Dear all,
Can anyone advise me on how to deal with the following data ? Any SAS
recommendations would be particularly of interest to me too.
Basically I have 4 raters (always the same) who evaluated 15 subjects with
an ordinal variable of three categories (very good, good, not good). Each
subjects was assessed by each rater twice but the raters did not know they
evaluated the same subject. The second rating was done after the raters
were given particular recommendations for rating.
As I understand the data, it looks like we could do a kind of extension of
the Mc Nemar test but with repeated measures.
Do you think a GEE approach with the rater and the subject as clustering
factors is appropriate to measure the impact of the recommendations the
raters were given ?
|