- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content
Posted 10-31-2017 01:42 PM
(1406 views)
Hi,
I used proc univariate to do signed rank test on a variable delta_1, which is the difference between two groups. I rescaled delta_1 to delta_2 where delta_2=delta_1*a constant and then did signed rank test on delta_2. I was expecting to obtain the same results. However both the signed rank and p value changed. I don't understand what drives the difference.
Thanks!
2 REPLIES 2
- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content
The test statistics will be the same only when Mu0=0. For other values of the location parameter, there is no reason to expect that the signed-rank test is invariant to scaling.
If that doesn't solve the problem, post your PROC UNIVARIATE code.
- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content
Yes show the code, both for how you scaled the variable and the univariate code for both tests.