Hi Donald,
thanks for your answer! Averaging the sessions definitely is the easiest way. However, I wonder about the comparability of the computed averages. Assuming that we have only 3 subjects. The first subject was measured once, the second twice and there are 3 sessions available for the third. All sessions are perfectly identical, no difference regarding the task. If I now average the contrast for each subject and put them into my second level analysis, I will treat all the averages equally. Isn’t this partially wrong, since my average contrast for subject 3 should be of better quality and more precise concerning e.g. my residual error than the ones for subect 1 or 2 which both include less sessions. What do you think about this issue and is there any chance of modeling this quality difference by somehow weighting the average contrasts according to the number of included measures?
Best,
Laurens
|