Search

When we conceived this book, we thought we knew what was meant by ethnography and how its practitioners went about approaching their trade. And we thought that we could categorize ethnographies by what their topics were: offenders, police, homeless persons, or other actors or organizations involved in criminology research. What we found, however, is that we were more able to explain what it is not: it is not based on administrative data, on primarily closed-form questionnaires, on statistical analyses, on percentages, and on p values.

This article describes the methods that social scientists use to measure and compare the risks that various populations face of being murdered. It demonstrates that the West was unusually homicidal compared to the rest of the Western world, except for the most violent areas of the South during Reconstruction.

Like AI Blumstein, I came to criminology via operations research (OR) and, earlier, via electrical engineering, when I was engaged in designing a radio system for the Boston Police Department (Maitz and Waldron 1968; Maitz 1970). I subsequently became an OR analyst for the National Institute of Law Enforcement and Criminal Justice (now NIJ) in 1969. After 3 years of seasoning, I started teaching criminal justice at the University of Illinois at Chicago Circle (now UIC without the Circle1). There I taught inter alia undergraduate and graduate statistics and research methods. Having never taken them, I had to learn them on the fly. But like most engineers, I had a strong background in dealing with data, so it was not too problematic.2 But I brought different ways of looking at data (with an emphasis on the word look). Engineers and most hard scientists are taught to plot the data as they collect them. In physics we added weights to a spring and plotted the spring's length against the applied weight; we dropped an object and plotted the time it takes to fall different distances. It is through such exercises that we were able to demonstrate the linear relationship in the first example and the quadratic relationship in the second. In becoming a criminologist, however, I found that I had to deal with different types of data, for the most part based on random samples. The National Crime Survey (NCS), and its successor, the National Crime Victimization Survey (NVCS), were based on samples, as were a number of other important data sets. I set out to learn the methods of dealing with

Lott and Whitley note that our analyses of the errors in the county-level UCR data used in More Guns, Less Crime (J. R. Lott, University of Chicago Press, Chicago, 1998, 2000) ignore the fact that all data have measurement error, that the largest errors were in counties with low populations, and that population-weighted regressions were used. We agree that this mitigates some of the effects of the errors, but does not take them fully into account. We also note that this is but one of the problems associated with the analysis. We therefore find no reason to alter our original conclusion, that in their current condition, county-level UCR crime statistics cannot be used for evaluating the effects of changes in policy.

Like so many others, I fell into a “life of crime” by accident. An electrical engineer by training, my first post-postdoc job was with an engineering consulting firm, where I was engaged in doing operations research, mostly for the US Navy. While I was there, in the late 1960s, the firm was commissioned to develop a new management information system for the Boston Police Department, and I was assigned to work on improving the BPD’s communications system.

County-level crime data have major gaps, and the imputation schemes for filling in the gaps are inadequate and inconsistent. Such data were used in a recent study of guns and crime without considering the errors resulting from imputation. This note describes the errors and how they may have affected this study. Until improved methods of imputing county-level crime data are developed, tested, and implemented, they should not be used, especially in policy studies.

Recently, a number of states have enacted laws that allow citizens to carry concealed weapons. This “natural experiment” was analyzed by John Lott and David Mustard, who found that these right‐to‐carry laws reduced violent crime, with a substitution toward property crimes, in those jurisdictions that adopted this law. Of particular importance, they found that homicide was reduced significantly, with even greater declines in larger jurisdictions. Their findings came at the same time that major reductions in homicide were occurring in many cities and states that did not change their gun‐carrying policies, which lead to questions of whether their finding was spurious, caused by problems with the data or methods. In this paper, we describe an analysis that looks at the effect of changing one aspect of their homicide analysis: disaggregating homicide data by weapon type, victim characteristics, and victim‐offender relationships. The results show that the liberalized carrying laws are associated with a...

The goal of statistical analysis is to find patterns in data. Most statistical methods rely on analyzing the effect of the same set of variables on the population under study, i.e., nomothetic analysis. Therefore, when data are collected in the social sciences, most often they are put in a framework that resembles a spreadsheet: each row represents a separate individual, and each column represents a separate characteristic (or variable) that pertains to that individual.

