Wow, that’s a great point. We can’t measure anyone’s “true” calibration by asking them a specific set of questions, because we’re not drawing questions from the same distribution as nature! That’s up there with the obvious-in-retrospect point that the placebo effect gets stronger or weaker depending on the size of the placebo group in the experiment. Good work :-)
Wow, that’s a great point. We can’t measure anyone’s “true” calibration by asking them a specific set of questions, because we’re not drawing questions from the same distribution as nature! That’s up there with the obvious-in-retrospect point that the placebo effect gets stronger or weaker depending on the size of the placebo group in the experiment. Good work :-)