Hello,

I normally use 3dDeconvolve to extract the beta weights of my variables of interest. For example, I code three different vectors with 1s where conditions 1, 2 or 3 occurred. Then I extract the beta weights for conditions 1, 2, and 3 and submit them to a group-level parametric analysis.

The problem is that a couple of my subjects do not have enough trials in one of the conditions to do a traditional 3dDeconvolve analysis where one needs a minimum number of observations/condition to yield a reliable beta.

To be able to use all my subjects, I am considering doing a correlation analysis of the BOLD signal with a single vector that codes for conditions 1, 2, and 3, where the amplitude of the signal is expected to vary with condition. For example, signal for condition 3 will be highest, condition 2 will be lower, and condition 1 will be the lowest. Then I can ask where brain activity covaries with this single vector, and do a group-level analysis on the t-scores (converted to z-scores) associated with this vector.

Can you point me to documentation about how to carry out this analysis? Or maybe tell me a better way of going about analyzing these 3 conditions when some of my subjects have < 8 trials in some of the conditions?

Best,
Christine