I don’t get this criticism.

The models he - and others like him - make are probabilistic. 70% probability Hillary would win was actually accurate. 30% probability events happen all them time.

The ones with “skin in the game” in 2016 said there was a 99% chance that Hillary would win. And that they’d eat their shoe, etc, if Trump won. They were in so much disbelief that Trump could win they built models just for confirmation bias.

▲

jstanley 7 hours ago | parent [-]

> 70% probability Hillary would win was actually accurate.

Based on what?

	▲	AftHurrahWinch 7 hours ago \| parent \| next [-]
		This: https://vincentarelbundock.github.io/Rdatasets/doc/dslabs/po...
	▲	softwaredoug 7 hours ago \| parent \| prev \| next [-]
		Based on historically common polling errors and how state polling errors correlate. 70% is far more likely than 99%.
	▲	wahnfrieden 7 hours ago \| parent \| prev [-]
		Probably hadn’t expected https://en.m.wikipedia.org/wiki/Russian_interference_in_the_...