Hi Statalist Community!
I am trying to run a bayesian analysis using the command bayesmh (I only have the 14.2 Version of STATA, so I can't use the bayes prefix). I would like to specify non-informative priors because I believe the probability (or weight) of each prior is the same and I thought this could be a way to do it. This means that my two hypotheses described by my two IVs would have the same probability of occuring (growth_1 and growth_2).
I just do not fully understand what this prior means, and if I should also flatten it or not to give the same prior odds to both variables.
. prior({sigma2}, igamma(.1,.1))
General analysis:
bayesmh voteshare growth_1 growth_2, likelihood(normal({sigma2})) ///
prior({voteshare: _cons}, flat) ///
prior({voteshare: growth_1}, flat) ///
prior({voteshare: growth_2}, flat) ///
prior({sigma2}, igamma(.1,.1)) ///
rseed(12345)
bayesstats ess
bayesgraph diagnostic {growth_1}
bayesgraph diagnostic {growth_2}
Could I also change the number of replications?
Thank you so much!
I am trying to run a bayesian analysis using the command bayesmh (I only have the 14.2 Version of STATA, so I can't use the bayes prefix). I would like to specify non-informative priors because I believe the probability (or weight) of each prior is the same and I thought this could be a way to do it. This means that my two hypotheses described by my two IVs would have the same probability of occuring (growth_1 and growth_2).
I just do not fully understand what this prior means, and if I should also flatten it or not to give the same prior odds to both variables.
. prior({sigma2}, igamma(.1,.1))
General analysis:
bayesmh voteshare growth_1 growth_2, likelihood(normal({sigma2})) ///
prior({voteshare: _cons}, flat) ///
prior({voteshare: growth_1}, flat) ///
prior({voteshare: growth_2}, flat) ///
prior({sigma2}, igamma(.1,.1)) ///
rseed(12345)
bayesstats ess
bayesgraph diagnostic {growth_1}
bayesgraph diagnostic {growth_2}
Could I also change the number of replications?
Thank you so much!
