Studying the Effects of Incarceration on Offending Trajectories: An Information-Theoretic Approach

• NCJ Number: 216639
• Author(s): Avinash Singh Bhati
• Date Published: July 2006
• Length: 105 pages
• Grant Number(s): 2005-IJ-CX-0008
• Agencies: NIJ-Sponsored
• Publication Type: Report (Study/Research)
• Award: Life Interrupted: Studying the Effects of Incarceration on Criminal Trajectories

Annotation

This study developed a research method that used criminal history to determine whether, and to what extent, incarceration deterred offenders from reoffending; the method was applied to a "real-world" dataset.

Abstract

The use of criminal history prior to incarceration was found to be a reliable predictor of whether or not incarceration would deter reoffending within 3 years after release. Application of this model to the "real-world" dataset found that for 56 percent of the offender sample, incarceration had a deterrent effect, i.e., their postrelease rearrest pattern showed a downward departure from what would have been expected based on their criminal history before incarceration. Forty percent of the offenders conformed to the postrelease offending predicted from their criminal history before incarceration, i.e., incarceration interrupted (incapacitation) their criminal-career trajectories, but did not alter the expected normal trajectory predicted from prior criminal history. For 4 percent of offenders, incarceration increased postrelease reoffending beyond what was predicted from their criminal history. Being older at release and having a plateau of offending prior to incarceration increased the likelihood that incarceration deterred reoffending. Having more prior accumulated arrests and having a later age at first arrest were both found to decrease significantly the likelihood of incarceration's deterrent effect. Being released to supervision did not significantly deter reoffending. These findings suggest that the proposed analysis of criminal history prior to incarceration enables corrections practitioners to identify who is and is not likely to be deterred from postrelease reoffending. Recommendations for future research are offered. Dated arrest histories of a sample of approximately 38,000 prisoners released in 1994 from prisons in 15 States were used to test the research method. 31 tables, 8 figures, 56 references, and appended model estimates and sample SAS code