I'm looking for an extension of the Cochran-Armitage test, which applies to ordered categorical and dichotomous variables of r x 2 tables, to two ordered categorical variables of r x c tables. Do you have any suggestions?
Now I consider: the extended Cochran-Armitage (this is based on Pearson product moment), Kruskal-Goodman and Kendall's tau-c. Which one is preferable and why?
You will find it on page 100 of the book Everitt, B. S. (1992). The analysis of contingency tables. CRC Press. It is not refered to as Cochran's for Cochran-Armitage but as "log-linear model for tables with ordered categories". Does any one have it programmed in Exel?