Date of Award
Master of Science (MS)
Amy Claxton Kallam
The current study utilizes data collected from a rural mental health center located in Great Bend, Kansas. Participants (N=103) were charged with DUI in Barton County between the period August 1995 to February 1997. DUI offenders that were charged again within three years of their DUI within this period were classified as recidivist. A variety of variables such as age, sex, face, marital status, prior criminal offenses, education level, results on substance abuse psychometric instruments, Blood Alcohol Content and employment status were studied utilizing logistic regression. To predict recidivism, a complex model was developed utilizing logistic regression. The developed model will assessed for accuracy with regards to the predicting the actual data. It was hypothesized based on previous research that BAL and previous criminal offenses would have considerable predictive value for identifying subjects that reoffend within the three-year monitoring period. In addition, multiple linear regression analysis was performed to examine possible relationships between psychometric subscales and DUI recidivism. Logistic regression was used to test for significance for each independent variable individually. The results indicate employment status, family income, years of education, and marital status were variables statistically significant in relationship to DUI recidivism. Each of these four variables was entered simultaneously in logistic regression to construct a linear equation to predict DUI recidivism. Years of education, marital status and employment status was used in the linear equation construction and in that order of importance based on the calculated R. The derived linear equation correctly predicted all study participants as recidivist or non-recidivist 67.37% of the time. Recidivist subjects were correctly identified 57.14% when utilizing the linear equation produced through logistic regress. The study also utilized multiple linear regression for six of the eight SASSI scales. None of the scales were found to be significantly related DUI recidivism.
Copyright 2000 Douglas C. Diel
Diel, Douglas C., "Predicting Dui Recidivism : Utilizing Factors Available to a Rural Mental Health Center" (2000). Master's Theses. 2773.