PREDICTING SENTENCES IN FEDERAL COURTS - THE FEASIBILITY OF A NATIONAL SENTENCING POLICY

• NCJ Number: 33686
• Author(s): L P SUTTON; M J HINDELANG; M R GOTTFREDSON
• Date Published: 1978
• Length: 52 pages
• Grant Number(s): 75-SS-99-6001
• Sponsoring Agency: National Criminal Justice Information & Statistics Service
• Corporate Author: Hindelang Criminal Justice Research Ctr
• Sale Source: National Institute of Justice/
• Publication Type: Statistics
• Language: English
• Country: United States of America
• Note: ONE OF A SERIES OF REPORTS ON CRIMINAL SENTENCING FROM UTILIZATION OF CRIMINAL JUSTICE STATISTICS PROJECT

Annotation

THE PREDICTIVE QUALITY OF MULTIPLE REGRESSION ANALYSIS IS EXPLORED IN AN EFFORT TO SEE HOW ACCURATELY SENTENCES CAN BE PREDICTED ACROSS INDIVIDUAL FEDERAL COURT JURISDICTIONS.

Abstract

THIS FINAL REPORT IN A SERIES ON CRIMINAL SENTENCES IMPOSED IN FEDERAL DISTRICT COURTS SHOWS THAT LINEAR SENTENCING MODELS YIELDED BY MULTIPLE REGRESSION ANALYSIS ARE NOT ONLY GOOD SUMMARIES OF THE INTERPLAY OF SENTENCING FACTORS, BUT ALSO ARE SUCCESSFUL PREDICTORS OF SENTENCE FOR SUBSETS OF CASES OTHER THAN THOSE FROM WHICH THE PREDICTION MODELS WERE DERIVED. WHEN OFFENSE WAS CONTROLLED, THE ACCURACY OF PREDICTION WAS FOUND TO VARY ACCORDING TO THE JURISDICTION, THE SPECIFIC TIME INVOLVED, AND THE PARTICULAR OFFENSE IN QUESTION. NARCOTICS SENTENCES WERE THE MOST PREDICTABLE. BY FOCUSING ON EACH OF THE 10 FEDERAL CIRCUITS AND 6 SELECTED FEDERAL DISTRICT COURTS AND BY USING NATIONALLY DERIVED OFFENSE-SPECIFIC MODELS, SENTENCES IMPOSED FOR NARCOTICS COULD BE MORE CONSISTENTLY PREDICTED THAN THOSE FOR ANY OTHER OFFENSE. ALSO, SENTENCING PATTERNS WERE FOUND TO CHANGE VERY LITTLE OVER TIME. WHEN AN EQUATION THAT SUMMARIZED FEDERAL SENTENCING PRACTICES IN 1964 WAS USED, SENTENCES IMPOSED IN 1971 COULD BE ACCURATELY PREDICTED. IT IS CONCLUDED THAT EQUITABLE SENTENCING BY WAY OF A CONCRETE SENTENCING POLICY--THAT IS, A POLICY THAT ASSIGNS SPECIFIC WEIGHTS TO VARIOUS OFFENDER, OFFENSE, AND PROCESS-RELATED FACTORS--IS TECHNOLOGICALLY FEASIBLE. FOLLOWING AN INTRODUCTION TO THE METHODOLOGY AND THE USE OF REGRESSION ANALYSIS TO DESCRIBE AND PREDICT OUTCOMES, THE USE OF A NATIONAL MODEL TO PREDICT SENTENCING AT THE CIRCUIT COURT AND DISTRICT COURT LEVEL AND THE CONSISTENCY OF SENTENCING OVER TIME ARE DISCUSSED. IMPLICATIONS FOR PREDICTION ARE CONSIDERED. INDEPENDENT VARIABLES ARE DEFINED IN AN APPENDIX. A USE EVALUATION QUESTIONNAIRE AND TABULAR AND GRAPHIC DATA ARE PROVIDED. REFERENCES ARE FOOTNOTED. SEE ALSO NCJ 33683-33685. (AUTHOR ABSTRACT MODIFIED--KBL)