Sign Test for skewed, non-normal paired sample comparisons

[istanbulemdoc]

Hi,

For the comparison of paired samples with non-normal distributions (or ordinal data) with skewness out of the boundaries of -2 and +2 it is suggested that the sign test is used instead of the wilcoxon test.

Is there a way to perform it with jamovi?

Thanks

[MAgojam]

Hey @istanbulemdoc,

in jamovi you do not find the sign test, but its versatility allows you to do it.
I'm not going to offer you some R code to use with version 2.0.1 of the Rj module (that's great), but two simple steps, which take the time of a blink of an eye.
Compute a new variable as the difference of the two of interest to create the signs you need and apply a binomial test to this new computed variable (you can find it in "Frequencies> 2 Outcomes> Binomial test".
I am attaching a screenshot as an example to give you the idea.

pain.PNG (57.93 KiB) Viewed 1339 times

There are two VAS "Visual Analogical Scale" ordinal variables for self-assessment of lumbar spine pain, pre and post rehabilitation treatment.
The third variable PAIN and the one computed with the decreased/increase signs as the difference between before and after used for the binomial test.

Cheers,
Maurizio