NCJ Number
75003
Date Published
1979
Length
21 pages
Annotation
The effects of 10 weighting schemes on the reliability and validity of a self-reporting delinquency measure were investigated.
Abstract
The weighting schemes consisted of unit weights (whether or not the event occurred once, twice, more than twice, or not at all); interval weights (whether the behavior had occurred within the last 6 weeks, the last year or whether it had ever occurred); and seriousness weights. The self-reporting measure was administered orally to 123 subjects 4 times during the course of an 18-week delinquency prevention project. For comparative purposes, archival data were collected from the probate court and from law enforcement agencies. Adding seriousness weights did not contribute appreciably to the ability of the instrument to detect outcome data. Frequency weighting schemes added somewhat to the instrument's reliability, but not at a significant level. The self-reporting instrument demonstrated consistent internal and test-retest reliability. Convergent validity was demonstrated through multitrait-multimethod procedures. The scale also demonstrated a positive relationship to court and law enforcement agency data. Although the seriousness weighting scheme added no significant validity or reliability to the instrument, it also failed to harm or distort the results of the self-reporting scale. The unit weights performed as well as the other, more popular, weighting schemes and were less complex to utilize. Twenty-two references and tabular data are included.