3 ms·
Is it not rational for doctors to take a Bayesian approach and weight their diagnoses accordingly? Not without having very detailed information on exactly how
by dagw 8y ago
Is it not rational for doctors to take a Bayesian approach and weight their diagnoses accordingly?
Not without having very detailed information on exactly how big the difference is and, equally important, how these differences are distributed within each sub-population.
- yasp 8y agoSo Bayesianism is good but only sometimes?
- dagw 8y agoBayesian reasoning is good when applied rigorously, but can be dangerous when applied sloppily and lazily, since people are terrible at subjectively estimating probabilities and tend to be way over optimistic when updating their priors with any new information they receive. Basically people will treat information that should realistically cause them to update their priors from 50-50 to 51-49 as if it was information that should cause them to update to 80-20 (and then they round off too 100-0).