This paper provides some examples of the utility of graphical methods in analyzing data. While such methods are not expected to supplant standard statistical techniques, they can help the researcher in understanding characteristics of the process in ways that cannot be replicated using the standard methods.

In the 1830s Simeon-Denis Poisson developed the distribution that bears his name, basing it on the binomial distribution. He used it to show how the inherent variance in jury decisions affected the inferences that could be made about the probability of conviction in French courts. In recent years there have been a number of examples where researchers have either ignored or forgotten this inherent variance, and how operations research, in particular mathematical modeling, can be used to incorporate this variance in analyses. These are described in this paper, as well as other contributions made by operations research to the study of crime and criminal justice.

Most of the methods we use in criminology to infer relationships are based on mean values of distributions. This essay explores the historical origins of this issue and some counterproductive consequences: relying too heavily on sampling as a means of ensuring “statistical significance”; ignoring the implicit assumptions of regression modeling; and assuming that all data sets reflect a single mode of behavior for the entire population under study. The essay concludes by suggesting that we no longer “make do” with the standard methodologies used to study criminology and criminal justice, and recommends developing categories that more accurately reflect behavior and groupings than the ones we currently use; looking at alternative sources of data, including qualitative data such as narrative accounts; and developing alternative methods to extract and analyze the data from such sources.

Criminal justice data are hardly ever linear, normal, and/or independent, but most statistical techniques rely on the assumptions of linearity, normality and/or independence. Moreover, since the data are often entered by hand, they are prone to error. Plotting the data permits the analyst to determine the extent to which the assumptions are valid and to catch obvious errors in data entry. Moreover, while standard statistical techniques are useful in testing hypotheses, visualization allows the data to tell its story and thus is useful in generating hypotheses. This chapter provides some examples of how visualizing criminal justice data contributes to a fuller understanding of the processes that produced the data.

One sees different features from different points of view. Flying over a region provides a view of geographical and geological features not visible from the ground, while ground-level observations show details not apparent from the air. Similarly, different units of analysis applied to crime data can bring out different patterns in crime. This chapter describes how two different units of time, one measured in years, the other in weeks, can be used to extract two different types of geographical patterns. One, the “wave” analysis, traces the actions of whole groups over a long time period. In particular, it should be of use in investigating the effect of in- or out-migration of racial/ethnic groups in neighborhoods in a metropolitan area and the effect that these population shifts have on crime. The “particle” analysis, on the other hand, focuses on the career trajectories of individuals, from an early age, as they experience noteworthy events during their lives and move (or are moved) from place to place in a city. This view of a neighborhood’s at-risk youths can be used to describe the role of residential mobility in crime.

Publisher Summary This chapter discusses the operations research in the public sector. Operations research is a scientific approach to decision making. It is concerned with scientifically deciding how to design and operate systems, usually under conditions requiring the allocation of scarce resources. The proper use of the operational research approach reduces the area of hunch to an absolute minimum, showing where judgment is needed and where it is unnecessary. The chapter describes the application of mathematical and statistical models to a variety of public sector issues, problems, and systems. The revolution that has been effecting all sciences—the existence of high powered desktop mini- and micro-computers, and the consequent availability of data, computational power and communications—has had an immense impact on the practice of operations research, particularly in the public sector. The availability of user-friendly software at every public official's desk makes it possible to use mathematical models, in all varieties of problem areas, without a need for consultants, analysts or experts.

Publisher Summary This chapter reviews the applications of operations research and management science (OR/MS) to problems of crime and justice. The chapter describes how OR/MS has been used in studying crime and criminal justice. The correctional system has several goals: deterrence, education, retribution, rehabilitation, and incapacitation. The models relating to incapacitation are explored. The recidivism of an offender, his/her continuing to commit crimes, is a crucial factor in criminal career research. Deterrence is the reduction in criminal activity by the general public attributable to a given intervention. It is not associated with specific individuals, so it cannot be measured by tracing offenders' careers. Another area of importance to prison administrators is the projection of future populations of offenders. Operations research has also been used to study crime and how the police react to it. Problems related to juries investigated using operations research include the elimination of intentional bias in empaneling juries; considerations related to increasing bias in a jury by using peremptory challenges; and studies relating to the reduction of juror waiting time and the cost of empaneling juries. The systematic, quantitative thinking exemplified by these approaches to problem-solving has been very effective in helping to improve the administration of criminal justice.

