NCJ Number
91774
Date Published
1983
Length
55 pages
Annotation
This paper demonstrates that the inclusion of legally relevant variables reduces correlations between case outcomes and the variables of race and socioeconomic status. It suggests that many inferences about the extent of discrimination may be erroneous and outlines structural modeling techniques as alternative research methods.
Abstract
A review of nine recent and influential empirical studies on discrimination in the criminal justice system notes that all suggest three factors as major influences in case processing: offense seriousness, quality of the evidence, and defendant's prior record. These researchers also found other legal and extralegal variables important at different stages and claimed the cumulative nature of extralegal factors placed black and lower status defendants at a considerable disadvantage by the time they reached the sentencing stage. The paper discusses problems associated with measurement of the key determinants that may obscure the true extent of discrimination. Alternative interpretations of results reported in the literature demonstrate the import of these statistical issues. The authors argue that better measurement of these determinants is not promising and present a structural equation model of nine different decisions affecting the criminal justice process which takes explicit account of measurement difficulties. The paper discusses the estimability of the model's parameters and how it helps efforts to obtain less ambiguous data summaries. Also explored are simulteneously addressing sample selection and measurement issues in future research and trade-offs in specifying alternative structural models of the criminal justice system. Equations, tables, footnotes, and 27 references are supplied.