3dDeconvolve \
-nodata \
-polort 1 \
-censor new_ideal_run1.1d[4] \
-num_stimts 3 \
-stim_file 1 new_ideal_run1.1D[0] \
-stim_file 2 new_ideal_run1.1D[2] \
-stim_file 3 new_ideal_run1.1D[3] \
-stim_label 1 c1 \
-stim_label 2 c2 \
-stim_label 3 c3 \
-stim_minlag 1 2 -stim_maxlag 1 6 \
-stim_minlag 2 2 -stim_maxlag 2 6 \
-stim_minlag 3 2 -stim_maxlag 3 6 \
-stim_nptr 1 10 \
-stim_nptr 2 10 \
-stim_nptr 3 10 \
-xout

3dDeconvolve error: Insufficient data for estimating 17 parameters;

==========================

I have 2 more questions :

1. If the design matrix doesn't invert can I do soe tricks to make it invertible? (e.g to consider one o the conditions as a baseline instead of fixation period that was supposed initially to be baseline and to consider the fixation as a new condition instead?)

2. How do I decide that one inverted matrix is better then another one. If I change 2-3 lines in my ideal file the matrix is invertible . What are the criteria of a good matrix?


Thank you.