I am sorry if this is a naive question, but I wanted to have your suggestions before considering anything. Suppose I have a group of "n" students and I have their score in three courses: A, B, and C. If the scores are not on the same scale, I can standardize them before comparison. For comparing the mean difference between two courses I could have done the Paired-T test, but I am not sure about the extension of paired T-test for comparing more than two means for dependent data. Any ideas? Thanks in advance!

  • 1
    $\begingroup$ Try the two-way ANOVA. Testing three paired-T test is not appropriate here. $\endgroup$
    – xjf
    Feb 2, 2017 at 22:46
  • $\begingroup$ If you standardize the grades within course, all courses will have a mean of zero. Obviously, then, there will no difference between courses in mean score. $\endgroup$
    – Joel W.
    Feb 3, 2017 at 0:28
  • $\begingroup$ @Jianfeng Could you please explain what will be the factors for two-way ANOVA? $\endgroup$ Feb 3, 2017 at 15:29
  • $\begingroup$ @JoelW. Thank you for pointing that out. By standardize I meant I will convert all the scores under the same scale (eg. score out of 100) $\endgroup$ Feb 3, 2017 at 15:31
  • $\begingroup$ @curiousmind, one factor is course and the other factor is student, since the scores from the same student are correlated. $\endgroup$
    – xjf
    Feb 3, 2017 at 16:29

1 Answer 1


A simple and intuitive solution is to just do three different paired t-tests and then correct for multiple comparisons using bonferroni correction. Approximately this means you just divide your alpha level with the number of comparisons you are doing.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.