The use of crime mapping as a tool for policing has a long history, and it has been adopted on a much broader basis since the advent of desktop computers made mapping dramatically easier. The adoption of mapping technology has not been problem-free, however. While the software has lessened the time and labor required to generate maps, the production of sophisticated maps and their integration into policing have not been straightforward.
An Abridged History of Crime Mapping
Geographer Borden Dent, in his article “Brief History of Crime Mapping,” traces the origin of the mapping of crime to France, where in 1829 Adriano Balbi and André Michel Guerry created maps that showed the relationship between violent and property crimes and educational levels. Within a few decades this approach—visually displaying differences in crime across geographic units—had spread to England and Ireland. In 1849, Joseph Fletcher created maps that showed the rate of male incarceration for serious property and violent crimes across counties in England and Wales, and in 1861, Henry Mayhew presented a number of maps displaying the English and Welsh county rates for a variety of crimes: rape, assault, bigamy, and abduction, among others.
These early maps are examples of choropleth maps—that is, maps that display quantities of things in areas. More specifically, in choropleth maps geographical areas are divided into multisided figures called polygons, which are then shaded depending on the value of the variable being displayed. Balbi and Guerry’s maps, for example, were shaded with crayon to show different levels in education. More modern choropleth maps are familiar to anyone who followed the most recent election returns: a map of the United States with each state shaded red or blue depending on the party of the senatorial victor is a very basic choropleth map. An example of a chloropleth map of midtown Anchorage is shown in Figure 2.
Sociologists, particularly those associated with the University of Chicago, began using mapping in the first few decades of the 1900s. Among the earliest were Progressive Movement social work educators Sophonsiba Breckenridge and Edith Abbott, who, in 1912, mapped where delinquent children had lived in Chicago over the period 1899 to 1903. This map, with each dot standing for one home, is an example of a point map—that is, a map in which points representing particular geographical locations, be they addresses or XY coordinates, are the main data element.
Perhaps the best known maps in criminology were created by the Chicago School sociologists Clifford Shaw and Henry McKay, who constructed a choropleth map using aggregations of addresses of close to 3,000 male delinquents in Chicago for the period 1927 to 1933. The map featured polygon shading to indicate rates of delinquency. Like Breckenridge and Abbott, Shaw and McKay also constructed point maps of the locations of the homes of about 10,000 male delinquents who had come before the juvenile court of Cook County in the years 1934 to 1940. Shaw and McKay noted that the spatial distribution of juvenile delinquents’ homes remained fairly constant over these differing time spans, despite the fact that there was a high degree of residential mobility in various areas of Chicago. Their work, with that of others, gave rise to the social ecology approach to studying crime. This approach assumes that crime is to a large extent caused by community- and neighborhood-level variables, such as land use, infant mortality rates, mental disorders, tuberculosis, and the percentages of minorities and families on social assistance.
It is worth noting that all the maps described above were made without the benefit of computers. The underlying base maps—that is, the maps showing streets, roads, and other major features such as water and railroads—had been drawn by hand. Each point was located manually, and polygons were shaded using ink, pencil, or crayon. Creation of a single map could take many, many hours of tedious labor. In one sense, crime mapping was an idea that arose before its time, before the requisite technology was available. It illustrates what Sean Gilfillan in The Sociology of Invention defined as the uselessness of premature invention—“an invention which for any reason did come before its time remains useless and undeveloped until its proper day dawns.” The “proper day” for crime mapping did not come until developments in computer technology made it feasible to run mapping programs on relatively inexpensive desktop computers.
Prior to the widespread use of desktop computers, the few police departments who did crime mapping relied on primitive techniques such as sticking thousands of pins into large maps attached to the wall. In his essay “Geographic Information Systems and Crime Analysis in Baltimore County, Maryland,” Philip Canter describes the county’s pin-mapping efforts as requiring twelve maps and 70 square feet of wall to cover the entire area of the jurisdiction. While these maps were reasonably good for detecting clusters of criminal activity, they did not permit more sophisticated analyses that incorporated other, nongeographic factors, such as modus operandi or time of offense.
Even with the advent of computers, generating a crime map was no small feat, as geographer and criminologist Keith Harries has noted in Mapping Crime: Principle and Practice. Mapping with gigantic mainframe computers was still extremely labor-intensive. First, there was the work involved in describing the boundaries of the map with numbered coordinates on punched cards. Then came the labor of keypunching the cards, followed by a similar process of coding and keypunching to put the data on the map.
