Skip to main content
SearchLoginLogin or Signup

Mapping crime prevention: What we do and where we need to go

Andresen, M.A. (2017). Mapping crime prevention: What we do and where we need to go. In B. Leclerc & E.U. Savona (Eds.), Crime prevention in the 21st century (pp. 113 – 126). New York, NY: Springer.

Published onJan 01, 2017
Mapping crime prevention: What we do and where we need to go

Abstract Crime mapping is an activity that dates back to the early 19th century in the context of understanding crime in France. Crime mapping has become an important component in crime prevention because it can help in the identification of crime problems and the evaluation of crime prevention initiatives. In this chapter, I cover the state of the art in mapping crime prevention as well as stating where it needs to go. The current primary methodology, kernel density mapping, is explained, followed by a discussion of the caveats of this technique. I then discuss local spatial analysis and its importance for crime prevention, emphasizing that these methods are easily applied. Lastly, the importance of temporal scales are discussed, showing how space and time are interconnected and one cannot be done without considering the other. This chapter concludes with a statement that complicated spatial analyses are not necessary for mapping crime prevention. Rather, the analyst can be sophisticated without being complex.

Keywords Crime prevention, Crime mapping, local spatial analysis, spatial-temporal


Crime prevention is an activity that can take many forms. Crime prevention initiatives can be considered developmental, community-based, rehabilitation, policing, deterrence-punishment, and through modifications of criminal opportunities often referred to as crime prevention through environmental design—see Tilley (2005) and other chapters in this volume for discussions of the various forms of crime prevention. Mapping crime prevention is most often associated with the latter of these types of crime prevention initiatives, but may also be part of policing and community-based initiatives. Most of the discussions below are most closely aligned with situational crime prevention (Clarke 1980, 1983, 1997, 2012).

Mapping for the purposes of crime prevention is instructive because it shows where the criminal event problem is, but it is also instructive for the purposes of evaluating various crime prevention initiatives. In this chapter, I will briefly cover the state of the art in mapping for crime prevention and then move into some considerations for the future of mapping crime prevention. I hope to show that adding sophistication to mapping crime prevention does not necessarily increase its complexity, but adds a significant amount of information that is instructive for understanding the criminal event problem at hand, identifying the appropriate crime prevention initiative, and performing an evaluation of crime prevention initiatives that can inform future crime prevention activities.

State of the art: what we do to map crime prevention

Though one may find a number of different methods to map crime for prevention purposes, the most common forms of doing so are dot maps and (kernel) density maps (Santos 2013). Dot maps can take two forms: 1) one dot on the map for every (criminal) event; and 2) graduated dots such that as the number of (criminal) events at each location increases the dot gets larger. This technique can be instructive when there are not many dots to place on the map but very quickly becomes difficult to interpret as the number of criminal events increases. Kernel density maps are based on dots on a map but create a surface representing the intensity of dots at any location in the study area. There are other basic and advanced techniques available, but are seldom used when mapping for crime prevention (Santos 2013).

The kernel density hot spot map is calculated as follows, and shown in Figure X.1. Though the details of how the calculation is made can vary, the general formation begins by placing a grid over the study area. For each cell in that grid a value is calculated that represents the intensity of criminal event activity. From each cell, a circle is drawn (the diameter of this circle is called the bandwidth) and the criminal events (points) that fall within that circle are counted and used to calculate the kernel, or the “height” of the hot spot. Generally speaking, the more criminal events within the circle, the higher the kernel will be for that cell. This is repeated for every cell within the grid placed over the study area that can then be used to generate a map showing hot spots: hot spots are usually identified by the presence of a number of cells that have high kernel values, as can be seen in Figure X.2a.

<Insert Figure X.1 About Here>

Considerations for mapping crime prevention

No discussion of the consideration for the mapping of anything, including crime prevention, would be complete without a brief discussion of the ecological/atomistic fallacy and the modifiable areal unit problem. The ecological/atomistic fallacy refers to the improper inference that can be made when analyzing spatial data, including maps, and the modifiable areal unit problem refers to how the results of an analysis can change when you change the spatial unit of analysis. We will discuss each, briefly, in turn.

