M.J.L.F. Cruyff

Juvenile Delinquency and Gender

This article considers differences in patterns of youth delinquency and problem behaviour between boys and girls. It uses cross-sectional surveys of self-reported youth offending in 11 European countries, and a similar survey covering various ethnic groups in Rotterdam, both carried out in 1992. These surveys show that there remains a substantial gap in the level of delinquency between girls and boys across all countries and ethnic groups. The findings confirm that weak social controls by family and school are an important correlate of delinquency for males and females in all country clusters and across all ethnic groups. On the whole, the correlates of delinquency are found to be similar in males and females, which suggests that there is no need for a different theory to explain delinquency in boys and girls. Social control explains part of the gap in delinquency between boys and girls, simply because social controls of girls tend to be stronger and tighter. Culturally determined differences in the strength of family-based social controls can also explain some of the variation in delinquency between ethnic groups.

Estimating the Size of a Criminal Population from Police Records Using the Truncated Poisson Regression Model

The truncated Poisson regression model is used to arrive at point and interval estimates of the size of two offender populations, i.e. drunk drivers and persons who illegally possess firearms. The dependent capture–recapture variables are constructed from Dutch police records and are counts of individual arrests for both violations. The population size estimates are derived assuming that each count is a realization of a Poisson distribution, and that the Poisson parameters are related to covariates through the truncated Poisson regression model. These assumptions are discussed in detail, and the tenability of the second assumption is assessed by evaluating the marginal residuals and performing tests on overdispersion. For the firearms example, the second assumption seems to hold well, but for the drunk drivers example there is some overdispersion. It is concluded that the method is useful, provided it is used with care.

Log-linear randomized-response models taking self-protective response behavior into account

Randomized response (RR) is an interview technique designed to eliminate response bias when sensitive questions are asked. In RR the answer depends partly on the true status of the respondent and partly on the outcome of a randomizing device. Although RR elicits more honest answers than direct questions do, it is susceptible to self-protective response behavior; that is, the respondent gives an evasive answer irrespective of the outcome of the randomizing device. The authors present a log-linear RR model that accounts for this kind of self-protection (SP). The main results of this SP model are estimates of (1) the probability of SP, (2) the log-linear parameters describing the associations between the sensitive characteristics, and (3) the prevalence of the sensitive characteristics that are corrected for SP. The model is illustrated with two examples from a Dutch survey measuring noncompliance with social welfare rules.

Point and interval estimation of the population size using a zero-truncated negative binomial regression model

This paper presents the zero-truncated negative binomial regression model to estimate the population size in the presence of a single registration file. The model is an alternative to the zero-truncated Poisson regression model and it may be useful if the data are overdispersed due to unobserved heterogeneity. Horvitz-Thompson point and interval estimates for the population size are derived, and the performance of these estimators is evaluated in a simulation study. To illustrate the model, the size of the population of opiate users in the city of Rotterdam is estimated. In comparison to the Poisson model, the zero-truncated negative binomial regression model fits these data better and yields a substantially higher population size estimate.

People born in the Middle East but residing in the Netherlands: invariant population size estimates and the role of active and passive covariates

Including covariates in loglinear models of population registers improves population size estimates for two reasons. First, it is possible to take heterogeneity of inclusion probabilities over the levels of a covariate into account; and second, it allows subdivision of the estimated population by the levels of the covariates, giving insight into characteristics of individuals that are not included in any of the registers. The issue of whether or not marginalizing the full table of registers by covariates over one or more covariates leaves the estimated population size estimate invariant is intimately related to collapsibility of contingency tables [Biometrika 70 (1983) 567–578]. We show that, with information from two registers, population size invariance is equivalent to the simultaneous collapsibility of each margin consisting of one register and the covariates. We give a short path characterization of the loglinear model which describes when marginalizing over a covariate leads to different population size estimates. Covariates that are collapsible are called passive, to distinguish them from covariates that are not collapsible and are termed active. We make the case that it can be useful to include passive covariates within the estimation model, because they allow a finer description of the population in terms of these covariates. As an example we discuss the estimation of the population size of people born in the Middle East but residing in the Netherlands

