Check the variance over time per regressor

AFNI version info (afni -ver): Precompiled binary macos_10.12_local: Apr 7 2023 (Version AFNI_23.1.00 'Publius Helvius Pertinax')


My experiment contains 8 runs and 8 regressors, of which 6 regressors appeared in the first 6 runs and 2 regressors only appeared in the last 2 runs. I am interested to see whether the subject-level variance of the two regressors in the last 2 runs is higher than the other 6 regressors due to fatigue. What will be a proper way to do such a test? Will -sresp in 3ddeconvolve serve the purpose?