Assessing Displacement

Determining displacement and diffusion should be central to understanding the impact of your POP project because the presence of extensive or malign displacement can overshadow any gains your response achieved. At the same time any level of diffusion stands to amplify response effects. In short, this means any determination of response effectiveness must directly measure the extent to which displacement and/or diffusion occurs. This has implications for how you design your assessment.

Designing your assessment to determine the presence of displacement or diffusion effects requires you to apply what you learned in the analysis of your project's displacement potential, particularly involving the types of displacement that might occur. For many projects spatial and temporal displacement will be most relevant; for others target, tactical, or crime type displacement may be more likely. A project could result in more than one type of displacement or diffusion. If you think this could be the case in your project, you need to assess the different types of displacement and diffusion that might reasonably occur.

Assessing most forms of displacement and diffusion requires the use of at least three different components:†

1. Response area – to determine any change in the crime or problem behavior as a result of the response. Many times this is a specific geographical area, but it can also be a system (such as transport systems), a group of facilities (such as convenience stores, banks, or retail businesses), or a specific time period (such as Friday or Saturday evenings).
2. Displacement/diffusion area – to determine whether the response resulted in the relocation (displacement) or reduction (diffusion) of crime or problem behavior elsewhere. Again, this may be a specific geographical area surrounding your response area (perhaps to another police jurisdiction), but it could also be another system (such as the relocation of problem behavior from a subway system to a bus system), a group of facilities (such as other convenience stores or retail businesses) or other targets, tactics, or time periods.
3. Control area – to determine whether areas that did not receive the response changed in ways similar to or different from those observed in the response area. The type of control area should be the same as or as similar as possible to that used for the response area. The control should also have not received the response, and there should be no possibility that it would have been influenced (or contaminated) by the response or by some other intervention. For instance, if the response area is a specific geographical location (such as a hot spot), the control area should be a similarly sized geographical area with similar traits. If the response is a sample of targets, the control should be some other similar group of targets not subjected to the response (secured convenience stores [response] versus unsecured convenience stores [control]). The key is that they are similar and unrelated in the sense that they are free from influencing one another.

† For a thorough primer on assessment design, see Problem-Solving Tools Guide No. 1, Assessing Responses to Problems: An Introductory Guide for Police Problem-Solvers, by John Eck; also see steps 46 to 53 of Clarke and Eck's (2005) Crime Analysis for Problem Solvers in 60 Small Steps.

You should consider three criteria when selecting suitable displacement/diffusion areas (see Table 3). The first criterion is that there is a logically specifiable reason to expect displacement or diffusion to that place, target, tactic, time, or crime type. For spatial displacement and diffusion this usually requires the area to be near the response area for many of the theoretical reasons discussed previously in this guide (e.g., awareness space and familiarity decay). There may, however, be times when it is logical for displacement to occur some distance away from the response area, and this should also be examined. Examples of this would be the relocation of problem behavior in a park to another park several blocks away or a street prostitution market that may shift to an existing prostitution market in another part of town. In such instances you need to assess the presence of displacement and diffusion in each location (e.g., the zone around the response area and the distant location).

For the other forms of displacement, the logical relationship between the former and the alternatives should guide you. For instance, in a project that gated alley ways to prevent residential burglaries accomplished through rear entry, did offenders change tactics and enter the residences from the front (e.g., tactical displacement)? Did offenders shift to burglarizing businesses rather than homes (e.g., target displacement)? Or was there an increase in nearby robberies (e.g., crime type displacement)?

The second criterion to consider in selecting the displacement/diffusion catchment area is size.32 The catchment area should not be too large relative to your response area as it could lead to the false conclusion that displacement did not occur. In other words, if the displacement/diffusion area is too large, the relocation of crime (displacement) might actually have occurred but the amount relative to the volume of crime in the (too large) catchment area may not be discernible. In this instance you would have concluded there was no displacement when in fact there was. The displacement/diffusion area should also not be too small so as to ensure that any relocation of crime is detectable. In practice it may be difficult to determine what exactly constitutes an appropriate-sized catchment area, and there is no right or wrong answer. You need to explore this with your crime analyst to see what works best. It may be useful to use multiple catchment areas, perhaps of different sizes such as one smaller and one larger.

Table 3: Criteria for Selecting Displacement/Diffusion Areas
Criterion	Purpose	Rationale
Logical Interrelatedness and/or Proximity	To ensure the observation is valid.	Selecting an area where displacement or diffusion would not logically occur may result in false conclusions that it did not occur when in fact it did.
Proportionate in Size	To increase the accuracy of your measures.	An area too small or too large could conceal the presence of displacement or diffusion if it occurs.
Contamination Free (Exclusivity)	To help ensure the observation is an effect of the response.	A contaminated area may lead to false determinations of displacement or diffusion when in fact it was the result of something else, such as another intervention.

