I just want to drop into this two good reasons for people making choices that don't look "good" to other people, as a warning against smugness:
1) RCTs are a very blunt instrument. What you're interested in is the (Bayesian) probability of a given action having a given outcome for you, given everything you know about yourself - not the frequentist probability of it having that outcome for a randomly selected member of a group you're in.
2) The crude outcome usually measured is not the only aspect of the outcome of interest to you; you're interested in all aspects of the outcome, including your anticipated regret, which is tied up with your own situation and psychology and not well second-guessed by others. For example, given a treatment that reduces the chance of death by not much, but increases its uncertainty and, if it fails, has the prospect of reducing the quality of life in the last months, do you take it? The choice is so personal and contingent that probably few of us know what our own answer would be, never mind what someone else's should be.
Well said.
