171: The easiest person to fool is yourself (with Daniel Simons and Christopher Chabris)
Everything Hertz
00:00
The Consistency of Random Data
People who want their data to be really clean like the idea that, oh, the two groups were identical to start with. They were always the same so that you can say any improvements are just due to what we are intervention. And that's just not what random data look like, right? So that's how that test works. No, not at all. There was a intrinsic lack of variation. Like the complete lack of something where there should be intrinsic variation and there just isn't. The next level of complication is probably intrinsic noise with patterns within the noise. Then the third level is probably something that's rule governed but difficult to discover. Which you also see on that basis.
Play episode from 36:17
Transcript
