How is it possible that it took until 2015 to discover the mistake in the original 1985 research (Gilovich et al.) "proving" that hot hands in basketball don't exist?
It became mainstream news, discussed in places like ESPN and has been repeated countless times in popular science books. Furthermore, the debunking in the Miller and Sanjurjo article referenced in the parent's blog post is not the result of arcane mathematical analysis, but uses extremely basic math that any undergraduate could understand.
It's almost shockingly hard to believe that no one noticed the problem before, when the result was the subject of such passionate, widespread controversy, and so many people had the tools to see the flaw....
And not only that, the edge case condition makes the flaw painfully obvious. From the Miller and Sanjurjo article's description of the original, flawed research: "...a streak is typically defined as a shot taken after 3, 4, 5, . . . hits". Looking at only the case of a shot taken after streaks of length 3, the miss rate is 100% (because a hit would imply the streak actually had length 4).
Usually when the popular press talk about a study, they don't link to the original study. So the number of people who talk about it is much higher than the number of people who might actually go and re-examine the study.
It became mainstream news, discussed in places like ESPN and has been repeated countless times in popular science books. Furthermore, the debunking in the Miller and Sanjurjo article referenced in the parent's blog post is not the result of arcane mathematical analysis, but uses extremely basic math that any undergraduate could understand.
It's almost shockingly hard to believe that no one noticed the problem before, when the result was the subject of such passionate, widespread controversy, and so many people had the tools to see the flaw....
And not only that, the edge case condition makes the flaw painfully obvious. From the Miller and Sanjurjo article's description of the original, flawed research: "...a streak is typically defined as a shot taken after 3, 4, 5, . . . hits". Looking at only the case of a shot taken after streaks of length 3, the miss rate is 100% (because a hit would imply the streak actually had length 4).