3 ms·
> Bayesian statistics can inform hypotheses at any level of evidence This is wrong. You probably can employ bayesian causal inference if you know that two obse
by bildung 6y ago
> Bayesian statistics can inform hypotheses at any level of evidence
This is wrong. You probably can employ bayesian causal inference if you know that two observed variables have a causal relationship and want to know about the direction. But if you already know that, most of the riddle has been solved a priori. Double-blind RCTs are the gold standard.
- jcims 6y ago> Double-blind RCTs are the gold standard. They are also extremely slow. Appropriate for efficacy testing of new treatments, not as much for public health messaging at the onset of a crisis. It’s important to have a methodological basis for decision making in those early hours where information is sparse and of dubious quality. Without that, people panic and make short-sighted decisions that may undermine the legitimacy of their organization for years to come.
- bildung 6y agoYes absolutely. If things have to move fast, people/institutions have to refer to heuristics. But in my understanding this also is pretty much what the article described.