Analyzing Multiple-Item Measures of Crime and Deviance I: Item Response Theory Scaling

Multiple-item measures of self-reported offending typically provide the principal outcome measures for individual level research on the causes of crime and deviance. This article directs attention to the substantial problems presented by the task of forming composite scores for these measures, and it presents a possible solution to those problems. We consider scaling by means of the graded response model from item response theory (IRT) as a potential means of overcoming the shortcomings of traditional summative scaling and of obtaining valuable information about the strengths and weaknesses of our measures. We illustrate this strategy through a scale analysis of a fourteen-item, self-report measure of delinquency, using three years of data from the Monitoring the Future study, an annual national survey of high school seniors. The graded response model proves to be consistent with the data, and it provides results that address important substantive questions about self-report measures. The findings are informative about the strengths and weaknesses of alternative strategies for developing self-report instruments, indicating that there is little to be gained by making fine distinctions in the frequency of individual delinquent acts.     

The association between psychopathic personality traits and health-related outcomes

Purpose

Psychopathy and psychopathic personality traits (PPT) have been linked to a long list of negative life outcomes. To date, however, few studies have provided a systematic analysis of whether psychopathic personality traits contribute to increased health burden. The current study was designed to address this gap in the literature.

Method

This study analyzed data from the National Longitudinal Study of Adolescent Health and employed a measure of PPT derived from the five-factor model of personality. Analyses were conducted using OLS, logistic, and Poisson regression techniques.

Results

The results revealed that relatively higher scores on psychopathic personality traits were associated with a slight increase in a wide range of negative health outcomes. These significant associations were detected for both males and females.

Conclusions

We speak to the importance of these findings for the potential to reduce health burden among psychopaths and those who score relatively high on measures of psychopathic personality traits.     

Reassessing the reliability and validity of self-report delinquency measures

Several issues related to the reliability and validity of self-report delinquency measures are raised and discussed. These include problems associated with the use of internal consistency as the measure of reliability, the level of reliability or precision required for different types of analyses, problems with the content validity of self-report measures, problems of overreporting and underreporting, problems with the use of official records as a validity check on self-reports, and the lack of any good criterion as a major obstacle in assessing the empirical validity of self-report measures. In the light of these problems, some cautions about the use of self-report measures are made.     

Measuring Low Self-Control and Reactive Criminal Thinking in the NLSY–Child Sample: One Construct or Two?

The purpose of this study was to determine whether a behavioral rating measure of low self-control and an attitudinal measure of low self-control can be viewed as measuring the same construct. It was hypothesized that the externalizing scale of the Behavior Problems Index (BPI-Ext), which served as a behavioral rating measure of low self-control in the current study, would display greater similarity to a 6-item self-report of antisocial, but not necessarily delinquent, behavior (SR-AB) measure than it would a 6-item attitudinal self-report measure of low self-control, labeled the reactive criminal thinking (SR-RCT) scale. This study was conducted on a sample of 6280 children (3144 boys, 3136 girls) from the National Longitudinal Survey of Youth-Child (NLSY-C). A pair of confirmatory factor analyses revealed that the BPI-Ext and SR-RCT scales appeared to form two distinct constructs. In addition, the BPI-Ext correlated significantly better with the SR-AB than with the SR-RCT and the BPI-Ext and SR-AB achieved moderate negative correlations with measures of attention, concentration, achievement, and general aptitude, whereas the SR-RCT achieved small positive correlations. These results indicate that behavioral and attitudinal measures of low self-control are measuring different constructs, the former impulsive behavior and the latter reactive criminal thinking.     

The Self-Appraisal Questionnaire: a self-report measure for predicting recidivism versus clinician-administered measures: a 5-year follow-up study

In this study, the effectiveness of the Self-Appraisal Questionnaire (SAQ), a self-report measure for predicting release outcome, is examined as compared to clinician-administered and widely used risk prediction measures, over a 5-year period. The SAQ was administered along with four similar, but clinician-administered, measures to 91 federally sentenced Canadian male offenders prior to their release to the community. Follow-up data were collected for a 60-month period. Outcome criteria measures were violent and general recidivism. Results indicated that the SAQ is at least as effective as the four other measures in predicting postrelease outcome. The advantages of using the SAQ as a self-report measure as opposed to clinician-administered measures are discussed.     

