Plotting credible intervals from a bayesian model

[Dougie Jones]

I have created a bayesian model and now I'd like to visualize the model's output by plotting the credible intervals for certain parameter. How would I do that?

sysuse auto, clear
set seed 12345

bayesmh mpg foreign, likelihood(normal({var})) prior({mpg:_cons foreign}, flat) prior({var}, jeffreys)

The above code is just a working example, but how would I plot the posterior mean of foreign and its credible intervals? Additionally, some assistance in making a triplot (likelihood, prior probability and posterior probability) would be much appreciated.

[Dougie Jones]

No one?

[Leonardo Guizzetti]

This is not a complete answer, and I am not yet familiar with Stata's Bayesian estimation commands. It may be interesting to you to examine the posted estimation results, using -ereturn list-. One of the listed results is called "cri" which seems to be the credible intervals. Following your example code, you can view the contents of -e(cri)- using -matrix list e(cri)- which gives the following output.

. matrix list e(cri)

e(cri)[2,3]
             mpg:       mpg:          
         foreign      _cons        var
Lower  2.0341353  18.271162  20.827042
Upper  7.7758433  21.358019  41.501285

Indeed, these are the credible intervals. I think this will give you enough to get started.