Robert Apel

Propensity Score Matching in Criminology and Criminal Justice

The propensity score methodology has become quite common in applied research in the last 10 years, and criminology is no exception to this growing trend. It offers a potentially powerful way to estimate the treatment effect of some intervention on behavior when the receipt of treatment arises in a nonrandom way – this is the selection problem. It does so by creating synthetic “experimental” and “control” groups that are equivalent on a large number of potential confounding variables. In this chapter, we first introduce the counterfactual framework on which the propensity score method is based and define the average treatment effect. We then outline technical issues that must be addressed when the propensity score method is used in practice, including estimation of the propensity score, demonstration of covariate balance, and estimation of the treatment effect of interest. To provide a step-by-step example of the method, we appeal to the relationship between employment and substance use in adolescence. Following a brief review of research in criminology and related disciplines that employ the propensity score methodology, we offer a number of guidelines for use of the technique.

Youth Behavior, School Structure, and Student Risk of Victimization

Although a growing body of research on student safety focuses on school disorder, school climate, and the intersection of community/situational factors, comparatively less research has focused specifically on the individual‐ and school‐level factors that put students at risk of victimization in the immediate school environ. The present study is an attempt to broaden our understanding of the contribution of schools and school behavior to the victimization experiences of students. We compare traditional routine activity constructs to understand whether and how they differentially influence the risk of community and school victimization. Additionally, we investigate what school‐related variables (behavioral and structural) explain variation in young people’s risk for school victimization.

A Job Isn’t Just a Job: The Differential Impact of Formal Versus Informal Work on Adolescent Problem Behavior:

Research consistently demonstrates a positive correlation between hours of employment and problem behavior for adolescents. In response, the National Research Council (1998) proposed limits on youth work involvement, and its recommendation forms the basis for proposed legislation to amend federal child labor provisions. An unanticipated consequence may be to increase the amount of time that youths spend in the informal labor market because child labor laws only govern youth employment in the formal labor market. In this article, the authors attempt to address this policy implication and fill a gap in the extant literature by examining the impact of both formal and informal employment on delinquency and substance use. Because work patterns tend to be very different by gender and race or ethnicity, the authors estimate separate models for these subgroups. The authors use longitudinal data to deal with the possibility that there are unobserved differences between those that work and those that do not.

Assessing the Effect of Adolescent Employment on Involvement in Criminal Activity

This article considers the problem of estimating the effect of a binary independent variable (employment) on a binary outcome variable (involvement in criminal activity) for a nationally representative sample of adolescents (ages 15-18). The authors’ bivariate analysis confirms a common finding from the literature, that adolescent employment is associated with increased risk of involvement in criminal activity. They then turn to the problem of assessing whether this association is sensitive to plausible assumptions about the impact of other variables (both observed and unobserved) on both employment and crime. This assessment reveals that both the sign and magnitude of the maximum likelihood estimate of the employment effect are quite sensitive to these assumptions. Based on this evidence, they conclude that future efforts to understand the adolescent work-crime relationship will benefit from resolving the ambiguities identified by their analysis.

Imprisonment length and post-prison employment prospects

This study considers the relationship between imprisonment length and employment outcomes. The data are a unique prospective, longitudinal study of Dutch pretrial detainees (N = 702). All subjects thus experience prison confinement of varying lengths, although the durations are relatively short (mean = 3.8 months; median = 3.1 months). This contrasts with prior research that was limited to the study of American prison sentences spanning an average of 2 years. These data thus fill a gap in the empirical base concerning short-term confinement, which is the norm in the United States (e.g., jail incarceration) and other Western countries. Using a comprehensive array of pre-prison covariates, a propensity score methodology is used to examine the dose-response relationship between imprisonment length and a variety of employment outcomes. The results indicate that, among prison lengths less than 6 months in duration, longer confinement is largely uncorrelated with employment. In contrast, among spells in excess of 6 months, longer imprisonment length seems to worsen employment prospects.

