Posted by Luiz Pessoa on March 19, 2001 at 13:03:38:
Hi,
I have a basic question on using 3dDeconvolve for IRF estimation.
When estimating two event types that have unequal frequencies,
should one correct for that in the GLT matrix? For example, if
I'm comparing the "area under the curve" for both event types
should the GLT weights be altered so that the frequencies are
taken into account (say by dividing +1 and -1 in the GLT matrix
by the total number of events of a given kind)?
(Alternatively, maybe the input stimulus functions could be
normalized beforehand.)
Thanks,
Luiz