Understanding Your Local Repeat Victimization Pattern

The information in this guide is only a generalized description of repeat victimization. Because repeat victimization varies in different locations and by the type of crime problem, you must combine the basic concepts of repeat victimization with a more specific understanding of your local problem. Analyzing the local problem carefully will help you design a more effective response strategy.

Estimating the amount and distribution of repeat victimization is necessary to develop an effective response.

Selecting the Problem

Depending upon the crime problem being examined-and its presumed relationship with other related problems-you may wish to examine a cluster of problems. For example, convenience stores victimized by robbery may also be frequent victims of burglary, shoplifting, or larceny. The choice of problems to be examined should be based on practical reasoning.

Although it may be necessary to begin with a broad problem because of the type of data being used, every effort should be made to reduce the problem to one with more similarities than differences and avoid an overly broad problem. For example, vandalism includes graffiti and destruction of property, but each should be examined separately.

For the problem being examined, determine the appropriate "unit of analysis" by deciding if the relevant "victim" is an individual or household, an address, a business, or a group of victims such as convenience stores or an individual chain store.

Selecting Data

Existing police data, such as crime reports, are most often used to document repeat victimization. In selecting data, attention should be given to:

• Determining the appropriate time frame. For most crime problems, it is best to use at least three calendar years of data. You can then define repeat victims as persons or places victimized more than once during the three years. If you use only a single year of data, an address burglarized in December will not be identified as a repeat if the next burglary at the address occurred in January. Shorter time periods of data will generally underestimate the amount of repeat victimization.
• Determining data sufficiency. As a general rule, it can be easier to detect repeat victimization for offenses that are common than for offenses that are rare. While the magnitude of problems varies from one jurisdiction to another, it may be useful to have 100 offenses or more per year to detect repeats. While one may find repeats even among 10 bank robberies per year in a small town, analyzing a larger number of robberies will produce more reliable findings. When there are few offenses of one type, you can increase the amount of data by incorporating multiple years, up to ten years or more, or by broadening the geographic scope of analysis, such as incorporating offenses that occur in neighboring jurisdictions.
• Choosing a denominator. To best calculate repeat victimization, it is useful to know the population or number of possible victims of a similar type. For example, if you want to determine the number of repeat robberies of convenience stores, it is helpful to know the number of convenience stores. For demographic information, such as households, such data can be found in the U.S. Census; for other denominators, sources such as business licenses or tax records can be used to determine the population.
• Identifying key variables. Data should routinely include victim name, address (including street number, building name or number, apartment or suite number, and direction), and date and time of the offense. If the victim is a business, the business name and the type of business should be examined. Variables examined should also include the outcome such as the type or value of property taken or damaged, whether the offense was attempted or completed, the amount of force used in the offense, the level of harm, and so forth. Additional variables may be useful, and you can identify these by reading the literature on the specific type of problem being considered.
• Determining data limitations. Police data routinely have limitations, including underreporting, delayed reporting, data errors, and imprecise address information. When selecting data, you should identify the limitations of the data you will use, think through the implications of these limitations, and determine whether the data can be improved with a reasonable amount of effort. Detecting the limitations of different data and techniques for improving data quality are discussed in appendices A and B.

† This task may suggest organizational changes that can be made to improve data quality. For example, agencies may modify offense reports, change nature classifications used in recording dispatched calls or standardize recording of victim names.

Analysis Tasks

Basic analysis for repeat victimization is straightforward and rarely requires any complex statistical procedures. Data are typically contained in a spreadsheet with a range of descriptive variables; the data are then sorted based on a particular variable so all related victims are grouped together.

Mapping locations. Many types of revictimization can initially be detected with point maps. These maps place dots or points on the locations of offenses, calls for service, and arrests for crimes such as burglaries, robberies, and assaults. Points that are scaled in size to reflect the number of incidents occurring at individual addresses are useful. Point maps are less useful in areas of high-rise apartments or office buildings where the population is dense. Point maps are also less useful in rural areas or areas with few addresses such as parks, farms, and large parking lots, unless global positioning system (GPS) coordinates have been established for these locations.

Sorting offense data by address. Mapping is essentially a method of putting a chart or table onto a spatial layer. We can accomplish the same task by sorting data by address-by sorting the street names and numbers, we create a table such as Figure 5. This table does not include the victim name, but the data would enable you to sort and then count the number of offenses occurring at each location.

Figure 5

† A tool for conducting this analysis is included on the popcenter.org website.

Sample Table for Sorting Addresses

Police Beat	Census Tract	Census Block	Street Number	Street Direction	Street Name	Street Type
12	12500	4005	645		H	ST
12	12500	4005	645		H	ST
31	13412	1004	4420		BONITA	RD
14	12303	2013	295		E	ST
13	13000	1000	352		H	ST
13	12700	2000	444		3RD	AV
31	13409	1000	386	E	H	ST
31	13409	1000	358	E	H	ST
31	13413	2003	1020		TIERRA DEL REY
31	13409	1000	354	E	H	ST
31	13409	5009	599		TELEGRAPH CANYON	RD
31	13409	5009	591		TELEGRAPH CANYON	RD
31	13409	5009	591		TELEGRAPH CANYON	RD
14	12302	1006	279		F	ST
14	12302	1006	279		F	ST
14	12302	1006	279		F	ST
14	12302	1006	279		F	ST

Sorting offense data by victim name. Police data can be sorted by victim name as the primary sorting characteristic, using address, and other unique information to verify and resolve any apparent errors in the database. Both calls-for-service and offense data can be analyzed in this way.

