I have run an within-subjects experiment in which each participants answers to 3 binary questions (A,B and C). This means that all participants answered every questions.
I want to compare the proportions of correct answers between each pair of questions: is the p(A) significantly different from p(B)? same for the other two pairs.
My first idea was to conduct 3 tests for each pair, i.e. tests which H0 are of the forms "p(X) = p(Y)" for X,Y = A,B or C. And that's what I believe does a Chi2 test when there's only two questions.
What else does provide a Chi-squared test conducted on the whole table? What is tests are best to do in this case?
