Dear all,
I am running a Bayesian Analysis using the bayesmh command in Stata 16.1. The command runs perfectly fine, but I would like the customized prior for my main explanatory variable "treatment" to have a skewed normal distribution instead of its current normal distribution.
The help file lists many prior distributions to choose from, but as far as I can tell no skewed normal distribution. For the variable of interest "treatment" I currently define a prior that is normally distributed with mean = 1.920304 and var = 0.3538458. I would like to adjust the prior such that it is skewed normally distributed with mean = 1.920304, var = 0.3538458, and shape = -0.1574642.
Do you know how to achieve this? Any suggestions are greatly appreciated!
Thanks in advance and best,
Sarah
I am running a Bayesian Analysis using the bayesmh command in Stata 16.1. The command runs perfectly fine, but I would like the customized prior for my main explanatory variable "treatment" to have a skewed normal distribution instead of its current normal distribution.
Code:
bayesmh y treatment x i.trade##i.region, likelihood(normal({sigma2})) /// prior({y:treatment}, normal(1.920304, 0.3538458)) /// prior({y:x}, normal(0, 2.5)) /// prior({y:i.trade}, normal(0, 2.5)) /// prior({y:i.region}, normal(0, 2.5)) /// prior({y:i.trade#i.region}, normal(0, 2.5)) /// prior({y:_cons}, normal(0.8740444, 3.875811)) /// prior({sigma2}, gamma(1, 1)) /// nchains(4) mcmcsize(60000) burnin(20000) rseed(78643) /// block({y:x _cons}) /// block({y:i.trade i.region i.trade#i.region}) /// block({sigma2})
Do you know how to achieve this? Any suggestions are greatly appreciated!
Thanks in advance and best,
Sarah
Comment