Hi Gang,
I code the vectors as a time series of 1s and 0s and then let 3dDeconvolve figure out the IRESP via min and maxlag. I do not preconvolve the vectors.
I like the idea of 3dMEMA, but I don't see an example where I could test for a linear trend in my 3 conditions. I would normally use 3dRegAna coding the three conditions with 1, 2, and 3. I don't see an equivalent way to do this in 3dMEMA. The closest thing I found was Example 3, below.
3dMEMA: Example 3 --- Paired type (difference of two regression coefficients or
general linear contrasts from each subject in a group). One scenario of
general linear combinations is to test linear or higher order trend at
individual level, and then take the trend information to group level.
In this case, I would test the linear trend with a glt within 3dDeconvolve and then use 3dMEMA to do a simple t-test of that beta vs. zero. Does that sound right? But would the beta and t-value associated with the glt reflect the fact that some subjects may have fewer trials than others for one of the three conditions tested with the glt?
Or is there a way to do a linear test within 3dMEMA on the betas and their t-values?
Christine