Hello Dr. Zalensky,

We are trying to perform ANOVA with our test sample set of 52 individuals (4 independent groups). I have a question about design matrix, contrast and FDR method. We followed the suggestion provided by you in this forum with reference to design matrix and contrast settings. We tested the following design and contrast settings.

Design 1:
1 1 0 0
1 1 0 0
1 -1 1 0
1 -1 1 0
1 0 -1 1
1 0 -1 1
1 0 0 -1
1 0 0 -1           Contrast set to [0 1 1 1]


Design 2: 
1 0 0 0
1 0 0 0
0 1 0 0
0 1 0 0
0 0 1 0
0 0 1 0
0 0 0 1
0 0 0 1              Contrast set to [1 1 1 1]

I ran both these settings with F-test and FDR set to permutations 50000. Both of them resulted in no significant results. However, when I tested them with permutation of 5000, design 1 gave me more significant edges than design 2. Considering both the matrices and their corresponding contrasts are equivalent, why would one see a discrepancy? Is it due to insufficient number of permutations? Is there a resource explaining why FDR requires more permutations than NBS?

Your thoughts would be really helpful!

Thank you
Charanya