Predicting overt and cyber stalking perpetration by male and female college students

In this study, self-report student surveys on early childhood maltreatment, attachment styles, alcohol expectancies, and narcissistic personality traits are examined to determine their influence on stalking behavior. Two subtypes of stalking were measured using Spitzberg and Cupach's (2008) Obsessive Relational Intrusion: cyber stalking (one scale) and overt stalking (comprised of all remaining scales). As t tests indicated that men and women differed significantly on several variables, OLS regression models were run separately for men (N = 807) and women (N = 934). Results indicated that childhood sexual maltreatment predicted both forms of stalking for men and women. For men, narcissistic vulnerability and its interaction with sexual abuse predicted stalking behavior (overt stalking R2 = 16% and cyber stalking R2 = 11%). For women, insecure attachment (for both types of stalking) and alcohol expectancies (for cyber stalking) predicted stalking behavior (overt stalking R2 = 4% and cyber stalking R2 = 9%). We discuss the methodological and policy implications of these findings.     

A framework for estimating probability of a match in forensic bite mark identification

In forensic dentistry, a human expert typically does the comparison and identification based on bite marks. Unlike DNA analysis, however, there is no quantitative basis with which to assign a probability for this given match. This paper proposes a framework for empirically estimating the probability of such a bite mark match and reports on initial experimental results. The methodology involved collection of dental population data (3D dental casts and bite mark images), image analysis for quantitatively measuring the degree of match (based on chamfer distance), and performing a logistic regression analysis using the collected population data to estimate the probability of match from the calculated degree of match. The model correctly predicted 35 of the 42 matches and 585 of the 588 mismatches. The method also has potential for use in other forensic applications in which the assignment of quantitative probabilities is important.     

Social Class and Delinquent Behavior in a National Youth Panel: 1976–1980

This article addresses the question of whether or not a more adequate measure of self-reported delinquency applied to a representative national sample would reveal class differences in r l e l i m that have not been found in earlier self-report studies. The methodological criticisms of earlier self-report measures are reviewed, a new self-report measure is described, annual sex-by-class-specific prevalence and incidence rates based on this measure are presented for a national youth panel for the years 1976 through 1980, and the implications of the class findings are discussed. Class differences in both prevalence and incidence are found for serious offenses. For males, class differences are also found in the incidence of nonserious offenses and global delinquency Class differences are more pervasive and stronger when using an incidence as opposed to a prevalence measure. Criticisms of earlier self report measures appear Justified, calling into Question conclusions about the distribution of delinquency in the adolescent population which are based on prim self-report data.     

A randomized experiment of a prisoner reentry program: updated results from an evaluation of the Minnesota Comprehensive Offender Reentry Plan (MCORP)

Using a randomized experimental design, this study evaluated the effectiveness of the Minnesota Comprehensive Offender Reentry Plan (MCORP), a prisoner reentry pilot project implemented in 2008. In an effort to reduce recidivism, MCORP attempted to increase offender access to community services and programming by producing greater case management collaboration between caseworkers in prison and supervision agents in the community. Results from Cox regression models showed that MCORP significantly reduced four of the five recidivism measures examined, although the size of the reduction in hazard ratios was relatively modest (20–25%). The findings further suggested that MCORP reduced costs. Sensitivity analyses showed, however, that the cost avoidance estimates were not robust across all assumptions that were examined.     

On using gait in forensic biometrics

Given the continuing advances in gait biometrics, it appears prudent to investigate the translation of these techniques for forensic use. We address the question as to the confidence that might be given between any two such measurements. We use the locations of ankle, knee, and hip to derive a measure of the match between walking subjects in image sequences. The Instantaneous Posture Match algorithm, using Harr templates, kinematics, and anthropomorphic knowledge is used to determine their location. This is demonstrated using real CCTV recorded at Gatwick International Airport, laboratory images from the multiview CASIA-B data set, and an example of real scene of crime video. To access the measurement confidence, we study the mean intra- and inter-match scores as a function of database size. These measures converge to constant and separate values, indicating that the match measure derived from individual comparisons is considerably smaller than the average match measure from a population.     

A test of recent subcultural explanations of White violence in the United States

Recent subcultural theories suggest that real and perceived status loss among some White males as a result of social change during the last four decades-such as increasing gender and racial equity and agricultural job loss-added to the traditional patriarchal and masculine characteristics of this group to create cultural adaptations that led to higher levels of violence. One would thus expect higher homicide rates in areas where these traits are more fully developed and/or where status threats are more strongly felt. To test the efficacy of these claims, this study used OLS regression, and (1) employed census 2000 data and both homicide victimization and arrest data to update and replicate a common model of social structure and homicide, and then (2) added several variables to this baseline model that represented the subcultural traits, conditions, and behaviors outlined by these theories. The analyses yielded little support for these ideas. Thus while they might be appropriate for explaining the variation of the strength of specific subcultures and types of violence (e.g., Far Right wing paramilitary groups and political violence), an extension of these theories to more general interpersonal violence did not appear warranted.     

