These statistics measures association between ordinal variables: gamma, Kendall’s tau-, Stuart’s tau-, and Somers’. If I am to report correlation between ordinal and nominal variables, what is appropriate statistic to report while using PROC FREQ of SAS to examine an association?
A nominal variable has no order, by definition, so I don't see why you would want to test a correlation between it and something. (see e.g. http://stats.stackexchange.com/questions/119835/correlation-between-a-nominal-iv-and-a-continuous-dv-variable). Are you interested in testing the differences between different levels of the variable?
Dear Moshood,
I agree with Stephen. This kind of design lends itself to a test of association with a two way chi-squared. If you have a validity difficulty due to low frequencies you may be able to combine some of your ordinal categories together - see: https://www.researchgate.net/publication/301230965_Chi-Squared_Test_for_Two-Way_Tables_Revised.
Data Chi-Squared Test for Two-Way Tables (Revised)
- Badges
- Science topic