U.S. flag

An official website of the United States government, Department of Justice.

NCJRS Virtual Library

The Virtual Library houses over 235,000 criminal justice resources, including all known OJP works.
Click here to search the NCJRS Virtual Library

Prediction of Delinquent and ADM Behavior From Other Delinquent and ADM Behavior (From Multiple Problem Youth: Delinquency, Substance Use, and Mental Health Problems, P 169-190, 1989, Delbert S Elliott, David Huizinga, et al -- See NCJ-119536)

NCJ Number
119542
Author(s)
D S Elliott; D Huizinga; S Menard
Date Published
1989
Length
22 pages
Annotation
This chapter examines the prediction of the various types of delinquent and ADM (alcohol, drug, and mental health) behaviors directly from one another, using data from the National Youth Survey for 1976-1983.
Abstract
The study first considers the prediction of "ever-prevalence" of delinquent and ADM behaviors from "ever-prevalence" of other delinquent and ADM behaviors. "Ever-prevalence" refers to the percentage or proportion of individuals who have committed an offense, etc., or on the individual level, to whether an individual has committed an offense, etc., in any one or more of the years under study. This is followed by an examination of substance use immediately prior to Index Offenses. The study then addresses the prediction of classification as serious delinquent, polydrug user, and mental health problem types as well as transitions between serious delinquent and polydrug user types. To examine whether the joint influence of delinquency, drug use, and emotional problems on future delinquency and drug use was additive or interactive, the study used the annual delinquency, drug use, and mental health problem types for 1976. The cross-classification of these three typologies, together with measures of future delinquency or drug use were used in a hierarchical loglinear analysis. As expected, delinquency and drug use were positively correlated. Except for the correlation between two types of drug use (marijuana and polydrug use), however, the correlations were not strong and reflected only 1 percent to 12 percent shared variance. Examination of conditional probabilities, however, indicated that despite low correlations, some strong (90 percent or more accuracy) predictive statements can be made about one type of behavior based on knowledge of another. 6 tables.