Evaluating Classification Schemes Used in Reporting Criminal Justice Data

• NCJ Number: 85071
• Journal: Evaluation Review Volume: 6 Issue: 4 Dated: (August 1982) Pages: 505-520
• Author(s): L J Hubert; R G Golledge; T Kenney; C M Costanzo
• Date Published: 1982
• Length: 16 pages
• Grant Number(s): 80-IJ-CX-0061
• Agencies: NIJ-Sponsored

Annotation

The two commonly used classification strategies assume that some variation exists among the aggregate classes defining the classification. A heuristic optimization technique may be a way of constructing better categorizations from secondary data sources.

Abstract

One common strategy groups the basic reporting units under study spatially and the statistical data reported for each unit within well-defined spatial classes. Another familiar strategy groups the units according to the values of some demographic variable. This report defines a strategy for evaluating the utility of a given classification based on spatial contiguity or population, for example. It suggests the heuristic optimization technique and presents a method for comparing the adequacy of any two different classifications. The article emphasizes the importance of empirical typologies obtained from earlier data sources as appropriate a priori classification schemes. The example used throughout the paper involves crime rate data for 40 cities in the Los Angeles area. Data tables, notes, and nine references are supplied.