Douglas A. Smith

Social Structure and Criminal Victimization

Using victimization data from 57 neighborhoods, this article examines the relationship between neighborhood characteristics and rates of violent crime and burglary. We argue that Shaw and McKays social disorganization theory provides a meaningful point of departure for examining the uneven distribution of criminal victimization across social units. Measures of three central theoretical elements in Shaw and McKays social disorganization perspective (poverty, residential mobility, and racial heterogeneity), and variables from the subculture of violence, social control, and opportunity perspectives are included in this research. Results indicate that core components of Shaw and McKays theory are important in explaining neighborhood victimization rates, although their influence is more conditional than direct and varies by type of crime. Evidence also emerges that supports social control models of aggregate criminal activity. Additionally, we examine whether parameters of macromodels of criminal activity vary across levels of urbanization. The implications of our findings for macrotheories of criminal activity are discussed.

Police Control of Interpersonal Disputes

This paper examines factors which influence police arrest decisions in interpersonal disputes. We analyze data from direct observation of police-citizen encounters and find that decisions are influenced by situational exigencies of these encounters and that the effects of certain situational variables depend upon the type of neighborhood in which disputes occur. For example, disputes involving cohabiting individuals are more likely to result in arrest in poor neighborhoods. Arrest decisions are also influenced by organizational characteristics of police agencies, such as bureaucratization and professionalism. We discuss the implications of our findings for a general theory of police dispute settlement.

Participation in and frequency of delinquent behavior: A test for structural differences

In this paper we address whether there are distinctive differences in the processes determining participation in offending vs frequency of offending. We develop a number of tests to examine not only whether the correlates of participation and frequency are similar but also whether the same underlying statistical model is consistent with the data on both these dimensions of a criminal career. The tests are applied using data from the first two waves of the National Youth Survey. While the findings are not without ambiguities, they do not support the idea that distinguishing among the dimensions of a criminal career is a fundamental requirement for conducting sound research on the causes of crime and delinquency.

Tobit models in social science research: Some limitations and a more general alternative

The use of tobit models to study censored and limited dependent variables has become increasingly common in applied social science research over the past two decades. Importantly, the likelihood function for a tobit model involves two distinct components: (1) the process that determines whether the outcome variable is fully observed or not and (2) the process that determines the score on the dependent variable for individuals whose outcome is fully observed. One limitation of the tobit model is its assumption that the processes in both regimes of the outcome are equal up to a constant of proportionality. In this article, the authors use Monte Carlo simulation evidence and an empirical example to illustrate the restrictive nature of this assumption and the consequences of disproportionality for the tobit model. They conclude that an alternative model proposed by Cragg should replace the tobit model as the estimator of first resort in situations such as those considered here.

Drug Use and Pretrial Misconduct in New York City

Is recent drug use significantly associated with pretrial misconduct? Does consideration of recent drug use enhance risk classification among a sample of persons who have time free pending the disposition of their cases? Using data on arrestees in Manhattan, this paper examines these issues and some related questions. To measure recent drug use, urine samples were collected from persons shortly after their arrest and tested for four drugs: heroin, cocaine, PCP, and methadone. Two measures of pretrial misconduct are considered: whether a defendant fails to appear for a scheduled court date (FTA) and whether a defendant is rearrested prior to case disposition. Censored probit models are used to estimate the statistical association between drug test results and pretrial misconduct. Results show that drug test results are significantly associated with pretrial misconduct over and above the information typically available to judges at the time release decisions are made. Some implications of these findings for pretrial decision making are discussed.

Latent-variable models in criminological research: Applications and a generalization of Joreskog's LISREL model

This paper examines issues in LISREL models involving latent, or unobserved, variables. We discuss estimating parameters in multiple-indicator models, various approaches to assessing goodness of fit, different types of linear measurement models, using LISREL to correct for attenuation in models involving single indicators with known reliabilities, and estimating parameters in structural equation models involving latent exogenous and endogenous variables. Finally, we present a generalization of the standard LISREL model which explicitly identifies implicit constraints on parameters in the standard LISREL model. We argue that the generalization can accommodate a wider range of models and provide additional diagnostic information on some assumptions in the standard LISREL model. By example, we show how the more general model can be estimated using Joreskogs LISREL program.

Applications and a Generalization of Mimic Models to Criminological Research

This article examines models involving multiple causes of a single latent endogenous variable. Such models, known as MIMIC models, combine psychometric and econometric features into a single unifying framework. Using an example concerning individuals subjective probability of victimization, we show how to estimate parameters in MIMIC models. Moreover, we present a formally equivalent generalization of such models that make explicit a number of implicit restrictions in standard MIMIC models. We show how to implement the more general MIMIC model using Joreskogs LISREL program and discuss the utility of such models to criminological research.