Information about specific crimes is not the only safety-related information available to (or useful to) the police or the community. Safety is a multifaceted issue, and crime is only one aspect (an important one, to be sure). This chapter focuses on the nature of incivilities, which are acts or situations that may not be criminal, but that have an adverse effect on a community. The extent to which police data can be used to provide information about incivilities, and thus improve community safety, is described as well. Also discussed are the nature of the community’s perceptions of dangerous locations, or hot spots.

MAPADS was implemented in the Area 5 violent crimes unit on a pilot basis. It was used only for robbery cases for a period of 5 months (January–June, 1988). Although the system proved to be effective even in this limited implementation, its effectiveness depended greatly on how information was entered and maintained in MAPADS and on the reward structure (both real and perceived) with respect to in-formation sharing by detectives. In this chapter we describe (a) how the volume of crimes relates to detective specialization, (b) issues relating to the sharing of information by detectives, (c) how MAPADS was used during the implementation, and (d) the data entry problems experienced in this implementation.

Our experience with the North and Central hot spot provided us with a deeper understanding of the intricacies of the police-community-crime-incivility nexus. In an attempt to obtain a profile of this hot spot, a community organization staff member was trained to work with merchants to record events they did not report to the police. Preliminary analysis of the data made it clear that crime in the area of North and Central is a complex problem. The intersection is situated in an area where racial diversity and the close proximity of middle- and low-income people can feed fears among residents and potential investors. Finding a solution requires more than the collection, analysis, and mapping of data. Strategies based on the experience of merchants, residents, and police must be devised. These strategies can only come out of a coordinated planning process that involves all relevant parties. Gains in the area can only be maintained if the basis for action is cooperative. This need for a nontraditional problem-solving approach suggests a team response, consisting of police officers and community representatives. The task is unfamiliar for most police agencies and cannot be handled in the course of normal patrolling duties.

One major but unanticipated result of the crime mapping project is what we have termed proactive management, defined by one official as an administrator having access to data so he can make decisions immediately. Rather than having to rely on central headquarters capabilities and wait for a cumbersome mainframe computer operation to eventually provide the requisite analysis, the district commander can now delegate the crime analysis officer to do the work, receive the results that day, and make management decisions as indicated. In current management science terminology, this is called a decision support system.

Implementing a crime mapping system is not a simple task. It involves an understanding of the intricacies of the data that are to be analyzed as well as what kinds of additional data might be usefully analyzed that are not currently used. One needs to understand the nature of the crime data normally collected and analyzed by police, how the data are collected and disseminated, how the police use the data for crime analysis and other purposes, and how community organizations have been involved in crime data analysis.

In this chapter we describe the context in which the project “Mapping Crime in Its Community Context” was developed. Because the focal point of the project is information about crime and public safety, we first discuss how information about crime, which generally comes from the community, is handled in a police department. In the crime mapping project the CPD was the grantee and CANS, a consortium of 10 community organizations, was one of the three subgrantees. A historical perspective is given of the relationship between these two organizations and of their relationship to the two other subgrantees, Northwestern University and the University of Illinois at Chicago.

A technical innovation such as MAPADS fosters a great deal of change, in both the police department that implements it and the community that the department serves. The manner of implementation can change the way the police relate to the community and the way they perform their routine patrol work. These changes should affect both the way the innovation is evaluated and the way police officers are trained. This chapter discusses some of the issues raised by considering the MAPADS system as an integral component of a police department.

The approach we have taken in implementing the mapping system has been to employ techniques most congenial to the users and to the nature of the data with which they must work. Sometimes this has led to formal quantitative approaches. Often, however, inferences based on an experienced officer’s eyeballing of the data to come up with a pattern have been more useful than inferences based on statistical significance.

The crime mapping research project was primarily concerned with conceptualizing, developing, and refining MAPADS as a system for use by police. Thus, it was not also possible to perform a formal evaluation of its capabilities, yet we do have strong indications of the benefits of MAPADS for patrol and tactical use. In order to explain the way in which the mapping system was used by patrol and tactical officers, this chapter describes the way they were assigned to anticrime patrols prior to the advent of the mapping system. A number of typical examples of MAPADS use are then given, showing how patrol deployment has been affected by MAP ADS. This chapter then describes one of the most important benefits of MAPADS in police patrol—its use as a beat’s memory.