The ecological/atomistic fallacy was formally identified by Robinson (1950) when he identified the conditions that must hold in order to claim that relationships found at one unit of analysis (the neighborhood, for example) would be the same for another unit of analysis (the individual, for example). He then went on to state how there is no evidence for these conditions to hold true in any analysis—Openshaw (1984a) showed that the differences between individual and area-level relationships could not be known, a priori. As such, the ecological fallacy occurs when a researcher or practitioner states that what is true of the whole is also true of all its parts; the atomistic fallacy occurs in the opposite direction such that what is true of the parts is also true of the whole. Of course, the fallacy is rarely, if ever, committed so overtly. Rather, an analysis (for crime prevention, for example) is undertaken using census level data (dissemination areas of census block groups) and then the researcher or practitioner attributes the relationships found to the individuals who live in that census area. This is actually an easy fallacy to commit, but it is also an easy fallacy to avoid.

The modifiable areal unit problem (MAUP) also emerges from the aggregation of data. In the census, for example, data are aggregated from individuals to census tracts and other spatial units of analysis for subsequent analysis. The trouble is that the aggregations of these data are usually deemed arbitrary in the sense that they do not represent “natural units” of analysis. This can emerge in two primary ways: the scale problem and the zoning problem. The scale problem emerges from analyzing different sized spatial units of analysis: census block groups, census tracts, and neighborhoods, for example. The zoning problem emerges when the size and dimension of the spatial units does not change but they are placed differently on the map—think of dragging census tract boundaries over by one block, or more. In both cases, different sets of individuals are aggregated into spatial units of analysis. The problem with these modifiable areas is that the results of any analysis can change, often unpredictably. Openshaw (1984b) showed that any desired results could be obtained through the modification of areal units, and Fotheringham and Wong (1991) showed that statistical results were not reliable and that their results were “depressing”—Wooldredge (2002) did have more promising results in a spatial crime analysis context.

This does not mean that mapping for crime prevention is inherently problematic. But it does mean that when undertaking mapping for crime prevention that the research and/or practitioner must be aware that what they are doing and any results that they find may be an artefact of the spatial units of analysis that they are using. Consequently, the use of at least two spatial units of analysis should be undertaken to (hopefully) show that results are not sensitive to the spatial unit used in the analysis. It would be recommended to start with the smallest unit of analysis available (the discrete criminal event point, if possible) and then aggregate, as necessary. This is important because neither the researcher nor the practitioner would want to improperly identify a location for a crime prevention initiative.

The limitations of kernel density mapping

As discussed above, kernel density mapping is one of the most common forms of crime mapping that is used in crime prevention. Though instructive for its use to identify crime hot spots, like any analytical technique, it has its limitations. Two of these limitations will be discussed here.

First, kernel density mapping, in its most common form, uses only criminal event data in order to identify hot spots. But what is a hot spot? John Eck and colleagues define as hot spot as “an area that has a greater than average number of criminal or disorder events, or an area where people have a higher than average risk of victimization” (Eck et al. 2005, p. 2). This definition has two components to it: 1) greater than average number of criminal events, and 2) higher than average risk. These can be two very different phenomena.

If the researcher or practitioner is concerned with the prevention of the volume of crime, then a kernel density map that only uses criminal event data is just fine. However, because criminal events occur when motivated offenders and suitable targets converge in time and space without the presence of capable guardians (Cohen and Felson 1979), where there are more convergences there are more criminal events. As such, just because there are a lot of criminal events does not mean that a person is actually at a high risk of victimization. This is quite common in central business districts and has been known for decades (Schmid 1960a, 1960b). As a consequence of this, if the intention of a crime prevention initiative is to reduce the risk of victimization then mapping only criminal events may be problematic. This is why we calculate crime rates: so we can compare rates across time and space when we know the population at risk of victimization changes.

