I’m using the 3dDeconvolve -nodata option to test some experimental designs. Quick questions.
- Each stimulus is 12s so I originally made a binary file as input as per the manual Doug Ward wrote in 2002 where I treat each stimulus as a sequence of 6 1s in a row (2s TRs) so each stim_file essentially looked like this - 0 0 0 0 1 1 1 1 1 1 0 0 0 - and there were 3 of those. However I wanted to specify a gamma function so I used the ‘make_stim_times.py’ tool on these files and essentially got ‘onsets’ at each place the 1s are (so e.g. stimulus 1 was 8 10 12 14 16 18 64 etc.). Is one way more accurate for testing multicollinearity than the other? As all 3 stimuli are quite long (12 s) and the TRs between them are short, relatively (8s on average) I want to make sure I can resolve each stimulus.
- I will also be shocking people which I’d like to input as a regressor of no interest so can I use a binary file with the -ortvec option to ensure there will be no shock contamination?