Some Comments To Console Edgar F. Borgatta

Abstract

IN "My Student, the Purist: A Lament," Edgar F. Borgatta (1968) examines the issue of using parametric statistical measures with ordinal data, specifically using Pearson's r with ordinal-ordinal data. The con-

The correlation between variate values and ranks in small size samples

“…A question then arises concerning the degree of distortion or error created by using a rank-order measure to represent an underlying continuous variable. One answer to this question that has been influential (see, for example, Aitkin & Hume, 1966; Jacobson, 1970;Kendall, 1973;Kim, 1975;O'Brien, 1979) is found in an article published in this journal over 25 years ago (Stuart, 1954). Although Stuart emphasized the application of his findings to the study of the asymptotic relative efficiency of various statistics, his results have also been employed to defend the use of rank-order measures of continuous underlying variables.…”

“…In spite of these cautionary comments, formula (1) has been repeatedly used to justify the 'replacement' of original, or underlying, variate values by their ranks (e.g. Aitkin & Hume, 1966;Jacobson, 1970;Kendall, 1973;Kim, 1975). Kendall (1973, p. 125) writes that '.…”

Conjoint Measurement and the Canonical Analysis of Contingency Tables

“…While naturally eager to extend the empirical range of the more powerful linear techniques, researchers have also shown their conservatism and caution when it comes to innovation in the area of statistical analysis-generally the eager pragmatist has to some extent been counterbalanced by the caution of the measurement purist (see the debate between Borgatta, 1968, andJacobsen, 1970). And this caution is not all scientific ritualism, as some have suggested.…”