CEC Theses and Dissertations

Title

An Evaluation of the Consistency of Judicial Sentencing Systems that Incorporate Subjective Factors

Date of Award

2005

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Graduate School of Computer and Information Sciences

Advisor

Sumitra Mukherjee

Committee Member

Maxine S. Cohen

Committee Member

Michael J. Laszlo

Abstract

Greater consistency in sentencing decisions can be achieved by implementation of a judicial decision support system (JDSS). Maryland has sentencing guidelines that judges use for sentencing decisions in criminal cases. The existing system in Maryland uses a sentencing worksheet with non-subjective factors. A recommended sentencing guideline range is issued for each case based on points given for the non-subjective factors. This research investigated whether the incorporation of subjective factors in the Maryland sentencing guidelines, operating as a judicial decision support system, improved the consistency of sentencing decisions.

A JDSS prototype was developed, which incorporated subjective factors with non-subjective factors from the existing system for sentencing decisions. Consistency was measured by the percent of sentencing decisions that were within the recommended guideline range for the existing system based on non-subjective factors versus the percent of sentencing decisions that were within the recommended guideline range for the JDSS prototype based on non-subjective and subjective factors. A random sample of 2,944 cases from 1998 to 2003 was used for this research. Cases were limited to Category V, Felony Theft cases. Multiple regression linear analysis displayed a nonlinear relationship between the non-subjective factors and the sentencing decisions for the existing system.

Results also indicated a nonlinear relationship between the non-subjective factors, the subjective factors, and the sentencing decisions using the JDSS prototype. Decision tree inductive learning was then used with a random sample of 500 cases. Thirty percent of the cases were used to train the data set. Results using decision tree inductive learning indicated greater consistency for sentencing decisions using the JDSS prototype versus the existing system. The incorporation of subjective factors in the JDSS prototype improved the consistency of sentencing decisions.

This document is currently not available here.

  Link to NovaCat

Share

COinS