For every subject in a study, I have 40 responses that can be: a,b,c. So it would look somthing like
sub | trial | resp
1 1 a
1 2 b
....
50 40 c
I want to check if their occurence rate differs (i.e a is much more frequent than b or c).
Is it correct to sum them up and use a linear model:
sums ~ factor + (1|subject)
