b).

The ability to identify differences or similarities in spatial point patterns has proven to be very instructive thus far and has direct applicability to the future of hot spots and policing because it may lead to a better understanding of this phenomenon and, hence, a better ability to reduce crime. For example, knowing where hot spots are located is clearly important information for the allocation of police resources. However, an understanding of why the hot spot emerged in the first place could better inform the police regarding their choice of action to reduce crime. Is the spatial pattern of the crime type at a hot spot the same as another crime type? Do these hot spots coincide with the spatial patterns of everyday routine activities that may be modified (the individuals or the actual hot spot location) to reduce crime? Do these hot spots coincide with socioeconomic factors (health risks, physical environmental, or municipal resources) that may be changed to reduce criminal victimization? Answers to these questions can emerge from the use of this particular spatial point pattern test because it allows for the identification of places with similar concentrations—see Andresen (2013b) for some more discussion of these possibilities.

The spatial point pattern test can be conducted on various spatial patterns using any geographically defined unit, for example, police beats, census areas, street segments, or simply a grid placed on top of a study area. The output of the test is a global measure of similarity for the entire study area, as well as local results that identify similarity for each of the units. The test is not a part of any geographic information system software, but it is freely available in a graphical user interface: http://code.google.com/p/spatialtest/. Technical details of the testing algorithm are available in Andresen (2009), but the general nature of the test is as follows:

Nominate one of your data sets as the “base” data set.

Repeatedly sample from the other (test) data set to generate a confidence interval for testing.

For each spatial unit, see whether the percentage of points within the base data is within the range of percentages generated for the test data using the sampling procedure and whether it is within the range that unit is deemed similar.

Calculate an index of similarity that is the percentage of units that are defined as similar, ranging from 0 to unity.

The purpose of this spatial point pattern test is to create variability in one data set, so that it can be compared statistically with another data set. The 85% samples generated, each maintain the spatial pattern of the test data set and allows for a “confidence interval” to be created for each spatial unit that may be compared to the base data set—the 85% sample is based on the research by Ratcliffe (2004). Therefore, statistically significant changes/differences are identified at the local level.

The output of the test consists of two parts. First, there is a global parameter that ranges from 0 (no similarity) to 1 (perfect similarity): the index of similarity, S, is calculated as

S = \frac{\sum_{i = 1}^{n} s_{i}}{n},

where s_i is equal to 1 if two crimes are similar in spatial unit i and 0 otherwise, and n is the total number of spatial units. The S-index represents the proportion of spatial units that have a similar spatial pattern within both data sets. Although there is no established rule of thumb regarding the value of the S-index that indicates similarity, we turn to the literature discussing correlations. In the context of the variance inflation factor (VIF) and multicollinearity, O’Brien (2007) states that a VIF ranging from 5 to 10 tends to cause concern. In a bivariate context, this would approximately lead to a correlation coefficient ranging from .80 to .90. We use the value of .80 to indicate that two spatial point patterns are similar. Second, the test generates mappable output to show where statistically significant change occurs, that is, which census tracts, dissemination areas, police atoms, and street segments have undergone a statistically significant change.

Results

As stated above, a previous evaluation of the police foot patrol initiative in Lower Lonsdale found evidence for decreases in mischief and commercial burglary, with no evidence of crime displacement. In fact, Andresen and Lau (2013) found evidence for a diffusion of benefits in the context of mischief: Crime also decreased in the potential crime displacement catchment area. With this in mind, any search for local crime displacement is not searching for increases in criminal activity after the implementation of the police foot patrol. Rather, a search for local crime displacement is concerned with changes in the spatial pattern of the remaining criminal activity. As such, despite decreases in criminal activity in all areas, the remaining criminal activity may have shifted to the catchment area to avoid the police foot patrol. A “traditional” search for displacement will not be able to identify such a change because it searches for changes in the levels of criminal activity in both areas: primary patrol area and catchment area.