In order to control for this in kernel density mapping, if it is a concern, a dual kernel must be calculated. This is a very simple technique to perform with the only complication being the need for population at risk data that varies across space at a spatial resolution that is appropriate for the crime prevention initiative. This is relatively simple when mapping for crime prevention at the level of the city because reliable data are available for the resident population at the dissemination area (census block group) that will give a reasonable approximation of the population at risk, but is not always a good representation of the population at risk—see Andresen and Jenion (2010) for a critique of such data. An example of how this impacts the hot spot map is shown in Figure X.2, for violent crime in Vancouver, British Columbia, Canada, using the ambient population that is more appropriate for considering violent crime but is not as readily available as census data.

We can see from Figure X.2 that using a single-kernel (criminal event data only) violent crime has a hot spot that is located within the central business district (the peninsula at the northern portion of the city) that exhibits a distance decay pattern as one moves away from the central business district. Using this information may lead the researcher or practitioner to target the high-density area for a crime prevention initiative. This is not incorrect because that is where the volume of violent crime is occurring, but when considering the dual kernel map that hot spot all but disappears. The location with the highest density is still in the central business district, but no longer in the center of the peninsula; this high-density area is now on the edge of the central business district itself and closer to skid row. Crime prevention initiatives imposed in the center of the central business district would be very different from those being implemented on the edge of skid row in any given city.

<Insert Figure X.2 About Here>

The second issue regarding kernel density mapping that needs to be made clear is that the method itself is not technically appropriate for criminal event data. This is because kernel density mapping is a surface generation technique for geographically continuous data. An example of geographically continuous data is temperature: temperature is everywhere but it is expensive to measure temperature everywhere. So, we measure temperature in a number of discrete locations (this would be determined based of the appropriate sampling model for the study) and then kernel density mapping or some other technique would be used to re-create the continuous surface.

The question we need to ask is whether or not this is a useful technique for identifying crime prevention initiative locations. If we are mapping for the purposes of crime prevention we must necessarily properly identify the correct locations to implement those initiatives. Based on the discussion above regarding how the kernel density map is created, it should be clear that a value can be calculated at a location that does not have (and never has had) any criminal events because criminal events occurred within the bandwidth of that location. Additionally, because of the bandwidth criminal events may be counted more than once such that any hot spots will appear to be larger than they really are. Indeed, this is the nature of kernel density mapping: all of the data are smoothed over a surface to generate an easy to interpret map. However, if the information that is generated in the map is not correct, or at least misleading, its value for the prevention of crime must be questioned. A problem that may emerge because of this is the inefficient use of limited resources for crime prevention.

This discussion should not leave the reader convinced that kernel density mapping is not instructive. On the contrary, such a mapping technique is quick, simple, and easy to implement and may provide very useful information to the researcher and the practitioner. It is important, however, to know the limitations of the method in order to be able to interpret the output (maps) with caution. One way to do this is to employ more than one method of mapping crime for the purposes of identifying the potential location for crime prevention initiatives, or evaluating those crime prevention initiatives.

The importance of local spatial analysis for mapping crime prevention: the importance of where

There are two aspects of local spatial analysis that will be discussed here in the context of mapping crime prevention. First, when considering the possibility of crime displacement an example of “local” versus “global” analyses will be provided—crime displacement is a well-known and heavily researched area in the field of crime prevention (Barr and Pease 1990; Eck 1993; Hesseling 1994; Weisburd et al. 2006). And second, an example will be shown regarding changes in spatial patterns of crime that uses both kernel density mapping and a local spatial analytical technique.

The first example with regard to crime displacement is in the context of a police foot patrol that took place in North Vancouver, British Columbia, Canada. Andresen and Lau (2014) undertook an evaluation of this police foot patrol initiative and found that there was a drop in the calls for police service of approximately 17 percent, with most of the reduction in criminal events revolving around mischief and commercial burglary. These authors considered the primary patrol area and a surrounding area to investigate the possibility of crime displacement from the police foot patrol. The surrounding area was considered as a whole to measure crime displacement. However, the surrounding area was relatively large such that it may be difficult to identify statistically significant increases in the number of criminal events; the primary patrol area was 0.90 square kilometers, whereas the potential displacement area was approximately 3 square kilometers. In order to address this concern, Andresen and Malleson (2014) performed a local analysis of crime displacement on this police foot patrol initiative and considered 32 units of analysis for crime displacement instead of just 1 unit of analysis.

