I've read the documentation linked to in the previous post, but I'm still unclear on a couple of things.
1.) Is it correct to say that the first of the two response models that 3dD outputs when using -stim_times_AM2 represents the average effect of a stimulus class, with the effect of the modulating variable (e.g., reaction time) removed? Or is the first response model the average effect of the stimulus class with the effect of the modulating variable present?
It seems pretty clear from the documentation that the second response model output by 3dD when using stim_times_AM2 represents the average effect of the modulating variable. Is that right?
2.) The data I'm looking at is from a decision making study in which the subject's response partially determines the stimulus class. As a result, I end up with empty stimulus classes for some subjects. So, when I run 3dD with -stim_times_AM2, it outputs all zeros for any regressor for which there are no instances of that stimulus class for a given subject. That's exactly as expected, and I can deal with this using -allzero_OK along with -GOFORIT.
However, when a given stimulus class contains only one instance (i.e., when a given stimulus class occurred only once in the course of the experiment), 3dD outputs all zeros for the second AM2 response model but not for the first. Putting aside for a moment the question of whether one would want to use a stimulus class that contains only one instance (perhaps not), is this normal behavior?