Over the past three decades increased attention has been placed on the prevention of crime through opportunity reduction. At the same time, researchers have embarked on the task of determining the impact of opportunity-reduction measures on the levels of crime and fear of crime within communities. While the focus of this research agenda is clear, less certain is the appropriate methodology that should be used in reaching usable findings. This paper discusses the current demands facing situational crime prevention (SCP) evaluation methods within the context of a pull toward randomized experimentation, the push toward scientific realism, and the expansion of developing technical issues such as crime displacement, diffusion and anticipatory benefits, among others. It undertakes an appraisal of the current body of SCP evaluation methods by examining over 200 evaluations of SCP efforts carried out from 1970 to 2007. It reviews the types of methodologies employed, reports on the conclusions of these studies, and discusses the implications for future SCP evaluation.

In problem-oriented policing (POP), assessment is an integral part of the process in which the effectiveness of the implemented measures is judged. When the POP philosophy was introduced, more of the effort was put into disseminating it and explaining its merits, than to instruction about the more technical parts of the process, like evaluation. The POP approach attracted a lot of interest, and many police forces started to practice it as a strategy to prevent crime and disorder. Rather soon, evaluative examinations showed that the implementation was often frail and that the quality of the projects frequently was poor. This goes especially for the assessment phase, which is technically demanding. Given the skills of practitioners, in most cases police officers, and the circumstances in which projects are carried out, a higher degree of sophistication of the evaluations is not to be expected. Evaluations almost always have a one-group before-and-after design, which is often the only feasible one, and which is sufficient for many purposes. However, examination of a number of projects that represent best practice shows that there is room for improvement. In order to achieve this, the police need to cooperate with academically trained analysts when practicing problem-oriented policing.

Interventions that block crime opportunity structures change crime signatures. Signatures are data patterns that describe how crime is associated with various features of the opportunity structure. The analysis of crime signature change, as part of crime prevention evaluations, can improve the internal validity of evaluation findings. This paper describes the logic of this argument, provides examples of how it works, and develops a four step procedure – SCEMA – for implementing this approach.

Problem-oriented policing initiatives seem well positioned to generate positive community reactions to police intervention because this approach focuses on dealing with specific crime and disorder problems that cause ongoing concern to affected community members. This paper examines the use of interviews with community members in small areas to assess the effects of police crime reduction strategies on citizen perceptions of disorder problems, police behaviors, and fear of crime. Our study uses citizen interview data collected as part of a randomized controlled experiment in Lowell, Massachusetts to evaluate the effects of problem-oriented policing strategies on crime and disorder hot spots. Our interviews with key community members revealed that selected place users noticed an increased police presence and that disorder problems were positively impacted in the treatment hot spots as compared to the control hot spots. However, the respondents did not detect any significant changes in police strategy, the willingness of the police to work with residents, or the demeanor of the police towards citizens. Our research suggests that small area interviews add considerable value to the evaluation of crime reduction strategies, but great care needs to be taken in analyzing and interpreting these data.

There are many meanings of "What" in "What Works?" Five are identified in this chapter. These relate to: particular interventions; classes of measure; mechanisms; strategies; and context-mechanism-outcome pattern configurations (CMOCs). Randomized Clinical/Controlled Trials (RCTs) are often treated as the gold standard for evaluation research in health. They have been used as the same standard in policing and crime prevention, though there are far fewer examples of them than there are in medicine. The "what" questions answered in RCTs relate to particular interventions or classes of measure. The adequacy of RCTs for evaluations in policing and situational crime prevention is considered herein through the detailed examination of an exemplary study. It is found to fail, in spite of its technical strengths. Its logic and assumptions are found to be seriously flawed. There are evaluation designs in health that do not involve RCTs. These also tend to answer different "what" questions from those asked in RCTs. Policing and situational crime prevention evaluations have more to learn from these than from RCTs.

Central to achieving the potential of evaluation is the ability to estimate the impact of crime prevention policies and programs on crime and to extrapolate the findings to other settings. The potential outcomes framework presented in this paper is the most compelling theory of causality, but its implications for extrapolating results have not been fully considered. The paper demonstrates the importance of balancing treated and untreated study samples as well as the study population and the target population for accurately estimating effects, generalizing them to the study population, and extrapolating them to other target populations. Alternative methods for producing unbiased estimates of program effects and the implications for extrapolating the estimates to other target populations are discussed.

According to Moore's law, increases in computing power are roughly exponential over time. As a consequence, the application of computational methods to old and new problems becomes ever more possible, even with desktop computing. Such methods, and in particular computer simulation, have considerable potential in the study of crime but their application is relatively novel at this time. Consequently, the aim of this chapter is to consider the possibilities with a particular focus on how they might be used to inform the evaluation of crime reduction activity. A number of different types of computational methods will be discussed and examples of the types of policy-related questions for which they might be used considered. The strengths and weaknesses of the approaches described will also be discussed.