Results of Sign Test

#1
Hi all,

I was hoping somebody could help me with understanding the results of a sign test.
I am not a statistician but a medic trying to understand statistics for medical research purposes.
After applying a sign test to 2 sets of data my results resulted in a negative score :
z = (p - π)/SE(p) = (0.4 – 0.5) = - 0.1/0.158 = - 0.63.

This shows that we have insufficient evidence to prove the null hypothesis, but I was wondering what is the significance of a negative score? What if the result was positive?

I appreciate any advice, many thanks!
 

hlsmith

Omega Contributor
#2
are you using the wilcoxon sign rank test or which one? what happens when you flip the order of the groups? have you graphed data with say a boxplot or histograms?
 
#3
Hi hlsmith,

thanks for your reply i really appreciate your time. I am using the basic sign test, not the wilcoxon. The data was 2 sets of 12 people rating how satisfied they were and comparing the 2 groups A and B. We used a Visual Analogue Scale scoring from 1 - 10. Results :

A B A-B Sign
7 5 +2 +
6 6 0
4 8 -4 -
8 7 +1 +
4 10 -6 -
1 5 -4 -
7 7 0
2 6 -4 -
5 8 -3 -
8 7 +1 +
8 9 -1 -
5 4 +1 +


So we used a simple sign test, which i understand but wondering why the negative result?
Thanks again for any help