Accounting for self-protective responses in randomized response data from a social security survey using the zero-inflated Poisson model

In 2004 the Dutch Department of Social Affairs conducted a survey to assess the extent of noncompliance with social security regulations. The survey was conducted among 870 recipients of social security benefits and included a series of sensitive questions about regulatory noncompliance. Due to the sensitive nature of the questions the randomized response design was used. Although randomized response protects the privacy of the respondent, it is unlikely that all respondents followed the design. In this paper we introduce a model that allows for respondents displaying self-protective response behavior by consistently giving the nonincriminating response, irrespective of the outcome of the randomizing device. The dependent variable denoting the total number of incriminating responses is assumed to be generated by the application of randomized response to a latent Poisson variable denoting the true number of rule violations. Since self-protective responses result in an excess of observed zeros in relation to the Poisson randomized response distribution, these are modeled as observed zero-inflation. The model includes predictors of the Poisson parameters, as well as predictors of the probability of self-protective response behavior.

Capture-recapture to estimate criminal populations

Methodology is presented that allows to estimate the size of population from a single register, such as a police register of offenders. A capture-recapture variable is constructed from Dutch police records and is a count of the police contacts for a violation. A population size estimate is derived assuming that each count is a realization of a Poisson distribution and that the Poisson parameters are related to covariates through the truncated Poisson regression model or variants of this model. As an example, estimates for perpetrators of domestic violence are presented. It is concluded that the methodology is useful, provided it is used with care.

Estimating the prevalence of crack dependence using capture-recapture with institutional and field data: a three-city study in The Netherlands.

The aim of this study was to estimate the prevalence of crack dependence in the three largest Dutch cities (Amsterdam, Rotterdam, The Hague), stratified by gender and age. Three-sample capture-recapture, using data (collected between 2009 and 2011) from low threshold substitution treatment (n = 1,764), user rooms (n = 546), and a respondent-driven sample (n = 549), and applying log-linear modeling (covariates: gender, age, and city), provided a prevalence rate of 0.51% (95% CI: 0.46%–0.60%) for the population aged 15–64 years, with similar estimates for the three cities. Females (23.0% of total estimate) and younger crack users (12.8% aged <35 years) might be underrepresented in drug user treatment services.

The multidimensional randomized response design: Estimating different aspects of the same sensitive behavior

The conventional randomized response design is unidimensional in the sense that it measures a single dimension of a sensitive attribute, like its prevalence, frequency, magnitude, or duration. This paper introduces a multidimensional design characterized by categorical questions that each measure a different aspect of the same sensitive attribute. The benefits of the multidimensional design are (i) a substantial gain in power and efficiency, and the potential to (ii) evaluate the goodness-of-fit of the model, and (iii) test hypotheses about evasive response biases in case of a misfit. The method is illustrated for a two-dimensional design measuring both the prevalence and the magnitude of social security fraud.

Sensitivity analysis and calibration of population size estimates obtained with the zero-truncated Poisson regression model

Zero-truncated regression models for count data can be used to estimate the size of an elusive population. A frequently encountered problem is that the Poisson model underestimates the population size due to unobserved heterogeneity, while the negative binomial model is not identified. A sensitivity analysis using the negative binomial model with fixed dispersion parameter might provide inside in the robustness of the population size estimate against unobserved heterogeneity, but as yet there is no method to determine realistic values for the dispersion parameter. This article introduces an R-squared measure and the use of the Pearson dispersion statistic to alleviate this problem. As a spin-off, a method is proposed for calibration of population size estimates in monitoring studies where the number of covariates varies over the measurement occasions. The performance of these methods is evaluated in simulation studies, and is illustrated on a population of drunk drivers.