Hello
I have found that there are a variety of hypothesis tests for testing the null hypothesis that two intra-class correlation coefficients are equal. However, each carries its own assumptions. Does anybody know of any such test which is suitable when all of the following conditions occur together:
a) the sample data are not Normally distributed
b) the two ICCs refer to the same sample
and
c) the sample data are not binary but rather are either ordinal or continuous?
I look forward to hearing of an appropriate test if it exists.
Best wishes
Margaret
---------------------------------
Inbox full of spam? Get leading spam protection and 1GB storage with All New Yahoo! Mail.
|