NCJ Number
82128
Journal
Law and Society Volume: 16 Issue: 1 Dated: (1981-1982) Pages: 147-162
Date Published
1982
Length
16 pages
Annotation
A new, easily applied method of scaling the severity of different types of criminal sentences and the seriousness of different types of crimes is described and illustrated.
Abstract
The application of canonical correlation scales sentences according to how the judicial system assigns them to crimes of varying seriousness and scales crimes according to how the judicial system assigns to them sentences of varying severity. Since the intended purpose is to scale both sentences and crimes according to actual judicial performance, it is not appropriate to rely on subjectively derived values for either scale. The statistical technique of canonical correlation allows researchers to estimate simultaneously scale values to the seriousness of different crimes and the severity of different sentences. Canonical correlation can be easily understood by anyone familiar with the multivariate methods of factor analysis and regression analysis. Canonical correlation extracts a linear combination, called a 'canonical variate,' from each of two sets of variables. Whereas factor analysis extracts factors that maximally explain the common variance among the variables, canonical correlation extracts pairs of canonical variates such that the correlation between the variates from each set of factors is maximized. This approach is an appropriate empirical method for research on a court system when the researchers desire severity and seriousness scales that are implicit in the system's performance, rather than scales based on consensual validity among experts or based on some a priori standards. Tabular data, 27 footnotes, and 22 references are provided.