Program Evaluation in the Context of Supervision Regime Change: Motivational Interviewing in Colorado

Abstract

This study assesses the effectiveness of an evidence-based counseling strategy employed by Colorado parole officers called Motivational Interviewing (MI), which aims to reduce parolee recidivism. I employ a quasi-experimental research design in which the caseloads of MI-trained officers are divided into two sub-samples: parolees supervised before the achievement of MI certification and parolees supervised after certification. Regression analyses show that MI is associated with recidivism reduction, but this relationship would not have been revealed without accounting for a high-profile crime that interrupted the study period—the murder of the Executive Director of the Colorado Department of Corrections by a high-risk parolee that had absconded from supervision. This crime caused a public relations crisis for the parole agency, which responded by lowering tolerance for misconduct. This reduction in tolerance increased recidivism rates among parolees in Colorado, obscuring the effect of MI on parole outcomes. Theoretical and policy implications are discussed.

Keywords:

• parole
• supervision regimes
• community corrections
• recidivism
• program evaluation

Notes

¹ The Colorado Department of Corrections reviewed their institutional movement records and claim that while Ebel did spend the majority of his prison time in solitary confinement, he was not released directly from administrative segregation to parole (personal communication with the Office of Planning and Analysis, Colorado Department of Corrections, September 29, 2015).

² Capitalizations are from the original headlines. Headlines shown in all capital letters were fully capitalized in the Denver Post.

³ Decisions to create composite independent variables were based on the frequency of positive responses in the LSI data. That is, I created composite variables (criminal acquaintances or friends, moderate or severe emotional interference, current or past mental health treatment) so as to have an adequate number of positive responses to reliably analyze.

⁴ Collinearity diagnostics revealed no issues between the exposure variable and the MI certification variable. In a linear regression model predicting rearrest within 18 months, the exposure variable displayed a tolerance of .530 and a variance inflation factor of 1.888. In the same model, the MI certification variable displayed a tolerance of .535 and a variance inflation factor of 1.868.

⁵ As discussed earlier, the effect of MI on parolee recidivism may, to some degree, reflect the effects of other available programming in various Colorado parole offices.

Additional information

Notes on contributors

Jeffrey Lin

Jeffrey Lin is an associate professor in the Department of Sociology and Criminology at the University of Denver. He studies crime and punishment in the United States, with an emphasis on correctional policies, community corrections, sex offenders, juvenile justice, and media coverage.