Such labor intensiveness meant that few police departments could afford to produce computerized crime maps. It wasn’t possible for most agencies to afford crime mapping until desktop computers became widely available in the mid-1980s to early 1990s and microprocessor speed increased.
The Advantages and Disadvantages of Desktop Crime Mapping
In their work, David Weisburd and Cynthia Lum argue that the expanding body of research on the context of criminal events (e.g., routine activities, situational crime prevention, and criminology of place) has contributed to a paradigm shift in policing. An increasing focus on small geographic areas has led to increased recognition of the value of a hot-spots policing approach, with crime mapping central to this approach. In addition, many agencies have experimented with problem-oriented policing. The value of crime mapping in assisting with problem scanning, analysis, and response assessment is well documented.
Technology change, specifically the increasing availability of desktop computers, has been the main reason for the recent rapid growth in crime mapping by American police departments. Desktop computers have made the creation of maps dramatically faster. Computerized crime mapping now relies less on labor-intensive processes; it is no longer necessary to draw maps by hand using special pens and India ink. Pressing only a few buttons on a keyboard produces similar effects. Desktop computers are also forgiving when one makes a mistake; it is a simple matter to redraw a map.
However, despite the advantages of computerized crime mapping, its adoption by police departments has not been straightforward. A study by the Police Foundation in 2000 found that although many departments could afford the equipment for mapping and had little difficulty finding personnel interested in learning the technology, “the learning curve is often underestimated and integrating the mapping into departmental routine requires substantial planning and effort.”
Studies by Thomas Rich and by Cynthia Mamalian and Nancy LaVigne of the National Institute of Justice have found that the majority of departments using crime mapping are creating automated pin maps and generating hot-spot maps. Along a range of difficulty, pin maps and hot-spot analysis are at the low end. For more advanced analysis, some knowledge of cartography and geographic information systems (GIS) is required. Leslie Kennedy in “Repeatable and Emergent Forms: Searching for Crime Spaces and Crime Places,” a paper presented in 2002, quotes an observer: “Police departments are going crazy. They thought all they had to do was put up MapInfo, and now consultants are coming along and telling them they need to do neural networks.” The complexity of these latest innovations may well prevent their adoption by all but the most technologically advanced departments.
With initial cost outlays for a basic mapping program low, the cost of crime mapping is not an impediment to most police departments. For example, ArcView, the popular software used by about 40 percent of departments doing mapping, according to Mamalian and La Vigne, retails for under $2,000. Most mapping software is relatively user-friendly, and there are many opportunities to receive training for those who wish to move beyond the manual. While these features of modern computerized crime mapping make it easy for some agencies to start crime mapping, particularly the larger ones with bigger budgets and staff who can devote time to accessing data and learning how to use the software, these features are not as useful for smaller agencies, which may be unable to free up personnel for training.
The Spread of Crime Mapping Use by American Police Departments
A small but growing body of research provides some insight on the use of crime mapping in American police departments and the associated problems. Initially, the technology spread quickly. In 1995, Thomas Rich reported the results of an informal poll conducted by the International Association of Chiefs of Police in “The Use of Computerized Mapping in Crime Control and Prevention Programs”: 30 percent of 280 police departments reported having mapping software. In the Police Foundation study mentioned above, a series of telephone interviews were conducted with personnel in 51 departments that had received grants from the Community Oriented Policing Services (COPS) Office to implement crime mapping. The study concluded that many departments found crime mapping considerably more difficult to implement than they had anticipated and had problems integrating the technology with problem-solving and community policing. The perceived need for technical assistance and training was great.
A more rigorous survey of 2,004 police departments in 1997-1998 conducted by Mamalian and LaVigne at the National Institute of Justice’s Crime Mapping Research Center (CMRC) found that large departments (i.e., those with 100 or more sworn officers) were more than ten times as likely to use computerized crime mapping than small departments (36% vs. 3%). Overall, the rate of use was 13 percent, and the average length of use was 3.3 years.
Weisburd and Lum further analyzed these data by graphing responses to the question, “For how long has the department been doing crime mapping?” The result is shown in Figure 1. These two researchers then conducted their own pilot study in 2001 of agencies with over 100 or more sworn officers in order to focus on the relationship between research and crime mapping innovation. They found that this pattern still held, even four years later, suggesting that crime mapping as an innovation has diffused very rapidly among large agencies.
