Hi Andrew,

Regarding to my last question, I also tried another design matrix and contrast:

intercept a b interaction
1 -1 -1 1
1 -1 -1 1
1 -1 1 -1
1 -1 1 -1
1 -1 1 -1
1 1 -1 -1
1 1 -1 -1
1 1 -1 -1
1 1 1 1
1 1 1 1

And contrast is [0 1 0 0] for main effect a, [0 0 1 0] for main effect b, and [0 0 0 1] for a*b interaction effect.

According to my understanding, this design and contrast should be equivalent to the ones in my last post (I also found that the FSL manual said these two are equivalent). However, I got totally different F-test results using NBS based on the two settings. Could you please help? Thanks.

Regards,

Jeff