Abstract
Association models have been presented for the analysis of singly-ordered contingency tables. The models contain some parameters regarded as category scores, and offer some additional insight into the meaning of these scores. We propose an association model with location and dispersion scores for the analysis of association in crossclassifications having singly-ordered response categories. Some issues of interest in this article are (i) two-dimensional graphical displays of parameter estimates and assigned scores, (ii) tests for homogeneity of row effects in the association model, and (iii) Monte Carlo simulations conducted to investigate the accuracy of the chi-square approximation to the null distribution and the power based on the proposed method. We show that the proposed method improves the power in testing for homogeneity of conditional distributions within rows under certain alternatives, comparing with other methods.