The discussions in earlier chapters all demonstrate the importance of the flexibility of the MAP ADS system. Because such flexibility is built into MAPADS, the organization becomes more responsive to changing local circumstances. One of the important characteristics of crime analysis is that the particulars of a given case, the facts that make it solvable, do not remain the same from case to case. The similarities between cases are mostly unusable in solving cases. By developing sets of routinized procedures, an organization intended to solve crimes may inhibit its own abilities to do so. To increase effectiveness, the organization must be able to adapt to the particular circumstances involved, both on a case-by-case basis (which is the reason for street-level discretion) and on a slightly more general basis, such as concentrating on auto thefts at one time and burglaries at another (which is the reason for command-level discretion). In order to make these adaptations, field units need access to contextual data about local conditions, and they need to be able to analyze that data in a changing variety of ways. These are exactly the benefits MAPADS provided the district personnel and was beginning to provide area detectives.

Two units within the CPD were chosen as sites for implementation of the project. Because the project’s goals relate to inferring patterns by mapping crime and community data on computer-generated maps, one of the units selected was a patrol district and another was a unit within the detective division that could take advantage of crime mapping. This chapter describes the manner in which the selection of units was made and the results of the implementations.

The technological advance represented by computerized mapping constitutes only part of the potential benefit of MAPADS. To be truly effective, MAPADS needs to work in an organizational and social context that permits the police and the community to interact while analyzing the data. This interaction requires trust and practice as well as an organizational and social infrastructure. We expected this infrastructure to be developed as the police and the community used MAPADS jointy in working on problems of crime and safety in the community. This did occur: the project made an important contribution to the development of this infrastructure. Cooperative working relations were established and information was exchanged regularly between the community and the police. This was an important achievement of this project.

At the moment, there are two primary types of data collected and displayed by MAPADS. The first type is the geographical data: the streets, blocks, railroad tracks, and so on that make up the portion of the city being displayed. The second type is crime data: type of crime, location of the crime, and time of occurrence. In addition, when information about hot spots or other safety problems are provided to the police, these are also mapped.

In a sense, criminologists are luckier than economists. Economists are asked to forecast what will happen to the economy, and a lot of them get it wrong. Criminologists are asked to back-cast, to explain what happened after it happens, and probably with about the same success as economists have at forecasting: even though we know what the outcome is, we often don't know why. But this is not for lack of trying. After reading the papers and hearing the presentations, I am struck by the care with which the authors have analyzed the homicide data at three levels: national, multi-jurisdictional, and single-city. My comments should in no way be seen as criticisms of their efforts; rather, they are suggestions as to additional steps that might be taken. My comments focus on two areas: disaggregation of homicide data, and considering whether "regression to the mean" might explain the recent decline in New York City homicides.

This paper compares the crime statistics generated by the FBI's Uniform Crime Reports (UCR) to those generated by the National Crime Survey (NCS), collected by the Census Bureau for the Law Enforcement Assistance Administration. The crime reporting process is described for both the UCR and NCS to determine the nature of their respective biases. Events within the processes are analyzed to develop a new estimate of crime incidence based on both the UCR and NCS. Numerical examples are then developed to point out the deficiencies in the NCS, and to make suggestions for the improvement of its accuracy.

Phillips, Ray, and Votey (1984a) used a model of traffic casualties incorporating three independent variables—miles driven, rainfall, and alcohol consumption. They applied it to British data to attempt to determine (1) whether the British Road Safety Act of 1967 had an impact on road casualties and (2) the relative contribution of the three variables to road casualties. They accomplished the first objective using interrupted time series analysis, but fell victim to the ecological fallacy in their interpretation of the second.

A marked decrease in arrest rates-up to 70%-was noted in the evaluation of a juve nile corrections program. This so-called "suppression effect" has been widely publicized as proof of the effectiveness of correctional intervention. In this article we show that the sup pression effect can easily be explained as a statistical artifact, regression to the mean, assuming that judges base their sentencing decisions on juveniles' prior records. We con clude that before/after comparisons should not be used to evaluate delinquency programs.