|
Session Title: Crime, Violence and IRT/Rasch Measurement
|
|
Panel Session 128 to be held in International Room on Wednesday, November 7, 4:30 PM to 6:00 PM
|
|
Sponsored by the Quantitative Methods: Theory and Design TIG
|
| Chair(s): |
| Susan Hutchinson,
University of Northern Colorado,
susan.hutchinson@unco.edu
|
| Discussant(s):
|
| Michael Dennis,
Chestnut Health Systems,
mdennis@chestnut.org
|
| Abstract:
This panel presents a variety of applications of the Rasch measurement model in the assessment of crime and violence (Global Appraisal of Individual Need, Dennis, 2003). Maps of crime and violence items and persons are presented that illustrate the symptom hierarchies on a linear, interval yardstick. Atypical cases are profiled, e.g., those who endorse very little crime and violence except some of the more severe instances such as prostitution, forgery and rape. Differential item functioning for adults vs. adolescents and men vs. women are presented. Each of these examples deals with methods to improve the reliability, validity and fairness of the measures we use to evaluate programs.
|
|
Interpreting Maps and Statistics of the Crime and Violence Scale
|
| Karen Conrad,
University of Illinois, Chicago,
kmconrad@uic.edu
|
| Barth Riley,
University of Illinois, Chicago,
bbriley@chestnut.org
|
| Ya-Fen Chan,
Chestnut Health Systems,
ychan@chestnut.org
|
| Kendon Conrad,
University of Illinois, Chicago,
kjconrad@uic.edu
|
| Michael Dennis,
Chestnut Health Systems,
mdennis@chestnut.org
|
|
In the 1930's Thurstone used his new scaling method to estimate the seriousness of crimes. In this presentation, Rasch analysis will be used to estimate the seriousness of the 31 crime and violence items of the Crime and Violence Scale (CVS) of the Global Appraisal of Individual Need (Dennis et al., 2003). Including a brief introduction to Rasch analysis, the presentation will include discussions of person and item reliability of the CVS. Then it will focus on the person/item map that depicts distributions of persons and items on the Rasch ruler of crime and violence. The discussion will include a comparison of the Rasch results to Thurstone's findings as well as to other theoretical expectations regarding the seriousness of various crimes.
|
|
|
Crime and Violence: Differential Item Functioning by Age and Gender
|
| Ya-Fen Chan,
Chestnut Health Systems,
ychan@chestnut.org
|
| Karen Conrad,
University of Illinois, Chicago,
kmconrad@uic.edu
|
| Kendon Conrad,
University of Illinois, Chicago,
kjconrad@uic.edu
|
| Barth Riley,
University of Illinois, Chicago,
bbriley@chestnut.org
|
| Michael Dennis,
Chestnut Health Systems,
mdennis@chestnut.org
|
|
Criminal and violent behaviors are prevalent among people presenting to substance abuse treatment. The specific types of crime and violence committed, however, might vary by gender and age. Using a Rasch differential item functioning test (DIF) this study aims to examine whether the severity of criminal and violent behaviors differed by gender and age among adolescents and adults who were treated for substance abuse. The DIF test showed that males and females differed significantly in their endorsement of items related to prostitution, illegal gambling, physical violence (slapped someone), and armed theft. Furthermore, adults were found to be more likely to endorse prostitution, rape, forgery, arson, property crime and DUI than youths. The presentation will demonstrate how measurement models can improve our understanding of crime and violence among substance users in treatment as well as how it can impact program planning.
| |
|
Detecting Misfitting Persons with the Rasch Model
|
| Barth Riley,
University of Illinois, Chicago,
bbriley@chestnut.org
|
| Ya-Fen Chan,
Chestnut Health Systems,
ychan@chestnut.org
|
| Karen Conrad,
University of Illinois, Chicago,
kmconrad@uic.edu
|
| Kendon Conrad,
University of Illinois, Chicago,
kjconrad@uic.edu
|
| Michael Dennis,
Chestnut Health Systems,
mdennis@chestnut.org
|
|
The GAIN Crime and Violence Scale (CVS) consists of four subscales: General Conflict Tactics, Property Crimes, Interpersonal Crimes, and Drug Crimes. Rasch principal components analysis of person residuals and person fit statistics were examined in order to determine the extent to which atypical patterns of responding to the CVS was present in a sample of 7,435 substance abuse treatment clients. Principal components analysis identified five components, with the first component explained 10.9% of residual variance. Among persons who loaded negatively on the first component (n=2,334, 31.4% of total sample), half (50.3%) had response patterns characterized by high outfit (MNSQ Outfit > 1.33). The average measure on the CVS for this subgroup (n=1,175) = -1.66 logits. Persons with misfitting response patterns tended to endorse previous involvement in prostitution (item difficulty=2.34 logits) or indicated that they had committed forgery (item difficulty=1.91 logits) without endorsement of less severe forms of crime and violence.
| |