In my data, the response variable is ordinal and there are 16 predicting variables which are also ordinal. I plan to perform an ordinal regression. But before doing that I want to graphically show the relationship between the response and predictor variables. In regular regression where the variables are numerical, we can produce a scatterplot matrix for that purpose. I am wondering whether for ordinal regression with ordinal predictors there is any graph that can be used to visually display the relationship before the formal analysis. Another possibility is to calculate the association between the response and the ordinal predictors before the analysis. Any recommendation on that part? Also, I plan to perform a proportional odds model using R. Can anyone recommend a R package or functions which can be used to test the proportionality and also used for the analysis?
Asked
Active
Viewed 288 times
1
-
2"*I am wondering whether for ordinal regression with ordinal predictors there is any graph that can be used to visually display the relationship before the formal analysis.*" -- see [this question](http://stats.stackexchange.com/questions/56322/graph-for-relationship-between-two-ordinal-variables/) for some suggestions, but of course any pairwise display has the same problem plotting $y$ vs $x_i$ does in an ordinary regression situation (the conditional relationship may be quite different from the unconditional one in such a display). – Glen_b Oct 22 '13 at 16:12
-
2As for measures of association, there are a number of ordinal-ordinal measures; one example is [tau-b](http://en.wikipedia.org/wiki/Kendall_tau_rank_correlation_coefficient#Tau-b) – Glen_b Oct 22 '13 at 16:14