I have data from an unpublished experiment on factors affecting calibration. People with higher education levels were better calibrated, and people from more “rational” occupations like “scientist” or “computer programmer” were better calibrated than people from less “rational” occupations like sales or design. People whose job involved working with risk and probabilities directly (eg investment banker, weatherman) were just below scientists.
There was an attempt to investigate whether gambling helped, but it got contradictory results: frequent gamblers were worse calibrated on questions like “What percentage chance France is bigger than Germany?” but better calibrated at predicting future events. I don’t understand why this happened.
This doesn’t distinguish between education and science education making people more rational, and more rational people getting more education and going into science, but it’s at least a little positive.
With the gamblers, there could be two factors fighting it out:
1.) Maybe you’re more likely to gamble lots in the first place if you’re badly calibrated because you think you’re more likely to win than you actually are.
2.) But once you gamble regularly (and if you keep doing so) you might need to develop the ability to predict the future states of games.
No 2 seems to explain the calibration with regards to future events. Maybe gamblers are more likely to display overconfidence with regards to other calibration tasks though as this overconfidence helps explain why they would choose to gamble.
Just a guess though, no solid reason to believe it.
I have data from an unpublished experiment on factors affecting calibration. People with higher education levels were better calibrated, and people from more “rational” occupations like “scientist” or “computer programmer” were better calibrated than people from less “rational” occupations like sales or design. People whose job involved working with risk and probabilities directly (eg investment banker, weatherman) were just below scientists.
There was an attempt to investigate whether gambling helped, but it got contradictory results: frequent gamblers were worse calibrated on questions like “What percentage chance France is bigger than Germany?” but better calibrated at predicting future events. I don’t understand why this happened.
This doesn’t distinguish between education and science education making people more rational, and more rational people getting more education and going into science, but it’s at least a little positive.
With the gamblers, there could be two factors fighting it out:
1.) Maybe you’re more likely to gamble lots in the first place if you’re badly calibrated because you think you’re more likely to win than you actually are.
2.) But once you gamble regularly (and if you keep doing so) you might need to develop the ability to predict the future states of games.
No 2 seems to explain the calibration with regards to future events. Maybe gamblers are more likely to display overconfidence with regards to other calibration tasks though as this overconfidence helps explain why they would choose to gamble.
Just a guess though, no solid reason to believe it.