Age, Criminal Careers, and Population Heterogeneity: Specification and Estimation of a Nonparametric, Mixed Poisson Model* (original) (raw)
Related papers
American Journal of Sociology, 1998
This article reviews questions about different categories of criminal careers, summarizes Poisson Iatent class regression models, describes procedures for evaluating the optimal number of latent classes, and applies this methodology to data from male cohorts taken from the cities of London, Philadelphia, and Racine. Four latent classes of offending careers is an appropriate number for the London cohort, but five classes can be justified for the Philadelphia data. In the case of the Racine cohorts, five classes may be detected for the 1942 and 1955 cohorts but only four for the 1949 cohort. Despite the varying numbers of latent offending classes, there clearly is a small number of typical age patterns. The Poisson distribution and Poisson regression models are fast becoming key methodological tools for the development of micromodels of delinquentlcriminal careers in the criminological literature (
Criminal Offending Frequency and Offense Switching
Journal of Contemporary Criminal Justice, 2004
A basic function of criminal career research is to describe patterns of criminal offending. During the history of the criminal career framework, researchers have divided the conceptual terrain of criminal offending into different dimensions. Two such dimensions include offending frequency and offense switching. Although research on both of these dimensions is extensive, there has been little investigation of their relationship to each other. In this article, the authors use juvenile police contact data from the 1945 Philadelphia birth cohort study to address this issue. In so doing, they extend an existing analytical framework that will allowthem to examine the joint distribution of these two variables. Two specific models are compared. The first model hypothesizes that the probability of switching between two categories of offenses (serious violent and all other offenses) will vary with individual offense frequency. The second model hypothesizes that the probability of offense swit...
Rethinking the "norm" of offender generality: Investigating specialization in the short-term
The life-course approach to criminal career research has devoted a good deal of attention to the generality or specialization of offending behavior. Typically, extant research demonstrates versatility on the part of offenders, yet such findings could be attributable, at least in part, to time and measurement aggregation bias. This work uses a temporally * The authors gratefully acknowledge both the guidance of Raymond Paternoster, under whose editorship this manuscript was initially submitted and the review process begun, and the helpful suggestions of the anonymous reviewers. Data for this study were accessed from the Inter-university Consortium for Political and Social Research (ICPSR). The original collector of the data, ICPSR, and the relevant funding agency bear no responsibility for use of this collection or for interpretations or inferences based on its use. Bibliographic Citation for Data Collection: Horney, Julie, and Ineke Haen Marshall. Crime Commission Rates Among
Australian and New Zealand Journal of Criminology, 2010
T his study examines the extent to which the positive association between past and future offending can be explained by either (un)observed population heterogeneity, or by state dependence. We use criminal career data derived from (1) a self-report survey administered to 2,900 Dutch individuals aged 15 years or older, and the Criminal Career and Life-course Study (CCLS) that has official data on a 4% (N = 4,684) sample of Dutch offenders convicted in 1977. Using hierarchical linear modelling we find both heterogeneity and state dependence partially explaining continuity in crime in both samples. In addition, we find that the impact of prior offending decreases over time and that that impact depends on the total number of previously committed offences.
Crime & Delinquency, 2011
Much of the knowledge base on offense specialization indicates that, although there is some (short-term) specialization, it exists amidst much versatility in offending. Yet this general conclusion is drawn on studies using very different conceptualizations of specialization and emerges with data primarily through the first two to three decades of life. Using data on a sample of Dutch offenders through age 72 years, this article introduces and applies a new method for studying individual offender specialization over the life course. The results indicate that although, in general, individual offending patterns over the life course are diverse, there is also evidence of an age-diversity curve. Linking offense frequency trajectories to the estimated diversity index, the authors also examine distinct specialization patterns across unique trajectory groups. Implications for theory and research are outlined. Crime Delinquency OnlineFirst,
Journal of Quantitative Criminology, 1999
An important theoretical problem for criminologists is an explanation forthe robust positive correlation between prior and future criminaloffending. Nagin and Paternoster (1991) have suggested that the correlationcould be due to time-stable population differences in the underlyingproneness to commit crimes (population heterogeneity) and/or thecriminogenic effect that crime has on social bonds, conventionalattachments, and the like (state dependence). Because of data andmeasurement limitations, the disentangling of population heterogeneityand state dependence requires that researchers control for unmeasuredpersistent heterogeneity. Frequently, random effects probit models havebeen employed, which, while user-friendly, make a strong parametricassumption that the unobserved heterogeneity in the population follows anormal distribution. Although semiparametric alternatives to the randomeffects probit model have recently appeared in the literature to avoid thisproblem, in this paper we return to reconsider the fully parametric model. Viasimulation evidence, we first show that the random effects probit modelproduces biased estimates as the departure of heterogeneity from normalitybecomes more substantial. Using the 1958 Philadelphia cohort data, we thencompare the results from a random effects probit model with a semiparametricprobit model and a fixed effects logit model that makes no assumptions aboutthe distribution of unobserved heterogeneity. We found that with this dataset all three models converged on the same substantive result—evenafter controlling for unobserved persistent heterogeneity, with models thattreat the unobserved heterogeneity very differently, prior conduct had apronounced effect on subsequent offending. These results are inconsistentwith a model that attributes all of the positive correlation between priorand future offending to differences in criminal propensity. Sinceresearchers will often be completely blind with respect to the tenabilityof the normality assumption, we conclude that different estimationstrategies should be brought to bear on the data.
THE EFFECTS OF LIFE CIRCUMSTANCES ON LONGITUDINAL TRAJECTORIES OF OFFENDING*
Criminology, 2005
This study, which is based on individual criminal careers over a 60year period, focuses on the development of criminal behavior. It first examines the impact that life circumstances such as work and marriage have on offending, then tests whether the effects of these circumstances are different for different groups of offenders, and finally examines the extent to which the age-crime relationship at the aggregate level can be explained by age-graded differences in life circumstances. Official data were retrieved for a 4-percent (N=4,615) sample of all individuals whose criminal case was tried in the Netherlands in 1977. Self-report data were derived from a nationally representative survey administered in the Netherlands in 1996 to 2,244 individuals aged 15 years or older. In analyzing this data, we use semi-parametric group-based models. Results indicate that life circumstances substantially influence the chances of criminal behavior, and that the effects of these circumstances on offending differ across offender groups. Age-graded changes in life circumstances, however, explain the aggregate age-crime relationship only to a modest extent. * We thank Daniel Nagin, John Laub, Rolf Loeber, Gerben Bruinsma, Catrien Bijleveld and Ray Paternoster and anonymous referees for numerous helpful suggestions. Please contact the authors at NSCR (Nederlands Studiecentrum Criminaliteit en Rechtshandhaving),
Crime and Young Men: The Role of Arrest, Criminal Experience, and Heterogeneity
2006
Using National Youth Survey (NYS) data, we examine the relationship of current criminal activity with past criminal activity and past arrests using an ordered probit model with unobserved heterogeneity. Past arrests raise current criminal activity only for the non-criminal type, while past criminal experience raises current criminal activity for both types. Also, the age crime profile peaks at age 18 for non-criminal type individuals, but for criminal type individuals, it continues to rise with age. Past research indicates that age arrest profiles rise till age 18 and then fall for both types, suggesting lower apprehension rates for criminal type individuals.