,
3 tweets,
1 min read
Read on Twitter
Your hypothesis relies on data that's (i) 8 months old (ii) was collected mostly online (fine but not gold-standard) (iii) involves a lot of statistical extrapolation (iv) is calibrated to 2018 rather than 2020 turnout (v) is based on approval rather than head-to-head matchups.
All of which is… fine… to get a rough approximation. But the real-world margin of error on those estimates is (at least) several points wide, which is much too large to come to many firm conclusions about the tipping point, especially if you're not working probabilistically.
There's also a fairly large volume of polling that shows Biden (and to some extent Bernie) performing quite well vs. Trump in head-to-head matchups in the Midwest. I wouldn't put a ton of stock in that data, either. But it's not just one poll in Ohio. 53eig.ht/2IWBgHX
