Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | "JVerkuilen (Gmail)" <jvverkuilen@gmail.com> |
To | statalist@hsphsun2.harvard.edu |
Subject | Re: st: Tobit regression Model |
Date | Sun, 2 Dec 2012 14:12:37 -0500 |
On Sun, Dec 2, 2012 at 1:51 PM, Carlo Lazzaro <carlo.lazzaro@tiscalinet.it> wrote: > I agree with Jay about the trade-off between research assumptions and > empirical evidence. > About Jay's second point, my smattering of the Bayesian framework is limited > and, at least in Italy, it is not common at all. I usually go (empirical) > Bayesian for running probabilistic sensitivity analysis (I'm a health > economist) and play on the safe side of the matter with conjugated > distributions. Others may find hard times in dealing with more intimidating > facets of Rev Bayes' machinery (eg: improper priors)! > Here you wouldn't be using improper priors, you'd need informative ones to accommodate the separation issue that likely lurks around, because the prior is providing information about the parameter that's not coming from the data. That's not very much fun either, as the resulting sensitivity analysis to ascertain the effect of the prior on posterior inference on the other variables won't be easy. Tools like OpenBUGS (http://www.openbugs.info/w/) makes it MUCH easier to do this than in the past, but it requires a great deal of specialist knowledge. A good Bayesian analysis takes quite a bit of time to do well, longer than that of a similarly well-executed ML analysis. JV * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/faqs/resources/statalist-faq/ * http://www.ats.ucla.edu/stat/stata/