The spillover effects of focused deterrence on gang violence

Background: Focused deterrence strategies attempt to increase punishment risks faced by violent gangs through the development of new and creative ways of deploying traditional and non-traditional law enforcement tools. In addition to increasing the swiftness and certainty of sanctions, these strategies explicitly communicate incentives and disincentives to deter likely gang offenders from violent behavior. Objective: This study seeks to determine whether focused deterrence strategies generate spillover deterrent effects on the gun violence behaviors of vicariously treated gangs that were socially tied to directly treated violent gangs. Research Design: A nonrandomized quasi-experimental design was used to evaluate the gun violence reduction effects of focused deterrence strategies on directly treated gangs and vicariously treated gangs. Propensity score matching techniques were used to identify balanced comparison gangs for the vicariously treated gangs. Growth curve regression models were used to analyze gun violence trends for treated gangs relative to comparison gangs. Unit of Analysis: Quarterly counts of fatal and non-fatal shootings involving specific street gangs between 2006 and 2010 served as the units of analysis. Measures: Key outcome measures included quarterly shootings committed by specific gangs, shooting victimizations suffered by specific gangs, and the total number of shootings involving specific gangs. Results: The focused deterrence strategy was associated with statistically significant reductions in total shootings by directly treated gangs and vicariously treated gangs. Conclusions: Our study finds that vicariously treated gangs were deterred by the treatment experiences of their rivals and allies. This suggests that focused deterrence strategies can generate spillover crime reduction effects to gangs that are socially connected to directly treated gangs.

The Effects of Jail and Prison Confinement on Cohabitation and Marriage

This study uses the National Longitudinal Survey of Youth 1997 to explore the relationship between incarceration and the stability of cohabiting and marital relationships. Self-report dates of relatively short confinement in jail or prison (median one month) are linked with data on cohabitation and residential partnerships, by month, from ages 18 to 32. I estimate the effects of incarceration on transitions into and out of cohabitation and marriage while controlling for other salient life events (e.g., employment, parenthood). Findings indicate that incarceration precipitates an immediate and persistent disruption in residential partnerships and is also a long-term impediment to the transition to marriage (but not the transition to cohabitation). The long-term disruption in existing residential partnerships applies equally to females and males, as well as to whites, African Americans (males only), and Hispanics.

A Situational Model of Displacement and Diffusion Following the Introduction of Airport Metal Detectors

Much of the discourse surrounding counterterrorism centers on the inevitability of displacement, or the substitution of another form of terrorist attack in place of the one that has been thwarted. Yet a longstanding tradition of research in situational crime prevention finds that displacement is far from inevitable, and often depends crucially on the specific features of the incidents in question. In fact, crime prevention efforts are often followed by a “diffusion of benefits” (i.e., crime reductions) to incidents, groups, or locations that were not the intended target of the intervention. The current study examines various forms of displacement and diffusion in response to airport metal detectors among terrorist groups that had been involved in the perpetration of aviation attacks prior to their implementation. Using data from the Global Terrorism Database, the findings from interrupted time series models suggest a complex set of displacement and diffusion effects with respect to alternative attack modes, target types, and weapon usage.

Adolescent Victimization and Violent Self-Help

A generation of criminological research demonstrates considerable overlap between victim and offender populations. Although there is compelling theoretical and empirical evidence that criminal offenders live a high-risk lifestyle that exposes them to a higher likelihood of becoming victims of crime themselves, we take as a point of departure the possibility that an individual’s experiences as a crime victim might also motivate them to engage in certain forms of violent behavior as a form of ‘‘self-help.’’ In this study, violent self-help is conceptualized to encompass gang membership, handgun carrying, and aggravated assault. An analysis of data from a nationally representative sample of adolescents (12 years of age at the initial interview) provides support for the proposition that experienced and vicarious victimization are precursors to later violent behavior, even among youth with no history of violent behavior.

Instrumental Variables in Criminology and Criminal Justice

Instrumental variables estimation is an econometric technique that is commonly employed by economists to overcome the problem of endogeneity in a causal variable of interest. It is a method that could be of some use to criminologists who also frequently confront simultaneity, measurement error, and selection bias. The method is sometimes referred to as a natural experiment because, like a classical experiment, it resolves these problems by rendering variation in the key independent variable exogenous or uncorrelated with the error term. It does so through the introduction of a variable that is correlated with the causal variable of interest but is otherwise uncorrelated with the outcome other than the one through the causal variable. This exclusion restriction is the key to causal identification and must be defended on substantive and theoretical grounds, not necessarily statistical ones. Following an intuitive description of the method, a short empirical example is provided, along with guidance about common pitfalls and potential problems with the method. Researchers interested in a more technical treatment of the method are pointed to accessible treatments in economics.