Dear Michael,
You are interested in differences between conditions, or better, differences between the average responses of the conditions, so you don't want to take into account number of trials. Thus contrasts like [1 -1 0 0] are the ones you want to specify.
In case you are interested in the average response *across conditions* you would go with [1/4 1/4 1/4 1/4], weighting the four conditions equally.
However, in case you were interested in the average response *across trials* for whatever reason you would take into account number of trials, which would result in e.g. [11 10 14 11]/45.
Best regards
Helmut
|