The third criterion is that the displacement/diffusion area is free from any possible contamination, which could occur when catchment areas overlap with each other, with control areas, or with response areas other than the one you are assessing.33 If there is a possibility that changes in the displacement/diffusion area are the result of factors other than the response you are assessing, it will be difficult for you to claim the change was displacement or diffusion as it could be caused by something else. This means you need to ensure the displacement area does not overlap with other areas of activity. Beyond ensuring the areas do not overlap, in practice (e.g., applied settings) it is nearly impossible to control all outside influences. At the least, you should strive to identify what those possible influences are and understand how they might influence your measures of displacement and diffusion.

Figure 2 provides a schematic of an assessment designed to examine spatial displacement and diffusion. Here all three criteria are met. First, the displacement/diffusion area is logically related (in this case of spatial displacement, adjacent to) to the response area. Second, it is also proportional to the size of the response area. Third, it is free from any overlap with the response area or the comparison area, which could contaminate the measurements.

Figure 3 shows how this might look in a field setting with more than one displacement/ diffusion area. In this diagram the displacement/diffusion areas are both blocks surrounding the response area located at a public park (e.g., hot spot) and a somewhat distant public park. Again each criterion is satisfied: a) they are logically interrelated and/or proximate to one another, b) they are proportionate in size, and c) they are free from contamination (exclusivity).

Measuring Displacement

To determine the presence of displacement and diffusion effects, take measurements both before and after the implementation of the response in each area you identified in your assessment design (e.g., the response area, one or more displacement/diffusion areas, and the control area). Doing this allows you to find any change in the problem behavior after the implementation of the response relative to before it was in place, and to identify any displacement or diffusion effects. You can use a variety of measures as long as they are relevant to the focus of the responses. It is better to use several measures to assess improvements made by your project because it more fully represents the multiple dimensions your project stands to impact. Some types of measures you can use are:

• Crime counts or crime rates (i.e., number of crimes per 100,000 population)
• Levels of disorder (such as graffiti, loitering, or vandalism)
• Citizen fears, attitudes, or perceptions
• Observations or arrest reports of specific offenders (to determine whether offenders in one area are observed or arrested in the displacement/diffusion areas)
• Motor vehicle accidents
• Levels of incurred harm
• Crime severity
• Financial costs, among others.

At a minimum you need one before and one after measurement, but you may want to take several after measurements to gauge the impact of your project over time. As a rule, the longer the time period after the response the better, but most existing evaluation periods of crime prevention efforts tend to be less than two years after the response is applied.34

For many projects it is sufficient to simply compare before and after changes in your response area to those in your displacement/diffusion area. An increase in the measured level of crime or problem behavior in your displacement/diffusion area suggests that displacement occurred. If the increase is less than the reduction achieved in your response area, the project still achieved some success. If, however, the increase in the displacement/diffusion area is greater than the reduction in the response area, the project was unsuccessful because the displacement erased all of the response affects. A decrease in the level of crime or problem behavior in the displacement/diffusion area suggests that diffusion occurred and the benefits of the project achieved in the response area were augmented.

Table 4: Comparing Displacement and Diffusion Effects to Response Effects
	Before	After	Difference	Estimated Cumulative Net Effect
Response area	200	100	-100
Displacement/Diffusion Area 1	100	150	+50	-50
Displacement/Diffusion Area 2	75	50	-25	-75

The hypothetical data presented in Table 4 provides an illustration. Let's say the implementation of a problem-led response results in a 100-crime reduction in the targeted area. One displacement/diffusion area experiences a 50-crime increase (indicating displacement) while a second displacement/diffusion area experiences a 25-crime reduction (indicating diffusion). The 100-crime reduction achieved in the response area is cut in half to a net reduction of 50 crimes due to displacement in the first comparison area. Even though displacement occurred here, it was not enough to wash out the response effects. The 25-crime diffusion effect in the second comparison area adds to the net effect of the project resulting in an overall net reduction of 75 crimes.

For some projects, such as those that are large in scale or where there is much at stake, a more in-depth and precise determination of displacement and diffusion effects may be needed. When this is the case you may need to enlist the services of someone within your agency, such as a crime analyst, who has sufficient statistical and evaluation skills or an expert at your local university. Because evaluation of displacement and diffusion effects is unique to the crime sciences, general experts on evaluation methods may find the information presented in this guide useful. To assist you, your crime analyst, and/or your enlisted expert, a series of formulas are provided in Appendix B, which allow for precise empirical determinations of displacement and diffusion relative to any response effects. Appendix C provides the purpose, rationale, and interpretation of these formulas, and Appendix D gives an example of the use of these formulas from an actual project that was a finalist for the annual Herman Goldstein awards.