2018 POP Conference
November 5–7, 2018
Providence, Rhode Island

How is the concentration of risk among facilities calculated?

Once a satisfactory measure of the problematic events for a defined group of facilities has been obtained, the following six-step procedure can be used to determine whether the 80/20 rule applies.

  1. List the facilities alongside a count of the number of relevant events (e.g. thefts, assaults, calls for service) at each facility. Remember, it is important to verify that each facility on the list is of the type being investigated and that every crime attributed to each facility did in fact occur at that facility. (See Box 2 for a discussion of creating such a list and verifying its content.)
  2. Rank the facilities according to the number of events associated with each, from highest to lowest. (Table 1 is a list of hypothetical pubs along with the associated number of reported assaults.) Determine whether there is something that differentiates the facilities at the top of the list from those in the middle or at the bottom. For example, are the pubs at the bottom of the list popular evening entertainment spots for young people? Are they all located in a downtown entertainment district? Are they all owned by the same company? If so, then these similarities might account for the problem. If there are clear and obvious differences, then divide this list into meaningful categories, with sepa­rate ranked lists for each. Each category may pose a distinct problem. For each separate category, continue with Step 3. (For this example, assume that there are no important differences.)

    † Reproduced with permission from Clarke and Eck (2003)

  3. Calculate the percentage of events that each facility contributes. For example, in Table 1 there are a total of 121 assaults. The first pub, the White Hart, contributed 31 of these. So the White Hart accounts for 25.6 percent of the problem. The third column shows the percentage.
  4. Cumulate the percentages, starting with the riskiest facility. This shows the proportion of events associ­ated with each percentile (i.e., worst 10 percent, worst 20 percent, and so on, up to 100 percent). The fourth column shows what is called the cumulative percentage; that is, the percentages from the third column are added starting with the White Hart and going down.
  5. Calculate the proportion of the facilities that each single facility repre­sents. In our example, there are 30 pubs, so each represents 3.3 percent of the pubs. Then cumulate these percentages in the same direction as in Step 4 (top down in column 5).
  6. Compare the cumulative percentage of facilities (column 5) to the cumula­tive percentage of events (column 4). This shows how much the riskiest facilities contribute to the overall problem.

Box 2: Defining and Listing Facilities

In order to analyze crime concentrations, it is first necessary to define the type of facility to be examined; only then is it possible to create a list of facilities that meets that the definition. Ideally, all places that fit the definition and that are in the area of study will be on the list once and only once. In addition, facilities that do not fit the definition will not be on the list. The further the list departs from this ideal, the more likely it is that the results will be misleading.

Identifying all facilities of a particular type in any given area can be troublesome: not only can it sometimes be difficult to develop an appropriate working definition of the type of facility at issue, but problems can also arise in regard to the data management practices of relevant public and private agencies. 

Here is an example of creating a list of facilities that illustrates these points. A research team at the University of Cincinnati, Ohio wanted to determine why a few bars had numerous violent incidents, whereas most of the others had none or only a very few. To do this, they needed a definition of “bar” and a list of facilities that met this definition.

Researchers defined “bar” as a place that met four conditions: (1) it had to be open to the general public, rather than restricted to members or rented out to private parties; (2) it had to serve alcohol for onsite consumption; (3) some patrons had to come to the place for the primary purpose of consuming alcohol; and (4) there had to be a designated physical area within the place that served as a drinking area. Locations that did not meet all four conditions were excluded from the study.

To obtain a list of locations meeting this definition, researchers began by consulting records from the Ohio Division of Liquor Control. These records showed that 633 places within the city limits were licensed to serve hard liquor. Based upon their personal knowledge, researchers were able to exclude a number of locations from consideration, reducing the list to 391 possible bars. To isolate the real bars, researchers then compared the remaining locations to the most recent bar guide in a local weekly tabloid that catered to young adults, which contained both a brief written description of the locations and numerous commercial advertisements. The tabloid information revealed that at least 198 of the 391 places fit the definition used. The tabloid list was incomplete, however, as there were an unknown number of city bars that were not reviewed by the tabloid staff. A check of the online Yellow pages verified several more bars. Private fraternal organizations were eliminated from consideration because they were not open to the general public. For most of the remaining places, researchers phoned or visited the sites, examining the physical locations and interviewing owners and employees. Onsite visits revealed several restaurants had areas that looked like bars, but these were eventually eliminated from consideration when it became clear from interviews that they were more decorative than functional or that they were used for other purposes (e.g., to hold carryout orders for customer pickup or to provide overflow seating where customers could eat).  Ultimately, researchers identified 264 facilities that fit the definition of bar. These then became the subjects of the study.

Table 1: The Distribution of 121 Assaults in 30 Pubs

  No. of Assaults % of Assaults Cumulative % Assaults Cumulative % Pubs
White Hart 31 25.6 25.6 3.3
Union 17 14.0 39.7 6.7
Feathers 13 10.7 50.4 10.0
Wellington 11 9.1 59.5 13.3
Black Prince 8 6.6 66.1 16.7
Angel 7 5.8 71.9 20.0
George & Dragon 6 5.0 76.9 23.3
Cross Keys 6 5.0 81.8 26.7
Saracen's Head 4 3.3 85.1 30.0
White Bear 4 3.3 88.4 33.3
Mason's Arms 3 2.5 90.9 36.7
Cock 3 2.5 93.4 40.0
Badger 3 2.5 95.9 43.3
Hare & Hounds 1 0.8 96.7 46.7
Red Lion 1 0.8 97.5 50.0
Royal Oak 1 0.8 98.3 53.3
George 1 0.8 99.2 56.7
Cross Hands 1 0.8 100 60.0
Rose & Crown 0 0 100 63.3
King’s Arms 0 0 100 66.7
Star 0 0 100 70.0
Mitre 0 0 100 73.3
Dog and Fox 0 0 100 76.7
Griffin 0 0 100 80.0
Plough 0 0 100 83.3
Queen’s Head 0 0 100 86.7
White Horse 0 0 100 90.0
Bull 0 0 100 93.3
Swan 0 0 100 96.7
Black Bear 0 0 100 100