A question of science. Does any of you smart people know
how precise is the average human in evaluating (not counting nor calculating) probabilities? The kind of experiment I imagine is: an observer watches two different repeating random yes/no events, what difference in likehood between the two events is necessary for the observer to determine that they are different? I guess this kind of things has been studied?
That is likely to have been studied within the field of psychology, but it is also equally likely that the methodology is skewed due to the selection bias. (http://www.slate.com/articles/health_and_science/science/2013/05/weird_psychology_social_science_researchers_rely_too_much_on_western_college.html)
RépondreSupprimerOf course. Trouble is 1. guessing the right keywords for looking it up myself 2. finding it in freely accessible articles.
RépondreSupprimerTerry Olson oh yes. I can't wrap my head around that one.
RépondreSupprimerLuke Wayland yes I know of Ariely's work. His popularization books are excellent.
It has been studied. People aren't very rational. So, they don't think in terms of probabilities. And, to answer your question, other factors would be more important than a difference in probabilities between the two events.
RépondreSupprimerYup! But I'd like some number. Can we, on average, distinguish 1/4 from 1/2? (i'd say yes) 3/8 from 4/8? (no idea, but not very optimistic) 49% from 50%? (certainly not) What is the average treshold, what is the variation range, which factors influence that?
RépondreSupprimerGherhartd Sildoenfein, I don't know if the wiki article touches on this, but the trick to the MH problem is to imagine a many-door limit rather than just 3 doors. For example, if there are 1000 doors, you have a 1/1000 chance of picking the right door. Monty opens 998 of the remaining 999 doors to show you they're empty. It's very likely (999/1000 chance) that the right door is the one Monty didn't pick. Since you picked before additional knowledge of bad doors, you chances of 1/1000 do not improve. Hope that helps!
RépondreSupprimer