I am not sure whether there is a name for my following procedure: Given that there is an ordinal variable with 4 levels (0,1,2,3).

Ususal contrast is:

  1. 0 vs 1
  2. 0 vs 2
  3. 0 vs 3

I re-group (contrast) them into binary:

  1. 0 vs >= 1
  2. <= 1 vs >=2
  3. <=2 vs 3

May I know whether the re-grouping is legit if I am going to do regression? I will replace the whole ordinal variable with binary coded variable (one per time, so totally three regression model will be built instead of one). If yes, are there any name for this kind of "treatment/ transformation"? It would be great if you can also talk about the name of first grouping method (the usual one), so that I can call them and distinguish them.



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.