NCJ Number
147611
Journal
Sociological Methods and Research Volume: 4 Issue: 3 Dated: (February 1967) Pages: 349-379
Date Published
1967
Length
31 pages
Annotation
This paper suggests numerous research situations in which two-group and multiple discriminant analyses might be productively applied; principal methodological problems attributed to discriminant analysis are summarized and possible solutions suggested.
Abstract
Frequently, sociologists encounter research problems that require assessment of the extent to which two or more categories of people or events can be maximally distinguished from one another regarding some common variable attributes. Predictive models that accomplish this, in addition to providing rules for the classification of new entities, have been absent from the literature. Discriminant analysis, a multivariate technique used extensively in physical anthropology, clinical psychology, and the biological sciences, has made little headway in sociology, even though many substantive investigations are amenable to it. To illustrate, the use of discriminant analysis for data on student politics is presented. The authors conclude that as a means of maximally distinguishing a priori groups on the basis of variable profile scores, multivariate techniques such as discriminant analysis are superior to methods based on consecutive univariate tests. Multiple- discriminant analysis, particularly, would be useful in studying configurational relationships among K groups in a multidimensional space. 38 references