I have data set of teacher's evaluation rated by 4 different raters. The teacher's are evaluated for 13 different categories for example (Interaction with students, lesson delivery etc). All of the 4 raters are rating teachers in 13 categories from 1 to 5.
I want to find the agreement level between observers using cohen's kappa. I know how to compute kappa for one category only but I am confused how can we do that for different categories? Do I have to compute it for each category separately? or is there any other method?