Assessing Readiness to Change among Perpetrators of Intimate Partner Violence: Analysis of Two Self-report Measures

Recent interest in the construct of motivation to change among male perpetrators of intimate partner violence (IPV) has led to development of two self-report measures of this construct: the University of Rhode Island Change Assessment for Domestic Violence (URICA-DV; Levesque et al. 2000) and the Safe at Home Inventory (SAH; Begun et al. 2003). We examined the internal structure of these instruments using confirmatory and exploratory factor analyses in a sample of 199 male IPV perpetrators waiting to begin court-mandated batterer intervention programs. Results indicated that the scales had adequate internal consistency. However, using confirmatory factor analyses, the hypothesized structural models of each instrument fit the data poorly. Subsequent models suggested by a series of exploratory factor analyses failed to improve model fit substantially for both scales. For the URICA-DV, a 3-factor model including items from the Precontemplation, Contemplation, and Action subscales fit the data adequately. No acceptable model fit could be found for the SAH. Intercorrelations among similar subscales across the two inventories suggested only a modest degree of intercorrelation (r’s = 0.16–0.52), and a factor structure more dependent on scale of origin than item latent content, suggesting that the scales assess different facets of the readiness to change construct. These results suggest that before the stages of change construct can be fully applied to the IPV area, more research and evaluation are needed on how to accurately assess abuser readiness to change.     

POVERTY,INFANT MORTALITY,AND HOMICIDE RATES IN CROSS-NATIONAL PERSPECTIVE: ASSESSMENTS OF CRITERION AND CONSTRUCT VALIDITY*

This article assesses the extent to which the infant mortality rate might be treated as a “proxy” for poverty in research on cross-national variation in homicide rates. We have assembled a pooled, cross-sectional time-series data set for 16 advanced nations from the 1993–2000 period that includes standard measures of infant mortality and homicide and contains information on the following commonly used “income-based” poverty measures: a measure intended to reflect “absolute” deprivation and a measure intended to reflect “relative” deprivation. With these data, we assess the criterion validity of the infant mortality rate with reference to the two income-based poverty measures. Also, we estimate the effects of the various indicators of disadvantage on homicide rates in regression models, thereby assessing construct validity. The results reveal that the infant mortality rate is correlated more strongly with “relative poverty” than with “absolute poverty,” although much unexplained variance remains. In the regression models shown here, the measure of infant mortality and the relative poverty measure yield significant positive effects on homicide rates, whereas the absolute poverty measure does not exhibit any significant effects. The results of our analyses suggest that it would be premature to dismiss relative deprivation in cross-national research on homicide, and that disadvantage is conceptualized and measured best as a multidimensional construct.     

Decisions without consequences: cost control and access in state Medicaid programs.

States have implemented a number of strategies to provide services, pay providers, and control Medicaid spending. We test the effects of some differences in state Medicaid policies on program enrollees' access to and use of health care services. Logistic and OLS regression analyses of cross-sectional data indicate that these policies exert significant influences on enrollees' access to health services but have a weaker direct effect on their use of them. However, we find evidence that utilization is affected indirectly (through increased access) by state policy decisions. Somewhat surprisingly, Medicaid policies designed to contain costs by limiting utilization appear to affect neither access nor utilization. Medicaid enrollees have greater access to a private physician in states with higher physician reimbursement and additional Medicare insurance for their enrollees. Other nonpolicy variables with pronounced impacts on access to private office physicians include race and the availability of private insurance.     

THE INTERACTION OF ANTISOCIAL PROPENSITY AND LIFE‐COURSE VARYING PREDICTORS OF DELINQUENT BEHAVIOR: DIFFERENCES BY METHOD OF ESTIMATION AND IMPLICATIONS FOR THEORY*

Recent criminological research has explored the extent to which stable propensity and life‐course perspectives may be integrated to provide a more comprehensive explanation of variation in individual criminal offending. One line of these integrative efforts focuses on the ways that stable individual characteristics may interact with, or modify, the effects of life‐course varying social factors. Given their consistency with the long‐standing view that person–environment interactions contribute to variation in human social behavior, these theoretical integration attempts have great intuitive appeal. However, a review of past criminological research suggests that conceptual and empirical complexities have, so far, somewhat dampened the development of a coherent theoretical understanding of the nature of interaction effects between stable individual antisocial propensity and time‐varying social variables. In this study, we outline and empirically assess several of the sometimes conflicting hypotheses regarding the ways that antisocial propensity moderates the influence of time‐varying social factors on delinquent offending. Unlike some prior studies, however, we explicitly measure the interactive effects of stable antisocial propensity and time‐varying measures of selected social variables on changes in delinquent offending. In addition, drawing on recent research that suggests that the relative ubiquity of interaction effects in past studies may be partly from the poorly suited application of linear statistical models to delinquency data, we alternatively test our interaction hypotheses using least‐squares and tobit estimation frameworks. Our findings suggest that method of estimation matters, with interaction effects appearing readily in the former but not in the latter. The implications of these findings for future conceptual and empirical work on stable propensity/time‐varying social variable interaction effects are discussed.     

