At least to me, it's not clear what kind of test might be appropriate. How much data do you have? For example, do you just have the two percentages you reported? Do you have a score for each respondent? Did the same respondents take each of the tests?
Thanks for your response. There were 21 respondents who all took the 2 tests, and yes, I can access the scores for each respondent.
It sounds like you would want to use a paired t-test or a pairaed wilcoxon signed-rank test. ... But I'm wondering about the fact that the two tests have a different number of items. Does it make sense to compare the results of the two tests ?
"Comparing" is a vague description of what you want to do. Is your hypothesis that people score higher on one test than another? Do you want to see what score on one test corresponds to a given score on the other test? Do you want to compare mean scores? (I'd be wary of the latter, given that both tests – certainly the first one – may have test ceiling issues. Or do you want to see if the tests agree – that the rank order of each person on one test agrees with their rank order on the other?
Tell us, and we can make more specific recommendations (or at least argue about them, which is most of the fun of these question fora!).
Thank you both for your responses. Yes, my hesitance in using a paired t-test was that the tests have different numbers of items. What I want to know is whether the students scored higher on one test than the other. It's not a directional hypothesis, though. Thanks again for your expertise.
- Badges
- Science method