I love that the first thing he writes is an essay titled “Statistical objectivity is a cloak spun from political yarn”, even more so because he quotes another intellectual hero of mine, Jim Scott.
“So, nowadays we work with millions and millions of observations, then we had 14 observations, so if you’re going to try to tell a story about this you’ve got to put a lot of structure on it because you only have a little data to estimate these points, so it’s almost like you’ve got a Christmas tree and you’re using the data to hang decorations on it or something and the data’s not going to speak very much.”
He drove headlines again when he and Anne Case published a study showing death rates are rising among middle-aged white men in the US. There was even a story about how it got rejected from several medical journals. It’s somehow comforting to know that even when you win the Nobel Prize, journal rejections sting.
If you read my FP piece on why Deaton deserved the Nobel, you might recall me mentioning his latest unpublished critique of randomized experiments. It remains unpublished as far as I know, but for an overview, read Tim Ogden’s interview with Deaton.