Andresen and Malleson (2014) used a locally-based spatial point pattern test developed by Andresen (2009) to test the similarity of the spatial patterns of criminal events before and after the police foot patrol as well as identify the locations in which the concentrations of crime increased or decreased. As such, this test allowed the researchers to find out which of the concentrations of criminal events changed even though it is known that criminal events decreased in both the primary patrol area and the potential displacement area.

In their analysis, Andresen and Malleson (2014) found that the spatial crime patterns had changed as a result of the police foot patrol. This, of course, is an expected result because of a police presence in approximately one-quarter of the study area—commercial burglary did not exhibit much spatial pattern change but the primary patrol area is where the vast majority of the commercial land use area is located so not much change could occur. When considering all crime types aggregated together, little information could be obtained from their results. Some places exhibited increases in the concentrations of criminal events and other exhibited decreases; moreover, these increases and decreases occurred in both the primary patrol area and the potential displacement area. The primary result of interest when considering all criminal events is that there appeared to be more increases in the concentrations of criminal events at the border areas of the primary patrol area where less time will be spent by those who undertake the police foot patrol. In the context of mischief, the crime type that exhibited the most significant decrease in activity, there was a moderate indication that the spatial pattern of this crime type shifted away from the primary patrol area into a small number of areas.

The importance of thinking local should be self-evident from these results. Global evaluation of the police foot patrol indicates that criminal events decreased with no evidence for crime displacement. However, the spatial pattern of the remaining number of criminal events had shifted to particular areas. As such, mapping crime prevention at the local level can allow the researcher and/or practitioner to identify next steps for the prevention of further criminal activity.

The second example is also in the context of the police foot patrol in North Vancouver, Canada. In this example, Andresen (2015) shows how the use of the more standard mapping technique, kernel density mapping, does not provide much insight regarding any changes in the spatial patterns of criminal events, but another local spatial statistical technique (local Moran’s I) does provide some insight. The maps of these two techniques are shown in Figure X.3 and represent mischief, the crime type most impacted by the police foot patrol—the results are similar for other crime types that Andresen (2015) analyzed.

Comparing Figures X.3a and X.3b, some new information can be obtained regarding any change in the spatial pattern of criminal events. There is a hot spot in both maps at the southern portion of the study area (contained within the primary patrol area) with moderate density locations shifting around slightly. If anything is to be identified, it is the emergence of a moderate intensity “hot” spot in the west side of the study area and the expansion of the hot spot in the southern portion of the study area; this latter expansion is relative because mischief decreased significantly because of the police foot patrol initiative. Comparing Figures X.3c and X.3d, there also is not a lot of change that occurs. However, the information is far more specific in the areas of potential concern. The location in the western section of the study area is specifically identified as a High-Low cluster, another High-Low cluster has emerged in the northern area of the study area, and the criminal event cluster area in the southern portion of the primary patrol area has changed its classification; formerly this are contained two Low-High crime clusters but now contains a High-High crime cluster. This is far more specific, and curious, information provided than the kernel density maps and may be because the kernel density maps smooth out the criminal events, as discussed above.

<Insert Figure X.3 About Here>

This example shows that the use of one crime mapping technique, particularly in the context of evaluating a crime prevention initiative, is probably not a good idea. This does not show that one technique is wrong and one technique is correct, but that more than one method of analysis should be undertaken to investigate/evaluate crime prevention. If there is consistency between the multiple methods then the researcher and/or practitioner can have confidence in the results. If not, caution must be undertaken before further action is taken.

