ABSTRACT
In this chapter, the author attempts to put measures of association for ordinal data into the framework of the general linear model, to accent their utility for data analysis. The use of the general linear model also helps to explain why these statistics can be zero even when there is a stochastic dependence between the variables. The author focuses on Kendall's tau, Goodman and Kruskal's gamma, Somers's d, and derived statistics to show how ordinal measures of association can be used in the framework of the general linear model. A considerable start in the direction has been made by Maurice G. Kendall, Robert H. Somers, H. T. Reynolds, and Roland K. Hawkes. While some of the results are tentative, enough is known to merit increased use of the general linear model with ordinal measures of association. While tau summarizes a linear relationship between pairs, gamma summarizes a monotonie one.