Counting victims and offenses. Once data are sorted and matched, most electronic databases such as Microsoft^© Excel and SPSS^© contain a procedure for counting the number of unique addresses or names.

The best way to display repeat data is to create a table that includes the number of offenses and the number of victims (see Figure 6.) Additional columns may be used to report the percentage of households or offenses in each row.

Figure 6

Sample Table for Recording Number of Victimizations:

Residential Burglary

Number of burglaries	Number of victims burglarized (address or households)	Total burglaries
0	F (Total population or households - E)	0
1	A	A
2	B	2 x B
3	C	3 x C
4 or more	D	F - (A + 2B + 3C)
Total	E (A +B + C + D)	F

Once the figures in this table are entered, percentages (as displayed in Table 6) provide an easy way to see the proportions of offenses experienced by repeat victims and to determine the proportion of all victims who have been revictimized.

Cleaning data. Sorting and matching data reveals many errors. These may appear trivial (for example when road type is listed in one record as "avenue" and in another as "street") but such errors may reduce the number of matches and lead to underestimations of repeat victimization. The errors can typically be repaired by using another variable to sort and verify the correct version. For example, in Figure 5, 599 and 591 Telegraph Canyon are both listed as addresses. A check of the victim's name may show that the 599 address was a data entry error that should be corrected to 591.

Calculating time course. The amount of time between an initial offense and a subsequent offense (or between the second and third, or third and fourth) is called the time course. Most software can easily compute the time course in days by subtracting the date of the second offense from the date of the first offense. The procedure is only slightly more complicated when the data set exceeds 12 months, because the calculation of year must be converted so the software interprets Jan. 1, 2001, as 365 days later than Jan. 1, 2000. Descriptive statistics are typically used to report time course-the average number of days between repeat events, the range of least and most time between events, and percentiles such as the proportion of repeat events occurring within one week, 30 days, six months, or 12 months. The choice of the time period for reporting percentiles should be based upon natural or meaningful breaks in the temporal distribution of data.

Calculating rate. For many crime problems, the amount of victimization and repeat victimization will relate to exposure. For example, if there are 118 convenience stores in a city and 75 of these are Handy Andy^® stores, there are likely to be more robberies of Handy Andy than any other store. Exposure may also be increased by longer operating hours, more residents, more vehicular or foot traffic, and so on.

Planning Further Analysis

The analysis of repeat victimization, its concentration, and time course may clearly point to specific responses that will eliminate or reduce a problem; however, it is likely that further analysis will be necessary. Without further analysis, most police will attempt to conduct surveillance or undercover operations to apprehend one or more offenders at the location of repeat victimization. This response may be appropriate when repeat victimization relates to the repeat offending but such efforts may require extensive resources and will not necessarily change the general conditions of properties or behaviors of persons who face higher risk of victimization.

Collecting additional information. The initial analysis of repeat victimization may shed light on obvious vulnerabilities related to repeat victims. Such vulnerabilities may relate to characteristics for which data will need to be collected: store hours, age of victims (elderly or youth), environmental conditions, management practices, crime prevention devices used, victim behaviors, and so forth. All analysis should relate to factors that could be modified through some action on the part of the police or others. Patrol officers and investigators will often have insight into factors that contribute to high rates of repeat victimization.

Examining victim-suspect relationships. For some offenses, the relationship between victim and suspect will shed light on the nature of revictimization and on the nature of the police response. These relationships may explain many repeats for personal and property victimization of individuals.

Determining the role of boosts. Some offenses such as domestic violence will tend to involve the same offender over time but the role of repeat offenders may not be immediately obvious. For analysis of repeat victimization, efforts should be made to determine the contribution of repeat offenders. If a single prolific offender-an employee, family member, or someone else-underlies much revictimization, this information will guide efforts to determine the most effective response such as a panic or other temporary alarm, or increased short-term surveillance. Repeat victimization that continues after an offender is apprehended reflects flag rather than boost explanations.

Comparing victims and non-victims. Focusing on repeat victimization often highlights the differences between victims and non-victims, or between one-time victims and those who are repeatedly and even chronically victimized. Although convenience store robberies may be numerous, some stores are never robbed; crime in budget motels is not distributed across all such businesses, and factors such as management practices probably explain more variation than does location. Comparisons can reveal the average number of offenses for particular groups of businesses, locations, or victims.

Variables to be examined can be determined through discussions with investigators and patrol officers, as well as a reading of literature on the topic. For property offenses, variables may include method of entry, type of property stolen, security practices employed, proximity to crime generators such as schools or bars, and so forth. For personal victimizations, such as sexual assault or domestic violence, variables such as victim-suspect relationship, and drug or alcohol use may be critical. For commercial offenses, variables such as management practices, security features, demographic characteristics of customers, or type of merchandise may offer insight into distinctions between victims and non-victims.

For specific types of victims, such as schools, bars, budget motels, banks, and convenience stores, information can be collected through public records such as tax and business licenses.42

Establishing correlations between key variables or using cross-tabulations may provide important information about differences in victimization. Depending upon the nature of the problem, however, you may want to seek additional assistance in more complex statistical models. Providing this information may be helpful in implementing responses that require changes in victim behaviors or management practices.