Sophistication without complexity

In their seminal article that outlined the fundamental elements of a criminal event, Cohen and Felson (1979) stated that a crime occurs when a motivated offender, a suitable target, and the lack of a capable guardian converge in space and time. As such, understanding space (where) and time (when) are critical for understanding crime patterns and, hence, crime prevention. Given that mapping is fundamentally spatial (it is temporal as well as we shall see below), mapping crime prevention must pay particular attention to where and when crime occurs.

Similar to the use of local analysis, just knowing where criminal events occur may be misleading, or at least limited, when trying to understand the crime problem. As will be discussed further below, with an example, knowing when criminal events occur is particularly important. This may seem to be an odd statement in the context of mapping for crime prevention but it is far more important than usually thought.

As outlined by Hirschfield (2005), it is not just where and when criminal events occur that matters, but the combinations of the two: in those places where criminal events occur, when are they occurring, and at those times when criminal events are occurring, where are they occurring? These two different questions may lead to very different interventions. For example, if a location of criminal activity is to be targeted because it is a “known” problem, in order to get the best results from a crime prevention intervention that intervention should be developed and implemented considering the timing of those criminal events. If assaults are the problem and there is a drinking establishment in the area, the timing of most of the assaults will probably be Friday and Saturday evenings. As such, a crime prevention initiative that (intentionally or unintentionally) targets a time frame that considers when teenagers are out of school (weekdays, 3pm – 6pm) may not be effective. Mapping different time frames can help with this identification.

It is also possible that a high volume of criminal activity occurs on particular days and at particular times. If this is the case, all criminal events (or those of interest by the researcher and/or practitioner) that occur within that time range should be selected and mapped for prevention purposes. It is quite possible that the locations that emerge will be different than where they would be expected. Perhaps not, but this needs to be investigated properly to prevent the misuse of scarce resources in crime prevention. Such different forms of analyses provide much more sophistication to the mapping of crime prevention without becoming complicated. More maps will have to be generated, but this extra information will be most instructive for the prevention of crime.

The importance of temporal aggregation for mapping crime prevention: the importance of when

As stated above, routine activity theory highlights the importance of the convergence in space and time for understanding criminal events. Of course, when Cohen and Felson (1979) were discussing these concepts they were speaking of very specific moments when motivated offenders and suitable targets converged because they were interested in understanding specific criminal events, and the aggregation of those events to understand crime patterns over time. However, we must also consider time when mapping crime, for crime prevention or not. Most often, this temporal consideration involves a time frame of criminal events to place on a map: do we map crime for a particular year, particular season, or a particular month? This choice obviously depends on the context of the investigation, but it does have some implications for mapping crime prevention that need to be explicated.

The literature that most commonly investigates these implications is that of seasonality and crime. Such investigations have gone back to early 19th century France (Quetelet 1842) and find that a seasonal pattern is quite often present—see Andresen and Malleson (2013) for a discussion of the different studies that do and do not exhibit seasonality in their criminal event data. If seasonality is present, there is usually a peak of some degree during the summer months that is explained using routine activity theory: during the summer months the weather is nicer (warmer and drier), children are out of school, and people take vacations that all leads to an increase in motivated offenders and suitable targets converging in time and space with the lack of capable guardians (Andresen and Malleson 2013). Because of this phenomenon, anyone evaluating a crime prevention initiative must consider the timing of that initiative because measuring the presence of before and after effects could simply be because of changing of the seasons. But there is another consideration when mapping crime prevention.

The other consideration is that crime patterns may change at different times of the year, or even different days of the week. As such, when mapping where criminal events are occurring, the researcher or practitioner must also consider when criminal events are occurring if s/he wishes to know the appropriate place (and time) to implement a crime prevention initiative, as discussed above. Unfortunately, there is very little research in this area to call on in order to show its importance. Overall, this research shows that during the peak season(s) for criminal events, there is a particular spatial pattern that emerges: criminal events appear to increase disproportionately in areas/neighborhoods that are of low socio-economic status (Breetzke and Cohn 2012; Ceccato 2005; Harries and Stadler 1983; Harries et al. 1984). Though instructive, this may not have any particular implications for mapping crime prevention because criminal events are most often over-represented in these areas anyway.

