Standard errors to standard deviations in mixed models

[Suzanne Kuijpers]

Dear readers,

For a study into differences (e.g. liver values) between different dosages of a specific drug I am using linear mixed models in stata. All results are given as Standard errors of the mean (SEM), however I need these results to be given with Standard deviations (SD) per dosage. I have tried the postestimation commands "estat summarize" but these values do not seem to be correct for my results. Also ending the command with "stddev" does not seem to lead to the correct results. Another possibility is to calculate it with the number of observations however this is hard in my mixed mode as results are clusteredl. Is there a command that I am missing? Or is there a different way to calculate the SD?

Thank you in advance!

[Joseph Coveney]

All results are given as Standard errors of the mean (SEM)

Where do you see that? I see "Coef. Std. Err." at the top of the fixed-effects section of the regression table after mixed, which is what I assume that you're using.

I need these results to be given with Standard deviations (SD) per dosage.

Perhaps you'd be better off computing a summary of the values for each patient or animal over time. Something like a time-weighted average or an area under the curve of values recorded during the time course of observation. (I assume that you're dealing with some kind of repeated observations, and that the liver values are all nonnegative.) Then you have a single independent observation summary value for each patient or animal, and you can compute conventional summary statistics.