![Profile Image](https://pbs.twimg.com/profile_images/378800000628384447/c9cb85dbdb73c8bac4f28ce4cc46f38b_normal.jpeg)
Jason Furman NARA @CEAChair44
Not obvious the right answer would be 50/50, especially when ex ante you would expect more measure error from GDI https://t.co/lwhXNKL1BD — PolitiTweet.org
Marc Goldwein @MarcGoldwein
.@CEAChair I'm surprised we need a study to show that averaging two imperfect measures improves accuracy. Isn't that the rule of averages?