The results of the spatial point pattern test in Table 1 show the similarity of the spatial patterns for aggregate crime in the Lower Lonsdale area for a number of different time spans, all for the summer months when the police foot patrol took place in 2010. Immediately apparent is that there are year-to-year changes in the spatial patterns of crime in the Lower Lonsdale area: The S-index values are never greater than 0.55 in these comparisons. This should come as no surprise, particularly because Lower Lonsdale is a low-crime area such that relatively small changes in the locations of criminal activity can lead to significant changes in the spatial patterns of crime. Although the 2010 versus 2007-2009 results indicate a very low degree of similarity, this is also present when comparing 2007-2009 with the individual years within that time period. As such, the global statistic of similarity here does not give any indication that something different occurred with the implementation of the police foot patrol.

Table 1.

S-Index Values, Aggregate of All Crime Types.

	2010	2007-2009	2009	2008	2007
2010	1	0.315	0.356	0.329	0.301
2007-2009		1	0.507	0.534	0.479
2009			1	0.356	0.315
2008				1	0.370
2007					1

The results from the spatial point pattern test for the individual crime types are presented in Table 2. Because of the lower volume of crime when considering individual crime types, the 2010 spatial patterns of crime are compared with the 2007-2009 temporal aggregate within each crime type only. There is quite clearly significant variation in the values of the S-indices across the individual crime types, particularly when compared with the aggregate of all crime types. This provides further support for the claims of Andresen and Linning (2012) in that aggregating crime types is most often inappropriate when considering spatial patterns. All but one crime type is significantly less than 0.60 (theft from vehicle is slightly less than 0.60 at 0.589). This indicates a moderate degree of similarity in the spatial patterns of crime before and during the police foot patrol in Lower Lonsdale—Commercial burglary, when rounded to two decimal points, reaches the similarity threshold of 0.80, stated above. Initially, this may appear to be a curious result given that commercial burglary was identified by previous research as exhibiting a statistically significant decrease with the implementation of police foot patrol. However, the commercial land use in the Lower Lonsdale area is concentrated along the primary arterial road, Lonsdale Avenue, that runs down the middle of the primary patrol area. Consequently, there are few areas in which criminal activity may be displaced outside of the primary patrol area.

Table 2.

S-Index Values, By Crime Type, 2010 Versus 2007-2009.

	S-Index
Aggregate	0.315
Assault	0.644
Burglary—commercial	0.795
Burglary—residential	0.658
Drug possession	0.699
Mischief	0.438
Theft	0.603
Theft from vehicle	0.589

The most notable result is the S-index for mischief, 0.438. This is the lowest magnitude result for similarity, which also happens to be the most significant decrease in criminal activity from the police foot patrol by Andresen and Lau (2013). This is the first indication that a meaningful change in spatial patterns of criminal activity occurred because of the police foot patrol. However, at this point, the analysis must turn to the output from the spatial point pattern test to see whether indeed a spatial shift occurred into the catchment area.

The local output from the spatial point pattern test is presented in Figures 3a and b, aggregate crime and mischief, respectively. For aggregate crime, Figure 3a, there is little that can be garnered from this output. The spatial pattern of criminal activity has increased during the police foot patrol both within the primary patrol area and the catchment area. The only change that appears to be occurring is an increase in the concentration of criminal activity in the border areas of Lower Lonsdale. These border areas include police patrol atoms that are in the primary patrol area. However, because of the nature of police foot patrol, less time will be spent at these border areas; assuming the majority of the police foot patrol was along Lonsdale Avenue, the police officers would walk to the end of the road, turn around, and go the other way spending less time on the edges.

Figure 3.

Spatial point pattern test output.

The local output from the spatial point pattern test for mischief, Figure 3b, appears to give a greater indication of a spatial shift in criminal activity. There are six police patrol atoms that exhibit a statistically significant increase in criminal activity during the police foot patrol; two of these police patrol atoms are in the “core” of the primary patrol area and two are on the edge. The other two police patrol atoms are in the catchment area. In addition, the majority of the police patrol atoms in the primary patrol area exhibit statistically significant decreases in the spatial pattern of crime. As such, aside from two police patrol atoms, the spatial patterns of mischief appear to have shifted away from the core of the primary patrol area to the edges of the primary patrol area and into the catchment area. Although there is no “strong” indication of a spatial shift in the patterns of crime (nothing but statistically significant decreases in the primary patrol area and increases in the catchment area), there is an indication that this general pattern has occurred in the present context. This shows the importance of considering not only global statistical results but also local statistical results to identify the impact of a crime prevention initiative such as police foot patrol.

