Paired ordinal (rank) data - compare across groups

Question

I have pre- and post- treatment survey responses measured on an ordinal scale (1-5). There are two treatment groups (control and intervention). I understand if I want to test whether there is a difference in the before and after response I will need to use the Wilcoxon-signed rank test - i.e. the non-parametric alternative to the paired t-test (performed for each group).

But what if I also want to test if the before and after difference in response is different between groups?

My thought here was then to compare the pre-post difference between groups using the Mann-Whitney U test (non-parametric alternative to the independent samples t-test).

Is this correct?

Thanks

Answer 1

An ideal approach is to use ordinal regression. This will allow you to have both a before/after effect and a group-a/group-b effect in the same model. Ordinal regression is relatively easy in R, but I can't comment on other packages.

Your idea about using the pre-post difference as the dependent variable is also a viable option, but note that to do that you need to make an assumption that the ordinal categories are equally spaced. That is, you have to assume that a 5 is as far from a 4 as a 4 is from a 3, and so on. Otherwise you wouldn't be able to subtract the numbers.