Estimating Juvenile Recidivism by Cross-Level Inference (From Inferring Individual Behavior From Grouped Data, 1981, by John Wanat - See NCJ-86079)

• NCJ Number: 86080
• Author(s): J Wanat
• Date Published: 1981
• Length: 17 pages
• Corporate Author: University of Illinois at Chicago Circle
• Sale Source: National Institute of Justice/
• Language: English
• Country: United States of America
• Note: Available on microfiche as NCJ-86079. Presented at the Special National Workshop on Research Methodology and Criminal Justice Program Evaluation, March 17-19, 1980, Baltimore, Maryland.

Annotation

This paper applies a nonanalytic technique currently under development to the estimation of juvenile recidivism from aggregate data.

Abstract

Where only aggregate level data on juvenile crime for two successive years are available, no precise estimate of recidivism is possible with present techniques. Yet by using only those data, usually ignored relations among the variables, and examining the map of solutions to the 3x3 table representing the recidivism situation, a solution that is 'most possible' can be identified. The best or 'most possible' solution is estimated by charting the frequency distributions of solutions to each cell entry in the 3x3 table. That information constitutes a probability density function for each cell. By making the intuitively justifiable assumption of independence, the joint probability of each solution is computed, and the solution with the highest joint probability is nominated the most possible estimate. Sensitivity testing and a 'Monte Carlo' approach permit assessing the confidence the investigator should put in the estimate. While this approach is not backed by sophisticated intellectual theory, it is intuitively pleasing and is superior to alternatives for estimating recidivism when time and cost are considered. Tabular and graphic data as well as mathematical equations are provided, along with seven references. (Author summary modified)