"Correlation" between multiple categorical variables!

#1
I'm new to these forums - working on a study involving agreement between physicians' treatment choices. It's a multiple response questionnaire. I want to know the level of agreement between physicians in their choices of these multiple categorical variables. Any idea what the best test is to use? Any help would be much appreciated!
 

CB

Super Moderator
#3
It sounds like generally what you're interested in is something like interrater agreement. Specifically Cohen's kappa or Fleiss' kappa might be helpful.

Ameeq - Likert scales are a data collection method, not a statistical test.
 
#4
Thanks for your response. Exactly re: inter-rater agreement. The only problem being that with the variables all being nominal, there doesn't seem to be a more glamorous statistical method than raw agreement indiceskappa. (ie proportion who chose A, C, and E; proportion who chose B and C; proportion who chose only D, etc). But I will look into the Cohen and Fleiss kappa.