Posthoc testing for multiple categorical variables

[#1]

Does anyone know how to do do posthoc testing of multiple categorical data in JMP. I am analysing tumour location (4 different locations) versus e.g. if the tumour occurred bilateral or not. I do the Fit Y by X and JMP produces the mosaic plot and then I can do Fisher's Exact Test, which tells me yes there is a significant difference between the locations. But how do I now find out where the significant difference is?

I would be very grateful for your help.

[#2 · Accepted solution]

Hi,
I'm not sure I understand what you are trying to achieve based on the limited information you provided. The Contingency analysis tests for similar proportions across categories and you can reject the null hypothesis if these proportions are sufficiently different across categories. Hence, the test is only meaningful when you compare all the categories to one another. Based on your post, it seems that you would like to know which location is most different from the other which can be achieved by using the sub-analysis "Analysis of Means for Proportions", is that what you are looking for?
Best,
TS

[#3]

In addition to the solution offered by @Thierry_S , you can use the red triangle menu next to the table to show the cell chi square. This distance highlights the counts that differ the most from the centroid. The square root of the cell chi square is the Pearson residual. It is not a test as @Thierry_S showed but is related information about your data.