Discussion and Conclusion

Police (foot or motorized) patrol has been evaluated for at least 40 years with mixed results. Much of the most recent and methodologically complex evaluations have indicated that police patrol decreases criminal activity. Moreover, much of this research has not only identified decreases in criminal activity in the primary area of interest but also little or no displacement in the catchment area—The net change is always a decrease in criminal activity. In this article, we analyze the local spatial pattern of decreases in criminal activity resulting from a police foot patrol. This is an important question because despite potential decreases in criminal activity in both the primary area of interest and its catchment area, the spatial patterns of crime may change because of the increased police presence. In our analysis of police foot patrol in Lower Lonsdale, we have found some moderate evidence for such a change in spatial patterns taking place. This is most evident for the crime type of mischief, which happens to be the crime type found to have the most significant decreases in an evaluation of this crime prevention initiative.

It is important to recognize here the significance of such a change in spatial patterns. Yes, a decrease in criminal activity is a positive result that needs to be recognized, particularly if there is no evidence of crime displacement or if there is evidence of a diffusion of benefits. In such a situation, one may assume that the relatively easy targets for victimization have capable guardians by the presence of the patrolling police officers. This has led to offenders reducing their criminal activity because there are fewer suitable targets and/or offenders no longer committing crime because of the increased risk. However, if the hypothesized spatial shift is present, new information may be learned regarding the various opportunities for crime and their locations within the urban landscape—the opportunity surface. Knowing where (reduced) criminal activity shifts can inform where more crime prevention efforts may be targeted and help understand the relative importance of factors leading to that criminal activity. In other words, what are the characteristics of the areas that are the first, second, third, and so on, choices for criminal activity?

It is interesting to note that there was evidence for this phenomenon in mischief and not other crime types. This may simply be because of the volume of other crime types: There was not enough information to be able to find statistically significant changes in the spatial patterns of crime. However, it may also be the case some crime types are more resilient to displacement than others. Some crime types may require the presence of particular characteristics to occur. For example, economically based crime types such as drugs and prostitution may require very specific characteristics that can only be found in very specific places—They cannot be displaced in close proximity, and their spatial patterns in the immediate cannot be altered significantly. This is an empirical question and may be able to be identified through investigations of the first, second, and third choices of areas for criminal activity.

The first obvious direction for future research is to undertake this research in another location where a police (foot) patrol initiative has taken place. Specifically, it may be instructive for such an analysis to occur in a location with a greater volume of crime. Such a local analysis does not need to use the methodology employed here, but some form of local spatial analysis that allows for the identification of a spatial shift in crime patterns in the presence of an overall decrease in criminal activity. Second, field research needs to be conducted in the areas that exhibit increases in criminal activity after the implementation of police foot patrol. What is it about these areas that attract criminal activity once risk is increased in the places where offenders prefer to commit crime?

Footnotes

Declaration of Conflicting Interests

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

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Martin A. Andresen is an associate professor at Simon Fraser University, School of Criminology and Institute for Canadian Research Studies. He is also an associate member of the Department of Geography, Simon Fraser University and a member of the Crime and Place Working Group, Center for Evidence-Based Crime Policy, George Mason University. His research areas include spatial statistical analysis, crime at places, and environmental criminology with recent research published in Annals of the Association of American Geographers, Applied Geography, British Journal of Criminology, Environment and Planning A, European Journal of Criminology, and Journal of Research in Crime and Delinquency.

Nicolas Malleson is a lecturer at University of Leeds, School of Geography. His research areas include crime at places, agent-based modeling, and geocomputation with recent research published in Applied Geography, Computers, Environment and Urban Systems, Crime Patterns and Analysis, Environment and Planning B, International Journal of Geographical Information Science, Journal of Research in Crime and Delinquency, and Urban Studies.

References

Andresen

M. A.

(2009). Testing for similarity in area-based spatial patterns: A nonparametric Monte Carlo approach. Applied Geography, 29, 333-345.

Andresen

M. A.

(2013a). The science of crime measurement: Issues for spatially-referenced crime data. New York, NY: Routledge.