INTERFACING INDICATORS AND CONSTRUCTS IN CRIMINOLOGICAL RESEARCH: A NOTE ON THE COMPARABILITY OF SELF-REPORT VIOLENCE DATA FOR RACE AND SEX GROUPS

This paper examines issues in using self-reports of involvement in violent acts to measure a latent construct (propensity toward violence). Three issues are addressed. First, the meaning of latent variables and the factor-analytic approach to modeling latent constructs are discussed. Second, the comparability of self-report violence measures across race and sex groups is assessed. Third, the consequences are discussed of estimating factor analytic models on pooled samples of data where the measurement component of the model does not apply equally for different subsets of the sample.     

An empirical assessment of the AIDS Risk Reduction Model (ARRM) employing ordered probit analyses

The AIDS Risk Reduction Model (ARRM) is a conceptual model designed to explain the influence of knowledge and attitudes about AIDS/HIV on risk-related behaviors. This article measures the influence of the primary concepts associated with the ARRM (labeling, commitment, and enactment) on high-risk adolescent's knowledge, attitudes, and self-reported behaviors. Data were collected from 393 adolescents incarcerated in juvenile detention facilities. Ordered probit and ordinary least squares (OLS) analyses were conducted to determine which of several independent variables have the most impact on acceptance of the ARRM. Results indicated that the ARRM is effective for understanding the factors that influence AIDS risk avoidance. Findings also indicate that previous life experiences should be formally integrated into the model. Moreover, specific forms of enactment are associated with different factors. Suggestions as to how the ARRM may prove useful for promoting behavioral change among high-risk adolescents are provided.     

Psychopathic Personality Traits and Delinquent and Risky Sexual Behaviors in Croatian Sample of Non-referred Boys and Girls

The current study investigated the association of psychopathic traits with violent and non-violent delinquency, delinquency versatility, and risky sexual behavior in Croatian sample of non-referred boys (n = 226) and girls (n = 480). Psychopathic traits were measured by the self-report Youth Psychopathic Traits Inventory (YPI). Consistent with the study’s hypotheses, Impulsive-Irresponsible and Callous-Unemotional dimensions were most consistently associated with all outcome measures. Results of the regression analyses showed that the Impulsive-Irresponsible behavioral style had stronger association with non-violent delinquency and delinquency versatility for boys. However, the Impulsive-Irresponsible dimension had stronger influence on risky sexual behavior for girls compared with boys. The results of a two-group confirmatory factor analysis revealed that the three-factor model of the YPI was invariant across gender.     

Design,development and validity testing of the gang affiliation risk measure (GARM)

This study aimed to create a measure of risk for gang affiliation, for use in the UK. A pilot stage invited gang affiliated and non-gang affiliated participants between the ages of 16–25 years to retrospectively self-report on 58 items of risk exposure at the age of 11 years. Based on performance of these items, a 26-item measure was developed and administered to a main study sample (n = 185) of gang affiliated and non-gang affiliated participants. Categorical Principal Component Analysis was applied to data, yielding a single-factor solution (historic lack of safety and current perception of threat). A 15-item gang affiliation risk measure (GARM) was subsequently created. The GARM demonstrated good internal consistency, construct validity and discriminative ability. Items from the GARM were then transformed to read prospectively, resulting in a test measure for predictive purposes (T-GARM). The T-GARM requires further validation regarding its predictive utility and generalisability. However, this study has resulted in the first measure of gang affiliation, with promising results.     

A latent structure analysis of delinquency

This paper uses latent structure analysis to develop a construct of delinquency based on a set of frequently used indicators of illegal behavior. Other methods, such as Guttman scaling and factor analytic procedures, are briefly reviewed as similar attempts to identify manifest (observed) variables that may act as indicators of a latent (unobserved or unobservable) variable, delinquency. The basic mathematical properties of the latent structure model employed in this paper are presented and then applied to a set of six self-report measures. The issue of external validity is also examined in a comparative analysis of an additive scale of delinquency and the latent class measure of delinquency. The latent class measure provides a considerable increase in the magnitude of association with such commonly studied correlates as attachment, peer delinquency, and instrumental attitudes toward the law.