How do I estimate confidence intervals / credible region for parameters of Beta Binomial distribution? Specifically: from the data with variable number of trial, e.g. there are several experiments with different number of trials and success counts in each. Also for parametrization I'd like to use the proportion and dispersion ($\mu$ and $\theta$) instead of the shape parameters ($\alpha$ and $\beta$) since I need confidence intervals on the proportion.
Asked
Active
Viewed 523 times
1
-
Relevant: http://mathematica.stackexchange.com/q/6498/57 – Sjoerd C. de Vries Jan 11 '17 at 20:23
-
Would you add in your definition of $\theta$ ? If you want credible intervals and/or regions, how are you characterizing the priors for $\mu$ and $\theta$. In short, I think you'll get more help if you add in a few more details. – JimB Jan 13 '17 at 02:16
-
Maybe a possible duplicate of http://stats.stackexchange.com/questions/288/estimating-beta-binomial-distribution ? – JimB Jan 13 '17 at 02:19