I Suzanne
if x only occurs during A then 1 0 0 0 tell you where x is active (and
it is necessarily during A)
now if you want to see where x is active while you also have a A effect
then indeed conjunction (likely null) is what you are after
cyril
> Greetings.
>
> I have what is effectively a mixed block/event related design. I have two block types (A and B) and two event types (x and y). The two events occur with equal probability in both of the blocks, so there shouldn't be any problems with double specifying the regressors (e.g., if block A only had x events and block B only had y events).
>
> My question is how to go about setting up the contrasts to be able to examine the events based on whether they occurred during block A or B. If my GLM is set up as {x y A B}, is the contrast for those x events that occurred during block A just [1 0 1 0]? This seems like this would give me the union of the two regressors, rather than the intersection (which is what I am after). Should I, instead, be looking at doing a conjunction between the contrast for all x events and the contrast for all A blocks?
>
>
> Thanks in advance,
>
> suzanne
>
--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.
|