Weisburd and Lum’s pilot study examined the characteristics of different stage adopters, classifying responding departments into those who were early adopters, those who were later adopters, and those who had not yet adopted mapping. Early adopters were significantly more likely to have members who attended crime mapping or crime analysis conferences; were involved in a crime mapping or crime analysis email discussion list; had easy access to academic books about crime mapping or crime analysis; and were aware of research on hot spots, crime mapping, or crime analysis. It was implied by this study, but not clearly stated, that there is a network of police departments that communicate about crime mapping through the Internet and at conferences.
A more detailed analysis of the CMRC data undertaken by the author of this article found that diffusion of crime mapping was much more rapid in large police departments than in those with under 100 sworn officers, particularly departments with fewer than 50 sworn officers. In addition, some areas of the country, particularly the Pacific, Mountain, and South Atlantic census regions, experienced much faster diffusion than others, such as, most notably, the New England region.
The Discontinuance of Crime Mapping Use by American Police Departments
To date, there have been only two studies examining the specific issue of why police departments stop using or discontinue crime mapping, both by this author. In the first, data from the Law Enforcement Management and Administrative Statistics surveys from 1997 and 1999 were analyzed in order to investigate the extent of discontinuance of crime mapping and possible explanations. Of the 615 police departments in the sample that reported doing crime mapping in 1997, 242, or close to 40 percent, had apparently stopped mapping by 1999. Department size seemed to be an extremely robust predictor of discontinuance. Large departments, particularly those with 250 or more sworn fulltime officers, were significantly less likely to discontinue crime mapping than were the smaller departments. Of the 111 larger departments, only 3 (or 2.7%) had discontinued, compared to 235 out of 486 (or 48.3%) of the smaller agencies. Lower levels of staffing in the technical support area were strongly associated with discontinuance, but when department size was introduced as a control variable, the relationship became insignificant. So too did the finding that discontinuers had higher levels of funding for equipment. Discontinuers were significantly less likely than continuers to have engaged in problem-oriented policing activities or crime analysis, but again, department size was a strong intervening variable.
In a 2002-2003 study of 347 municipal police departments in New Jersey, The Adoption and Discontinuance of Computerized Crime Mapping by Municipal Police Departments in New Jersey, the author found that 12 of 48 departments that had undertaken crime mapping had since discontinued use of the innovation. Departments that discontinued mapping were found to differ significantly from continuers on a number of demographic variables. Discontinuers were significantly smaller than continuers, in terms not only of population served but numbers of sworn and civilian personnel. Continuers tended to be in jurisdictions with more crime events, although with respect to rates this difference was only statistically significant for violent crime. Neither population density nor geographic size was significantly different for continuers and discontinuers. In general, continuers scored higher on variables related to urbanity; in this sense, discontinuers resemble departments that never adopted mapping in the first place. This finding is consistent with what Everett Rogers noted on discontinuance in Diffusion of Innovations: discontinuers are much more similar to those who adopt an innovation very late than to early adopters.
In giving reasons for discontinuance, two respondents in the study cited technical difficulties, such as system failure, manpower demands, or incompatibility with other computer systems. Another department, which had received free mapping software, subsequently abandoned it after only five months because of problems obtaining current base maps.
Other respondents cited nontechnical reasons for discontinuance. One police department in a suburban community of about 10,000 in the northeastern part of New Jersey commented that they “did not find mapping to be of great benefit,” while another agency, which reported having done crime mapping for decades, lost its equipment in a flood and had not yet replaced it, although the department intended to take up the technology again.
This small group of discontinuers provides some support for notions about discontinuance presented in the diffusion literature. It is argued that innovations are abandoned for two reasons—replacement or disenchantment.
With this study, disenchantment, or dissatisfaction, was essentially the primary reason for discontinuance given by respondents. Departments that were unhappy with the results of crime mapping or those that had technical problems experienced this disenchantment. Other surveys of crime mapping departments have discovered similar implementation problems. Rich found that difficulties with acquiring data and ensuring data quality were impediments to the increased use of mapping software among the departments in his study. Problems associated with moving data between computer systems and geocoding were also identified by the Police Foundation’s 2000 survey.
As this brief review indicates, the adoption of crime mapping as a widespread policing tool has not been problem-free, despite the available technology. Although more departments are now able to produce and use fairly basic maps, more sophisticated mapping incorporating multiple variables is not as easily achieved. It is not clear, however, that police departments have much need for complex maps, which may help explain the research finding that departments sometimes abandon mapping after trying it for a period.
Sharon Chamard is an assistant professor with the Justice Center. This article is largely based on her 2003 doctoral dissertation, The Adoption and Discontinuance of Computerized Crime Mapping by Municipal Police Departments in New Jersey.