When we want to analyse variables with 2 variable, one variable is scale nominal, other's variable ordinal, pre and post test (two tail), can we choose wilcoxon signed rank test?
89
Data cannot be parametric or nonparametric. Some statistical procedures are parametric and some are nonparametric, and some are both!
You should base your statistical analysis on the research question,not the data type. The data type will guide you to your final choice of statistical procedure, but cannot be used to select a procedure.
What is your question, and what have you measured?
If you can pair each pre-test to a post-test ---- that is, if you know student A got this score on the pre-test and you know the score the same student got on the post-test---, then you want the Two-sample Rank-sum Test for paired data.
The names of these tests get confusing, but you want the Wilcoxon test for _paired_ data, if you can pair the data.
There is an example here:
http://rcompanion.org/rcompanion/d_10.html
Option 2:
If you have the capability to do ordinal regression, I would recommend that you use that instead. You still want to pair the data, by Student. Either use Student as a blocking variable, or as a random blocking variable.
- Badges
- Science topic