> When collapsing these two together using method 2, we
> are concerned that when assigning 1s to these two
> conditions, the 76% and 24% receive equal weighting
> when clearly more trials are contributing to the correct
> estimate.
Such an equal weighting does not invalid any percent signal change estimation or contrast testing. Instead the effect is mainly on the precision of coefficient and contrast estimation. Putting in words with more statistical flavors, with a ratio of correct : incorrect answers = 76%:24%, the signal averaging process in 3dDeconvolve would render the estimation for correct answers with less standard error than that for incorrect answers by a factor of sqrt(76/24) ~ 1.78. In other words, the estimation for correct answers is 1.78 times more precise.
Gang