Dear Natalie,
if both patients and controls differentiate in terms of response time, I would suggest modelling the group analysis as 2-factor design with factors "group" (levels "patients" and "controls") and "response time" (levels "fast", "immediate" and "slow"). For within-group analyses, you may want to run separate analyses that only model the factor "response time".
Since each subject belongs to exactly one cell in this factorial design (either "patient" or "control", either "fast", "immediate" and "slow" responder) there should be no need to take this into account during first level analysis - except for stimulus modelling of course.
Hope this helps
Volkmar
|