I have a question about how to analyze some data. It involves the performance of two equally sized groups in a questionary with 30 arithmetic questions that I scored with two possible outcomes (correct/ incorrect) plus possible omissions (unsolved question). My first approach to these data has been to calculate each subjects' frequency of correct, incorrect and non-solved questions which I aimed to compare by means of a chi-square association test (plus an a posteriori inspection of the standardized residuals).
On a second step, as I do not expect differences between groups, I would like to observe whether participants (regardless of the group) provided more/ less correct answers depending on the type of arithmetic operation involved (addition, subtraction, multiplication or division). I intended to ascertain this question by means of chi-square goodness of fit and, again, an a posteriori inspection of the standardized residuals
However, I have been told that my approach might be incorrect because I should treat correct, incorrect and non-solved questions as interval variables. Therefore, I am quite puzzled at the moment and I would appreciate expert advice on the correct approach to deal with this two experimental situations.
(I am using SPSS, just in case this information might be relevant here)
