Calculation of SEof mean in STAN


#1

I am interested to know the explanation of SE(mean) in STAN model output for model co-efficient. How it is being calculated?

Any help would be highly appreciated!

Thanks!


#2

An MCMC sampler produces an estimate of the mean. The error in that estimate depends on the number of effective samples \mathrm{N_{eff}}.
The SE column in the output is the Monte Carlo sampler error (MCSE) which is proportional to 1/\sqrt{\mathrm{N_{eff}}} instead of 1/\sqrt{\mathrm{N}}. As \mathrm{N_{eff}} increases, the MCSE approaches 0 and the estimated parameter mean approaches the true mean.


#3

Thanks Morris!