People are often *really* bad at seeing high-vol scenarios.
A few hours into the election, it was clear that:
1) Trump might win the election with >270
2) Trump might lose the election with < 200 EV
A 2% swing in the national climate would have swung ~100 EV. twitter.com/adamscochran/s
Conversation
With the electoral seat balance, Trumps best scenario was likely clinching a 278, way less paths to victory & the states he had the best chance in also had the strongest chance for 3rd party candidates.
This was all about skewed sampling of early vote which gen pop didn't see
1
4
Replying to
eh agree once more votes had come it--but when it was just FL there was a significant chance that the polling error in Miami-Dade was more generalized than it turned out to be.
1
1
1/2 That was bad pollsters reducing people to numbers. Cuban Americans make up a huge chunk of Miami-Dade, Trump hit South Florida with millions in Spanish Ads decrying socialism which packs a bigger weight with Cuban-Americans compared to other LatinX groups. But..
1
2
2/2
No national pollster did a specific Miami-Dade poll & very few national pollsters do Spanish language polling.
They LatinX pop across the rest of FL is more diverse, younger and from countries with less Socialism baggage.
Old analyst adage - no such thing as “avg user”
2
2
Replying to
yeah mostly agree and I think this was a totally reasonable interpretation
but I think it's hard to see TRUMP outperform by 4% in the end and think that, from early results where he did even better in an idiosyncratic case, the odds were *really* low that it would have been 5%
Totally, very hard. For that to even be on the radar (never mean to seem plausible/likely) takes familiarity with stats backgrounds, lots of demographic research, lots of understand of US politics, etc, etc.
That’s why the the prediction market v experts argument has gaps.
1