DETERMINANCY VERSUS RISK: A CRITIQUE OF CONTEMPORARY STATISTICAL METHODOLOGY IN SOCIOLOGY

Abstract

The contention of this paper is that contemporary sociology has been misled into relying on tests of significance to the virtual exclusion of measures of association. A glance at any recent issue of the American Sociological Review will show that the bulk of the research is analyzed by the computation and interpretation of levels of significance. Further, even if a measure of association is computed it is usually analyzed in terms of a test of significance rather than in terms of its own numerical value. This reliance means that our discipline annually discovers and reaffirms the probable existence of many interesting relationships; however, it also means that the strengths of these relationships are not known.