How to extract SD for fixed effects in mixed models (using R, lmer)? I would like to have the standard deviation for the fixed effects output in R. By default, the lmer function from the lme4 library only gives me the standard error. Is there a way to achieve this?
 A: I'm not sure that information is useful or even obtainable, and it certainly isn't common practice to estimate. The reason you get standard deviations for random effects in these models is because it allows you to summarize the fluctuation in many random intercepts/slopes. The same can't really be done with fixed effects because by definition they are the overall effect of the fixed predictors and aren't disaggregated into different clusters. You could get something like a standardized $\beta$ coefficient to estimate their independent effects on the outcome, but that wouldn't really give you the same information as standard deviations and doesn't account for the random effects structure anyway.
If your goal is simply estimating the standard deviation of the fixed effects after estimating the random effects, this can be somewhat achieved by using by-group estimates of standard deviations of the predictors (using whatever you used as random effects clusters for example). I would otherwise be unsure of how this information could be obtained in a meaningful way.
A: (I (OP) wanted to reply to Shawn Hemelstrand, however I can no longer access the unregistered account).
Your answer helps me a lot. I was asked by a reviewer to include the SD as part of an initial data analysis. I assumed that there is a reason why the SD are not automatically output, but was unsure and tried to comply with the request, but could not find any corresponding approaches. Your explanation makes sense to me and underlines that it is not a common procedure. Thanks a lot!
