Comment

Seth Meyers: The Trump Administration Gives Up on Controlling the Coronavirus Pandemic

146
No Malarkey!10/26/2020 9:55:23 pm PDT

re: #143 Love-Child of Cassandra and Sisyphus

It’s one of the problems with Nate’s model: it has built in preference for regions/sets-of-states to be more sensitive to a few select outcomes in particular states, based on past performances.

Now the past influences us today. But the usual disclaimer forced upon the financial industry is past performances do not guarantee future results, and that ought to be added in heavy doses to the 538 model.

See above. The model appears to also link historically the migration into Iowa, as the population of Iowa was early on settled by people from Ohio, whose families a generation earlier came from PA. Demographical connections tied to history.

Nate links them because the key error Wang made was treating each state election as an independent variable, making it extremely unlikely Trump could overcome the odds in all if them, when in reality it’s a national election so demographically similar states will move in tandem, so if Trump was gaining in one of them, he was probably gaining in all if them.