Meta-analysis of standard deviation using the `metafor` package in R: can we distinguish between the different types of variability? I am doing meta-analysis of standard deviations of parameter X (gastric pH). My collated studies are conducted on humans and each study report mean plus/minus SD of gastric pH in the studied population. 
I already did the meta-analysis of the standard deviations and got a meta-SD estimate. I used the metafor package in R. 
I assume that the meta-estimate of SD is a composite of between subject variability, within subject variability, between occasion variability in gastric pH as well as between study heterogeneity. 
My question:
Is there a way to differentiate between these different types of variability? or at least differentiate between (between study heterogeneity in the meta-SD estimate versus other types of variability?)
 A: Since you say that you are a learner you might be interested in the distinction between fixed effects (FE) and random effects (RE) models. Although these are often treated as just variants of one another in fact they are distinct models. Let us imagine their proponents speak to us.
Doctor FE says: I believe that the studies I have collected all estimate the same underlying parameter and so I will weight each estimate by its precision in order that the more precise studies get more weight.
Doctor RE says: I believe that there is no single underlying parameter and that my job is to estimate the parameters of its distribution. Since I assume it to be normal I can estimate its mean and variance. I shall then use weights based on both the variation between the studies and the precision of the individual studies. In the extreme this will lead to equal weights when there is much heterogeneity.
Note though that the heterogeneity is not accounted for in the sense of explained by the model. As you notice @Wolfgang in his comment carefully said it was '... taken into consideration ...'. If you have suitable moderator variables you may go on to explore the heterogeneity further of course but that is another story.
