Thanks a lot Alfonso, that certainly makes sense. As I am using this test only as a preliminary screening step (and I do want it to be liberal), I was wondering if there might be a more liberal equivalent within a single contrast (other than changing the p threshold of course)? Also, how is the p value corrected for the N hypotheses? Is it simply that if I have 3 lines the p is divided by 3? 

Thanks,

L.