This article, written just before Trump's 2016 victory, is required reading for 2020https://www.google.com/amp/s/m.huffpost.com/us/entry/us_581ffe18e4b0334571e09e74/amp …
Show this thread
-
tldr; the author argues that a model that shows Clinton grabbing 265-340 electoral votes in 95% of simulated elections — and NO result <240 EVs — passes "the smell test" because it "produces plausible output" Quick refresher: Clinton wound up with 227 EVspic.twitter.com/VGZizRJIdS
Show this thread
I softened the blow a bit here — this author rallied his 20 years of experience in statistical model-building to question @NateSilver538's "professional competency" and "responsibility in reality checking" for *only* assigning Clinton ~2/3 odds of an electoral victory
-
-
The biggest irony in the end is that when people look back and sneer at the models that assigned such low prob to Trump, 538 is usually called out first, when in fact they they were the only ones in the mainstream saying "this ain't over, guys"
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
I wondered if that was where you were going. The slamming Nate got at the time was amazing. He still gets it for 2016, but from Trump supporters now. He was the least wrong of anyone and went out of his way more than once to flag the high uncertainty.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
If you told me this is satire, I’d believe you
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Yep. But for me the main takeaway is that they were both wrong. Silver merely less so. Of course, the election was hacked. 2016 proved that we can never rest easy on polls again.
-
You are underestimating Comey's last minute surprise and its effect on turnout. Sadly, Trump has plenty of support here in PA.. high rural turnout combined with poor turnout in heavy Democrat areas gave him a win here.. no hack necessary.
End of conversation
New conversation -
-
-
Why spend time in Michegan, if the election is pretty much in the bag. It's not like fascism is really really good at having people under estimate its dangers historically.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
He’s right about something - making a model “client ready” fir IBs often means removing a bunch of things the client doesn’t want to hear, then justifying it by calling it “garbage”. No one closes the tab when you’re busy confirming their priors
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.