Gang Wrote:
-------------------------------------------------------
> > Oddly enough, the mainF, interaction, and
> intercept results without treating ratings as a
> random effect in
> > the lme command match my results from 3dLME
> while treating ratings as a random effect.
> >
> > Why is this?
>
> Michael, it looks like you could read the mind of
> 3dLME. You've done exactly what 3dLME does:
> whenever a model with a complicated random-effects
> structure fails, 3dLME switches to a model with
> just a random intercept.
>
Okay great. That helps a great deal. My results from lme match most of the results from 3dLME when there are no missing subjects.
> > How does 3dLME handle missing voxels?
>
> How do you code a missing value in brain imaging
> data? I thought that NA would not be an option?
>
> If you run lme in R, use option
> na.action="na.omit", which basically removes that
> data point from the analysis.
I've tried that option, but still was unable to replicate the 3dLME results. Is that what 3dLME uses? Also, I'm confused because I thought Mixed Models can handle missing timepoints within a dataset. If a repeated measures design has 3 levels, and one of the levels is missing (NA), wouldn't the model still include the other 2 timepoints? Not omit them?
