Dear SPM users, I have a question relative a t-comparison. I have a design with an unequal number of trials between conditions. In my case, there are 30 trials in conditions A and 70 trials in condition B. I would like to compare A>B. Is it correct to scale the contrast vector according to the number of trial in this way: [1 -3/7]?
Intuitively, it permits to compensate the unequal number of signal/event but does it make sense relative to the t-comparison and its interpretation?