I enjoyed your style and humor in this article! There are so many links to works that I'd like read. Maybe I'll send this to my neighbor with the "I believe in science" bumper sticker.
See also Richard McElreath's "Statistical Rethinking - Lecture 10, BONUS: Simpson's Paradox" (02/01/23): https://www.youtube.com/watch?v=jokxu18egu0&t=4246s. He agrees we should probably forget about the term because it doesn't really distinguish anything and there is no paradox. But adds a nice counterpoint about not being a definitional scold. And highlights the larger insight that teaching scientific logic is the cure to this sort of confusion that arises from our lack of statistical intuition.
I enjoyed your style and humor in this article! There are so many links to works that I'd like read. Maybe I'll send this to my neighbor with the "I believe in science" bumper sticker.
Many thanks from a bored epidemiologist.
A wonderful essay, thank you. Well-written and a pleasure to read...has motivated me to get back to J Pearl's book.
Michael J. Ward asks about citing the line "Don't be clever... Tool up and take every little step visually, for we are silly monkeys." - https://twitter.com/Michael_J_Ward/status/1618963716611112960
Richard McElreath clarifies "Avoid being clever" is from his lectures: https://speakerdeck.com/rmcelreath/statistical-rethinking-2023-lecture-06?slide=2.
And that he gets the notion of avoiding clever solutions from Edwin Jaynes's *Probability Theory: The Logic of Science*, Chapter 8: Sufficiency, Ancillarity, And All That: https://twitter.com/rlmcelreath/status/1618971253020454913.
Of which chapters 1-3 are here: https://bayes.wustl.edu/etj/prob/book.pdf.
And Chapter 8 is here: http://www-biba.inrialpes.fr/Jaynes/cc08n.pdf.
Nice piece, many thanks for sharing.
I'm wondering if some terms (like "Simpson's paradox"?) can still be useful in communicating the thruth, even if their name is wrong/misleading?
The problem with continuing to use "Simpson's paradox" is exactly that it's not useful, because it could refer to different phenomena.
See also Richard McElreath's "Statistical Rethinking - Lecture 10, BONUS: Simpson's Paradox" (02/01/23): https://www.youtube.com/watch?v=jokxu18egu0&t=4246s. He agrees we should probably forget about the term because it doesn't really distinguish anything and there is no paradox. But adds a nice counterpoint about not being a definitional scold. And highlights the larger insight that teaching scientific logic is the cure to this sort of confusion that arises from our lack of statistical intuition.