In an investigation of changing spatial patterns for different seasons and a variety of crime types, Andresen and Malleson (2013) found that the spatial patterns of crime were quite dissimilar at different times of the year—only sexual assault and robbery at small spatial units of analysis (dissemination areas, equivalent to the census block group) had little change in their spatial patterns from season to season. As shown in Figure X.4, criminal events increased during the summer in very predictable places: the beach, the central business area, large parks, large shopping centers, and the summer fair (PNE).

<Insert Figure X.4 About Here>

This analysis shows that if crime prevention is truly the goal during a crime prevention initiative—it could be community building, for example—the time frame chosen to identify problem areas is very important. Do you want to decrease criminal events based on criminal events that occur over the course of the year? Or do you want to focus on criminal events that occur during a particular time of year because that is when most of the criminal events occur? It should be clear that the choice made will impact the effectiveness of the crime prevention initiative, because if the “wrong” or “inappropriate” criminal events are mapped, that crime prevention initiative may be implemented incorrectly because the prevention of crime is highly situational (Clarke 1980, 1983, 1997, 2012).

The example shown here has used a rather coarse temporal unit of analysis, the season. Criminal events are also known to have different temporal patterns by day of the week (more assaults during the weekend, for example) and even within the day (residential burglaries tend to occur during the day and commercial burglaries during the night) (Andresen 2014). If spatial patterns of crime vary based on these temporal units of analysis, the implications for crime prevention become even stronger. If we are to map for the purposes of crime prevention we must map the temporal dimension properly as well.

Concluding thoughts

This chapter has considered the importance of mapping for crime prevention. Rather than simply discussing how to undertake that mapping within any particular software program, some considerations for what should be done have been discussed as well as some cautions for the various techniques was discussed.

I have tried to cover aspects of mapping for crime prevention that do not involve any specialist knowledge. Of course some training in mapping software is necessary, but no specialist training is required. All of these techniques are readily available in packaged software, most of which in the most common ArcGIS software program.

The most important aspect of these discussions to consider is that one does not have to get into complicated analyses when mapping for crime prevention. However, more sophisticated analyses will prove to be instructive to understand the local crime problem as well as provide more precise guidance as how to address that problem with a crime prevention initiative. Mapping for crime prevention is a fruitful endeavour that can very easily aid in the effective use of scarce resources for the prevention of crime.


Andresen, M.A. (2009). Testing for similarity in area-based spatial patterns: a

nonparametric Monte Carlo approach. Applied Geography, 29(3), 333 - 345.

Andresen, M. A. (2014). Environmental criminology: Evolution, theory, and practice.

New York, NY: Routledge.

Andresen, M. A. (2015). Identifying changes in spatial patterns from police

interventions: the importance of multiple methods of analysis. Police Practice

and Research, in press.

Andresen, M. A., & Jenion, G. W. (2010). Ambient populations and the calculation of

crime rates and risk. Security Journal, 23(2), 114 - 133.

Andresen, M. A., & Lau, K. C. Y. (2014). An evaluation of police foot patrol in Lower

Lonsdale. Police Practice and Research, 15(6), 476 - 489.

Andresen, M. A. and Malleson, N. (2013). Crime seasonality and its variations

across space. Applied Geography, 43, 25 - 35.

Andresen, M. A., & Malleson, N. (2014). Police foot patrol and crime displacement: a

local analysis. Journal of Contemporary Criminal Justice, 30(2), 186 - 199.

Barr, R., & Pease, K. (1990). Crime placement, displacement and deflection. Crime

and Justice: A Review of Research, 12, 277 – 318.

Breetzke, G. D. and Cohn, E. G. (2012). Seasonal assault and neighborhood

deprivation in South Africa: Some preliminary findings. Environment &

Behavior, 44(5), 641 – 667.

Ceccato, V. (2005). Homicide in San Paulo, Brazil: Assessing spatial-temporal and

weather variations. Journal of Environmental Psychology, 25(3), 307 – 321.

