Hi,
I have a fast even related design and was hoping to get a little bit of input on how to setup my regression model.
The task involves seeing subtle and overt emotional expressions (as well as neutral expressions). Subject must indicate if the expressions is emotional or neutral. Accuracy for the overt expressions and for the neutral expressions is very high (approx. 95%) and accuracy for the subtle expressions is around 50%. In my current model, I have the following regressors:
correct-subtle
incorrect-subtle
correct-overt
correct-neutral
Should I also include incorrect-overt and incorrect-neutral regressors if there are only a small percentage of these stimuli there were incorrect? It can't hurt the statistics of interest, correct? My ultimate goal is to look at glts for correct/incorrect-subtle vs. correct-neutral and correct-overt vs. correct-neutral.