Hi, I have fitted a glm with three different groups, using mri_glmfit (FS v4.0.3). I use the beta.mgh to get the adjusted means within an ROI, for each group. I'd like to make a bar or box diagram with the adjusted means/betas, but I'm not sure how to get the standard error of the betas so I can make error bars or a confidence interval.
It seems that mris_anatomical_stats won't take the beta.mgh as input. Is there another way to do this in FS? Or will I have to do it outside FS? All my contrasts have only one vector, so each comparison should reduce to a t-test (according to the Wiki). In general, given the p (sig.mgh) and df of a t-test, I can get the t-value, and given the difference between the betas, the df and the t, I can get the SE. But I assume this is a little more complicated than that? I guess I can't use the rvar.mgh or rstd.mgh, since these are based on the whole model and not just the two groups. Which brings me to a related question: In mri_glmfit, when you have more than two groups in your model, and perhaps also e.g. age as a continuous regressor, and you set a contrast to compare two of the groups, does the variance estimate (SE) include the variance of all the variables, i.e. all groups that are in the model including the ones not included in the comparison + whatever covariates there may be? If so, isn't it really closer to a Tukey's or Scheffe's than an ordinary t-test? But back to my main question: Is there a way to get a measure of dispersion or confidence for the betas associated with each of the three groups in my analysis? (And if there is, I probably can't just average that measure across all vertices in the ROI, right?) Thank you! -- yours, LMR
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