Clarke, R. V. G. (1980). Situational crime prevention: Theory and practice. British

Journal of Criminology, 20(2), 136 – 147.

Clarke, R. V. (1983). Situational crime prevention: Its theoretical basis and practical

scope. Crime and Justice: An Annual Review of Research, 4, 225 – 256.

Clarke, R. V. (1997). Situational Crime Prevention: Successful case studies (2nd ed.).

Monsey, NY: Criminal Justice Press.

Clarke, R. V. (2012). Opportunity makes the thief. Really? And so what? Crime

Science, 1, Article 3.

Cohen, L. E. and Felson, M. (1979). Social change and crime rate trends: A routine

activity approach. American Sociological Review, 44(4), 588 – 608.

Eck, J. E. (1993). The threat of crime displacement. Criminal Justice Abstracts, 25(3),

527 – 546.

Eck, J. E., Chainey, S., Cameron, J. G., Leitner, M. and Wilson, R. E. (2005). Mapping

crime: Understanding hot spots. Washington, DC: National Institute of Justice,

Office of Justice Programs, United States Department of ustice.

Fotheringham, A. S. and Wong, D. W. S. (1991). The modifiable areal unit problem in

multivariate statistical analysis. Environment and Planning A, 23(7), 1025 –


Harries, K. D. and Stadler, S. (1983). Determinism revisited: Assault and heat stress

in Dallas. Environment & Behavior, 15(2), 235 – 256.

Harries, K. D., Stadler, S. and Zdorkowski, R. (1984). Seasonality and assault:

Explorations in inter-neighborhood variation, Dallas 1980. Annals of the

Association of American Geographers, 74(4), 590 – 604.

Hesseling, R. B. P. (1994). Displacement: A review of the empirical literature. Crime

Prevention Studies, 3, 197 – 230.

Hirschfield, A. (2005). Analysis for intervention. In N. Tilley (Ed.), Handbook of crime

prevention and community safety. Cullompton, UK: Willan Publishing, 629 –


Openshaw, S. (1984a). Ecological fallacies and the analysis of areal census data.

Environment and Planning A, 16(1), 17 – 31.

Openshaw, S. (1984b). The modifiable areal unit problem. CATMOG (Concepts and

Techniques in Modern Geography) 38. Norwich: Geo Books.

Quetelet, L. A. J. (1842). A Treatise on man and the development of his faculties.

Edinburgh: W. and R. Chambers.

Santos, R. B. (2013). Crime analysis with crime mapping (3rd ed.). Thousand Oaks, CA:

Sage Publications.

Schmid, C. F. (1960a). Urban crime areas: Part I. American Sociological Review,

25(4), 527 – 542.

Schmid, C. F. (1960b). Urban crime areas: Part II. American Sociological Review,

25(5), 655 – 678.

Weisburd, D., Wyckoff, L. A., Ready, J., Eck, J. E., Hinkle, J. C., & Gajewski, F. (2006).

Does crime just move around the corner? A controlled study of spatial

displacement and diffusion of crime control benefits. Criminology, 44(3), 549

– 591.

Wooldredge, J. (2002). Examining the (ir)relevance of aggregation bias for

multilevel studies of neighborhoods and crime with an example of comparing

census tracts to official neighborhoods in Cincinnati. Criminology, 40(3), 681

– 709.

Figure X.1. Kernel density calculations

Source. Andresen (2014).

Figure X.2. Kernel Density, Single Versus Dual Kernel

a) Single kernel

b) Dual kernel

Source. Andresen (2014).

Figure X.3. Kernel Density Versus LISA

Kernel density maps

a) Before

b) After

Local indicator of spatial association: Local Moran’s I

c) Before

d) After

Source. Adapted from Andresen (2015).

Figure X.4. Spatial changes in crime patterns, summer versus yearly aggregate

  1. City of Vancouver, Major Points of Interest

  1. Spatial changes in crime patterns, all crimes

Source. Adapted from Andresen and Malleson (2013).

No comments here
Why not start the discussion?