Andresen

M. A.

(2013b). Translating the analysis of patterns into police practice: An application of a new spatial point pattern test. Translational Criminology, Fall 2013, 7-9.

Andresen

M. A.

Lau

K. C. Y.

(2013). An evaluation of police foot patrol in Lower Lonsdale, British Columbia. Police Practice & Research. Advance online publication. doi:10.1080/15614263.2013.805870

Andresen

M. A.

Linning

S. J.

(2012). The (in)appropriateness of aggregating across crime types. Applied Geography, 35, 275-282.

Andresen

M. A.

Malleson

(2011). Testing the stability of crime patterns: Implications for theory and policy. Journal of Research in Crime & Delinquency, 48, 58-82.

Andresen

M. A.

Malleson

(2013a). Crime seasonality and its variations across space. Applied Geography, 43, 25-35.

Andresen

M. A.

Malleson

(2013b). Spatial heterogeneity in crime analysis. In Leitner

(Ed.), Crime modeling and mapping using geospatial technologies (pp. 3-23). New York, NY: Springer.

Barr

Pease

(1990). Crime placement, displacement and deflection. Crime and Justice: A Review of Research, 12, 277-318.

10.

Bowers

W. J.

Hirsch

J. H.

(1987). The impact of foot patrol staffing on crime and disorder in Boston: An unmet promise. American Journal of Police, 6(1), 17-44.

11.

Cornish

D. B.

Clarke

R. V.

(1986). Situational prevention, displacement of crime and rational choice theory. In Heal

Laycock

(Eds.), Situational crime prevention: From theory into practice (pp. 1-16). London, England: HMSO.

12.

Cornish

D. B.

Clarke

R. V.

(1989). Crime specialisation, crime displacement and rational choice theory. In Wegener

Lösel

Haish

(Eds.), Criminal behavior and the justice system: Psychological perspectives (pp. 102-117). New York, NY: Springer-Verlag.

13.

Eck

J. E.

(1993). The threat of crime displacement. Criminal Justice Abstracts, 25, 527-546.

14.

Esbensen

F.-A.

(1987). Foot patrols: Of what value? American Journal of Police, 6(1), 45-65.

15.

Hesseling

R. B. P.

(1994). Displacement: A review of the empirical literature. Crime Prevention Studies, 3, 197-230.

16.

Kelling

G. L.

Pate

Dieckman

Brown

C. E.

(1974). The Kansas City preventive patrol experiment: A technical report. Washington, DC: Police Foundation.

17.

Kelling

G. L.

Pate

Ferrara

Utne

Brown

C. E.

(1981). The Newark foot patrol experiment. Washington, DC: Police Foundation.

18.

Koper

C. S.

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

19.

Nasar

J. L.

Fisher

(1993). Hot spots of fear and crime: A multi-method investigation. Journal of Environmental Psychology, 13, 187-206.

20.

National Research Council. (2004). Fairness and effectiveness in policing: The evidence. Washington, DC: Committee to Law and Justice, Division of Behavioral and Social Sciences and Education.

21.

O’Brien

R. M.

(2007). A caution regarding rules of thumb for variance inflation factors. Quality & Quantity, 41, 673-690.

22.

Pate

A. M.

(1986). Experimenting with foot patrol: The Newark experience. In Rosenbaum

D. P.

(Ed.), Community crime prevention: Does it work? (pp. 137-56). Beverly Hills, CA: Sage.

23.

Ratcliffe

J. H.

(2004). Geocoding crime and a first estimate of a minimum acceptable hit rate. International Journal of Geographical Information Science, 18, 61-72.

24.

Ratcliffe

J. H.

Taniguchi

Groff

E. R.

Wood

J. D.

(2011). The Philadelphia Foot Patrol Experiment: A randomized controlled trial of police patrol effectiveness in violent crime hotspots. Criminology, 49, 795-831.

25.

Sherman

L. W.

Weisburd

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

26.

Trojanowicz

R. C.

(1986). Evaluating a neighborhood foot patrol program: The Flint, Michigan, Project. In Rosenbaum

D. P.

(Ed.), Community crime prevention: Does it work? (pp. 157-178). Beverly Hills, CA: Sage.

27.

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, 549-591.