02-16-2017 02:17 PM
Hi, I have been reading up on the MULTTEST procedures available in SAS. I will be conducting multiple comparisons of 2 samples on categorical data (i.e., sample 1 vs sample 2 on yes or no responses for question #1, sample 1 vs sample 2 on yes or no responses for question #2, etc). I would like to conduct chi squares for the comparisons, but I would like to correct for multiple tests. Most of what I've read in relation to chi squares and the MULTTEST procedures is to input the raw p-values and then adjust using FDR or something else. However, I have not come across any examples of using the bootstrap procedure. Is it possible to do this at all? If I were to run the chi squares within the MULTTEST framework, would it work? I have seen a Fishers test example for permutation (so not bootstrap, but still a MULTTEST procedure), but no chi square. Any help is appreciated!
02-16-2017 04:25 PM
I don't know the answer, but I will point out all examples in Chapter 12 of Multiple Comparison and Multiple Tests Using the SAS System (which discusses using PROC MULTTEST for binary data) uses the PERMUTATION option on PROC MULTTEST for resampling. There is also a documentation example for binary data and two groups.