|
Abstract
|
VioLit summary:
OBJECTIVE:
The purpose of the research conducted by Visher et al. was to develop a model which predicts the recidivism of youthful offenders.
METHODOLOGY:
The authors employed a quasi-experimental, cross-sectional research design. The subjects used in their study were 1,949 randomly selected male youth (under age 25) who were released to parole in the state of California between July 1, 1981 and June 30, 1982. The subjects began their criminal activities at a young age (average age of 14.2 years) and more than 80% had 4 or more arrests.
The authors defined recidivism (or failure) as a subsequent arrest in California within 3 years of being released to parole. Eighty-eight percent of the subjects (1,710) were rearrested within this time frame. The dependent variable in their study was time to rearrest. They had over 30 independent variables grouped under the headings of criminal history, current commitment (offense type, length of commitment), substance abuse and school problems, family background, and environmental (community crime rate and crime clearance rate). Their analysis was based on a hazard model which predicted the likelihood that a given youth would fail in a certain time period given that he had not failed in a prior period. The model distinguished between youths who failed before 36 weeks and those who failed after 36 weeks.
The model was tested for accuracy using 4 subjects from the original sample: one with the highest probability of failing, one with the lowest, one from the middle of the probability distribution, and one subject whose hazard function showed the greatest variation over time. Data were analyzed using chi-square statistics.
FINDINGS/DISCUSSION:
The researchers found that the criminal history variables and variables describing the current commitment had a larger impact on the hazard function than the other independent variables. Other variables were also significantly related to risk of recidivism including substance abuse and school problems, family background variables (i.e., intrafamily abuse, parent criminality), and 6 out of the 7 variables which described county-level property and violent crime rates and clearance rates. The variables which were dropped from the model were the number of arrests, violence criminal history score, family violence, and parent criminality.
When the model was tested for accuracy, the researchers found "reasonably good agreement" between expected and observed recidivism for the 4 subjects at the end of 4 weeks as well as at the end of 36 weeks. However, there was a great disparity between observed and expected recidivism as time progressed toward three years with the hazard model overestimating failures. The authors also found that the model was useful when subjects were divided into high- and low-risk groups with results varying depending on where the line was drawn between the two groups.
AUTHORS' RECOMMENDATIONS:
The authors proposed that parole officers and correctional administrators could use their model to determine how to divide up the case load among officers such that the level of supervision required by a youthful offender could be determined by the risk level assigned to him/her.
(CSPV Abstract - Copyright © 1992-2007 by the Center for the Study and Prevention of Violence, Institute of Behavioral Science, Regents of the University of Colorado)
1980s
California
Recidivism Prediction
Offender Recidivism
Juvenile Offender
Juvenile Male
Juvenile Crime
Juvenile Delinquency
Juvenile Violence
Male Crime
Male Delinquency
Male Offender
Male Violence
Adult Crime
Adult Male
Adult Violence
Adult Offender
Offender Characteristics
Parole
Community Risk Factors
Family Risk Factors
Individual Risk Factors
Recidivism Causes
Recidivism Risk Factors
Crime Causes
Crime Risk Factors
Delinquency Causes
Delinquency Risk Factors
Violence Causes
Violence Risk Factors
03-05 |