Gang Wrote:
-------------------------------------------------------
> > 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?
>
> If one subject has missing data for a level, you
> just don't provide that specific level for that
> subject under -dataTable in 3dLME (instead of put
> a row with NA). And you can do the same in the
> dataframe for your R code.
Okay. Getting back to the case where a voxel is missing for one subject. It is not specified as NA in the -dataTable. This voxel could be on the outer edges of the brain, where it is not present for a subject(s) in comparison to other subjects in the analysis. Does 3dLME treat that as NA? I tried doing that for a peak voxel coordinate with one of our ROI's in R with lme. However, I get a different Z-value for a contrast. Nothing is specified as NA in the dataTable because we don't have any missing subjects. Just a few cases for some ROI's, the peak voxel coordinate location is not present for a few subjects.
Also, is there a way to generate Standard deviation/Standard error for the mean value of a factor/level in 3dLME or from 3dLME's output?
Thanks.
Michael
