afni.nimh.nih.gov

> Using 3dLME can we also omit the 3 way interaction (diagnosis X load X direction) and only look at the
> 2 way interactions (diagnosis X load & diagnosis X direction & all the two-way interactions with age)?

Just curious: is the three-way interaction result too disturbing?

You can try:

-model 'Dx*Direction*Age+Dx*Number*Age+Direction*Number*Age'

Gang