2

I am wondering if the Wilcoxon signed rank test is an appropriate test for my data, that does not follow a normal distribution.

I have 150 datasheets of numbers of drugs measured at two points in time.

  • 40 negative ranks (reduction in number of drugs)
  • 10 positive ranks (increase in number of drugs)
  • 100 zero differences

Considering the high prevalence of zero differences in around 2/3 of datasheets, is it still valid to use Wilcoxon signed rank test? Is there any reasonable alternative for nonparametric and non-continuous data?

kjetil b halvorsen
  • 77,844
StefanM
  • 21
  • Likely to be related: https://stats.stackexchange.com/q/103162/3277 – ttnphns Jul 24 '23 at 15:27
  • 1
    Also: https://stats.stackexchange.com/questions/73533/why-do-zero-differences-not-enter-computation-in-the-wilcoxon-signed-ranked-test – jbowman Jul 24 '23 at 15:39

0 Answers0