0

In R, I am using the survey package to conduct these tests on paired data:

  • Weighted Wilcoxon Signed Rank test
  • Weighted Sign test

However, I am struggling in doing them.

By defining a design and then using the functions

signed_rank<-function (x) {sign(x) * rank(abs(x))}
summary (svyglm(signed_rank(variable) ~ 1, mydesign))

I have been able to conduct a non-paired wilcoxon.

But I have not been able to conduct the paired version as well as the Sign test. Any idea?

Thanks

UseR10085
  • 7,120
  • 3
  • 24
  • 54
Matteo Podrecca
  • 1
  • This is a cross-post of a question answered on CrossValidated, where it is more appropriate: https://stats.stackexchange.com/questions/613027/paired-sign-and-wilcoxon-tests-with-weighting/613164 – Thomas Lumley Apr 17 '23 at 07:24

0 Answers0