Can anyone think of a decision which might come up in ordinary life where Baysian analysis and frequentist analysis would produce different recommendations?
The core difference between B and F is what they mean by “probability.” If you go to the casino, the Bs and the Fs will interpret everything the same way, but when you go to the stock market, the Bs and the Fs will want to use their language differently. It seems likely to me that most of the uncertainties that show up in everyday life are things that Bs would be comfortable assigning probabilities to, but Fs would be hesitant about.
When it comes to an action you must structure your knowledge in Bayesian terms to use to compute an expected utility. It is only when discussion detached knowledge that other options become available.
??? This isn’t true unless I misunderstood you. There are frequentist decision rules as well as Bayesian ones (minimax is one common such rule, though there are others as well).
Can anyone think of a decision which might come up in ordinary life where Baysian analysis and frequentist analysis would produce different recommendations?
The core difference between B and F is what they mean by “probability.” If you go to the casino, the Bs and the Fs will interpret everything the same way, but when you go to the stock market, the Bs and the Fs will want to use their language differently. It seems likely to me that most of the uncertainties that show up in everyday life are things that Bs would be comfortable assigning probabilities to, but Fs would be hesitant about.
When it comes to an action you must structure your knowledge in Bayesian terms to use to compute an expected utility. It is only when discussion detached knowledge that other options become available.
??? This isn’t true unless I misunderstood you. There are frequentist decision rules as well as Bayesian ones (minimax is one common such rule, though there are others as well).
In what sense is minimax frequentist?
From Wikipedia:
ETA: While that page talks about estimating parameters, most of the math holds for more general actions as well.
I don’t think that “non-bayesian” is a common definition of “frequentist.” In any event, it’s not a useful category.