using Mann-Whitney / Wilcoxon tests for ordinal data

[adaren]

Is there a way to use Mann-Whitney (independent groups) or Wilcoxon (paired samples) for ordinal dependent variables?
This is possible for Kruskal-Wallis, as the dialog box accepts ordinal data. Not possible for Friedman test, though.
As I understand the only way is to redefine an ordinal variable as continuous...?

[reclusivestupid]

Yes, you are correct. The Mann-Whitney U test (for independent samples) and the Wilcoxon signed-rank test (for paired samplesbasketball stars) are nonparametric tests commonly used when the dependent variable is ordinal or when the assumptions of parametric tests are not met.

[adaren]

Right. I was asking why ordinal data is not allowed - that is, the windows for selecting dependent variables for these tests (found under t-tests) does not accept ordinal data. I know this all is under t-tests menu, but it could be be e.g. independent samples t-test & Mann-Whitney).
Similarly no ordinal data allowed for Friedman test.

[decaux]

I would say that it should be allowed.
In the meantime you can change the measure type of your data to continuous (even though it is ordinal) and then the dialog box will accept it.