Hi, I tried to run a three level analysis in FEAT on longitudinal fMRI data. Level 1: run - fixed effects Level 2: combining time 1 and time 2 runs for each person - fixed effects Level 3: running group analysis on combined data - FLAME mixed effects The problem that I encountered was an error in the "difference" contrast at level 2. I had hoped to created an "average" contrast (1 1) and two difference contrasts (1 -1 and -1 1). My guess is that - even though I've stayed a fixed effects analysis - the fact that you can fit a line perfectly between two points is upsetting FEAT. The average contrast seems to work, but not so for either of the differences. Any thoughts? Do I have to switch to a paired t-test two level design? Thank you for your time Kx Sent from my iPhone, please excuse any typos or excessive brevity