I need to compare data from paired groups, but my distributions are non-normal, and my sample size is small. Are there any tests I could apply?

I've been looking into non-parametric alternatives like WMW, but I don't know if it applies to paired groups. On a time crunch, and would appreciate any advice.

  • 3
    $\begingroup$ Wilcoxon signed-rank test $\endgroup$
    – user158565
    Nov 11, 2018 at 19:13
  • 3
    $\begingroup$ For a paired t-test, the assumption is that the differences are normally distributed. $\endgroup$
    – Peter Flom
    Nov 11, 2018 at 19:13
  • $\begingroup$ An alternative to signed-rank test would be to go for the one-sample permutation t-test. $\endgroup$
    – Michael M
    Nov 11, 2018 at 19:30
  • $\begingroup$ If the distribution of the difference is not symmetric, use sign test. $\endgroup$
    – user158565
    Nov 11, 2018 at 20:57


Browse other questions tagged or ask your own question.