Hello,
I have a nominal variable in a content analysis called that is divided in 21 distinct values. Three independent coders coded this variable.
I used Deen Freelon’s ReCal online utility (http://dfreelon.org/utils/recalfront/recal3/) to calculate Cohen’s k and Krippendorff’s a for this variable: .712 and .715, respectively. Now I need reliability numbers for individual values, not just the overall variable.
First, is this even possible? Isn’t this like asking about the reliability of male vs. female vs. other? Second, if it’s possible, how do you go about it on SPSS or any other way that isn’t by hand, which I’m just not capable of doing?
Thanks.