Hi,
I was wondering if you could let me know if orthogonalization is appropriate in the following context:
We ran an experiment in which subjects received a stimulus every 30 seconds (duration = 30s), alternating with periods of rest (duration = 30s). As the subjects were receiving the stimulus, they were also making ratings (every TR, they made one rating). Upon looking at the data, we realized that the rating mirrored the stimulus (i.e. when the stimulus was ‘on’, the ratings were high, and when the stimulus was ‘off’, the ratings were low). We put EV1 as the stimulus (30s on, 30s off), and we put EV2 as the ratings (30s with high ratings and 30s with low ratings). When we put both EVs into the model, however, the signal went away; most likely because both EVs occurred simultaneously and varied together. So, at the subject level, we orthogonalized EV2 with respect to EV1. Then, in the group analysis, we used the EV2 contrast to look for a group effect, and found very interesting results.
My question is, how to interpret this? Is EV2 (ratings) a covariate, and this analysis is showing where EV2 covaries with the signal? Is there a better way I should be doing this?
Thank you for your time, and happy holidays!
Jodi
|