Wilcoxan test

Question

Hello

Full disclosure- VERY new to JMP and a super novice at stats.

I am trying to run a Wilcoxan test and I'm not sure of two things:

1. Does it matter what variable is listed first on the title? I have been trying to enter the Pre data before the Post data as detailed on a blog but every time I run the test the title of the dat comes up "Difference: PortPost-PortPre.

2. At the bottom of the page are the p values but I don't know which one I am supposed to use?

Thank you!

data.jpg

txnelson · Answer

Right click on the header for your PortPost-PortPre column and select

Column Properties=>Value Order

You will be able to set the order of the displaying of the data in the dialog that pops up

NJB187 · Answer

Thank you for replying. The pre is before an intervention and I'm expecting the post numbers to be lower (numbers are complications). I cannot get the Pre data to be listed first. Can you help me with that?

Mark_Bailey · Answer

Generally, the 'sense' of the difference matters. For example, if Before = 3 and After = 10, the difference is 7, not -7. The response increased. So, yes, the order of the variables is important - to you.

JMP does not know what your alternative hypothesis is. The null hypothesis is that the difference = 0. Are you testing for any difference, negative or positive change? Then use Prob > |S|. If you are testing for a negative change, then use Prob < S. If you are testing for a positive change, then use Prob > S.

You should have decided the appropriate alternative hypothesis before your analysis. The test is a 'one shot.' The data is good for one test, not for repeatedly testing with different hypotheses until you get a significant decision. You should 'fish' for the hypothesis that you like.