Bayesian Approach: Limitations
No matter what prior you pick, people will attack you
- “Principle of Flat Priors: Use flat priors as an approximation when the prior precision is small in comparison with the sample precision” (Berry 1996, p. 354)
- “Using flat priors is conservative when the actual prior evidence is consistent with the sample data” (Berry 1996, p. 354)
Not simple when doing joint probability
Bayesian results cannot be averaged, added, etc.
- an average or a sum must be recomputed from first principles