This is a new one on me. Applying the BIC
retrospectively makes an analysis Bayesian?
My prior probability is 0.999 that the
answer to Brian's question is No.
Stata is not much of a vehicle for doing Bayesian
things; nor do Bayesians much use Stata for being Bayesian.
(John Carlin is a case in point.)
By the way, in 2002 I implemented -cij- for
using the so-called Jeffreys prior for binomial
confidence intervals (itself a mishmash of Bayesian
and frequentist thinking). When StataCorp
took the same idea aboard in -ci-, all the Bayesian terminology
was thrown away, and my hidden option for supplying
your own prior also disappeared.
Nick
[email protected]
Paul Millar
> I have written a several routines that make use of the BIC statistic
> for significance of individual variables within a model:
>
> scc describe bicdrop1
> ssc describe bic
> ssc describe listmiss
> >Has anyone written an analysis in Stata for doing Bayesian analysis
> >to see if model A fits the data better than model B in the case
> >where model B is not a subset of the variables in model A but rather
> >a different model such as a power law versus a log-normal?
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
