NCJ Number
61251
Journal
Social Forces Volume: 58 Issue: 1 Dated: (SEPTEMBER 1979) Pages: 212-227
Date Published
1979
Length
16 pages
Annotation
THE THEORETICAL UTILITY AND EMPIRICAL VALIDITY OF RATIO CORRELATION AS A FORM OF ANALYSIS IS EXAMINED BECAUSE OF RECENT CRITICISMS OF ITS USE IN SOCIAL SCIENCE RESEARCH.
Abstract
SEVERAL WRITERS HAVE SUGGESTED THAT SERIOUS METHODOLOGICAL PROBLEMS RESULT FROM CORRELATING RATIO VARIABLES CONTAINING COMMON TERMS IN THEIR NUMERATORS OR DENOMINATORS, OR WHEN A RATIO VARIABLE IS CORRELATED WITH ONE OF ITS COMPONENT TERMS. AS A RESULT, THESE WRITERS HAVE CHARGED, MANY STATISTICAL CORRELATIONS REPORTED MAY BE MORE SPURIOUS THAN REAL. HOWEVER, THESE CRITICISMS ARE NOT NECESSARILY VALID. IF THE HYPOTHESIS BEING TESTED IS FORMULATED CONCEPTUALLY IN TERMS OF RATIO (E.G. PER CAPITA) MEASURES, THEN SPURIOUSNESS IS NOT AN ISSUE. SPURIOUSNESS IS A PROBLEM MAINLY WHEN COMPONENT VARIABLES ARE STANDARDIZED BY A COMMON TERM FOR REASONS EXTRANEOUS TO THE HYPOTHESIS. EVEN IN THIS CASE, THE INFERENCES ABOUT RELATIONSHIPS RATHER THAN THE EMPIRICAL CORRELATION ARE SPURIOUS. IN SOME RESEARCH CONTEXTS, USE OF TRADITIONAL RATIO VARIABLES DOES HAVE CONCEPTUAL AND INTERPRETIVE ADVANTAGES, AS WELL AS SPECIFIC STATISTICAL ADVANTAGES, OVER THE ALTERNATIVE PROCEDURES PROPOSED BY CRITICS. COMPARISONS OF RATIO CORRELATION AND OTHER METHODS IN ACTUAL STUDIES ILLUSTRATE THE METHODOLOGICAL ISSUES INVOLVED. TABLES, FOOTNOTES, AND A BIBLIOGRAPHY ARE INCLUDED. (CFW)