Hi,

I have three conditions with equal number of trials each. When I want to compare A vs. B+C, should t-contrast be [1 -1 -1] or [1 -0.5 -0.5]? 

Some time ago I have already asked a similar question and I got a definite answer that when I compare two conditions with different number of trials no weights should be applied (SPM handles it internally). In fact, my current problem is that with  [1 -1 -1] the result just does not make sense to me, whereas [1 -0.5 -0.5] gives what seems to be right. At the end, I know what I get for A vs. B and A vs. C separately. 

Thanks for help,

John