Sentencing the Federal Drug Offender - An Experiment in Computer-Aided Sentencing

• NCJ Number: 90950
• Author(s): I Potas; J Walker
• Date Published: 1983
• Length: 119 pages
• Corporate Author: Australian Institute of Criminology
• Sale Source: Australian Institute of Criminology; National Institute of Justice/
• Language: English
• Country: Australia

Annotation

Mathematical models for sentencing Federal drug offenders in Australia not only describe, but also prescribe, sentences with a remarkable degree of accuracy. The neat fit between cases analyzed and the models suggests that sentencing disparity involving drug offenses is not as serious a problem as imagined.

Abstract

A careful review of the 1980 Australian Law Reform Commission's Sentencing Report does not support its conclusion that a disparity problem exists, but does reveal the inadequacy of the system for gathering data on which to base a reliable evaluation. The method detailed in this report for identifying and measuring disparities has three stages: it first identifies factors that appear to consistently impact sentencing decisions and measures the extent of this influence; applies this information to the circumstances of each case to derive a notional or predicted sentence; and finally, calculates the difference between the actual sentence and the notional sentence. The study used a computer to analyze 253 drug offense convictions under section 233B Custom Act offenses and materials from a previous study profiling Federal prisoners. Judges appeared to use one small group of variables to determine severity level in sentencing, while the totality of information was used to fine-tune the precise sentences. They followed a similar process in determining the nonparole period. Three key factors effectively determined the basic sentence: drug quantity, prior criminal record, and rehabilitation prospects. The multiplicative model predicts 2.3 years for lenient decisions, 5.2 years for normal cases, and 10.2 years for cases demanding severe treatment. Correlations between the model's sentences and actual sentences handed down are sufficiently high to indicate that the model adequately describes the decisionmaking process. These models could be used as guidelines, particularly for inexperienced sentencers, or as monitoring devices. The document provides tables, graphs, 12 references, worksheet forms, and coding schedules. (Author abstract modified)