4 ms·

"Clinton has 90 percent chance of winning: Reuters/Ipsos States of the Nation" - Nov 8 - http://www.reuters.com/article/us-usa-election-poll-idUSKBN1322J1 http:

"Clinton has 90 percent chance of winning: Reuters/Ipsos States of the Nation" - Nov 8 - http://www.reuters.com/article/us-usa-election-poll-idUSKBN1322J1 http://www.reuters.com/article/us-usa-election-poll-idUSKBN1...

iriche 10y ago
Which news outlet didn't say that?
- swang 10y ago
  seriously. fox news. cnn. msnbc. EVERYONE said it. 538 said it was 66/33 or 75/25 or something similar (i forget, but it wasn't 90/10 in favor of Clinton) and they got roasted over it when they don't even control any of the polls.
- dragonwriter 10y ago
  The most well-regarded poll-based predictor, 538, gave Clinton a much lower than 90% chance and pointed out the key error of those predictors giving that high of a chance -- they used models which assumed, contrary to evidence, that deviations from polling results in different states in the same Presidential election are independent, when in fact they are strongly correlated.
khuey 10y ago
Trump winning doesn't prove they were wrong. That's not how p-values work.
joekrill 10y ago
OK, and your point? They said 90 percent. Not 100.
- alexc05 10y ago
  Exactly - I've played enough Dungeons & Dragons to know that it is possible to roll 1 or 2 on a D20. If there was a 90% chance of winning, it is the 10% which happened. That's pretty clear by how crazy close it wound up being to. Just 80,000 specific votes in specific states are outweighing more than 3,000,000 across the rest of the United States. That really feels like something that had about a 10% chance of happening (or less really).
akvadrako 10y ago
You got my up-vote; thanks for reminding us. Clearly lots of people still want to think Trump's winning was a fluke. But the LA Times polls, which had a unique methodology, predicted it. And Scott Adams gave quite reasonable explanations of why. Any news outlet that didn't give him at least close to a 50% chance wasn't doing their job. Personally, I made a few grand betting on https://www.predictit.org https://www.predictit.org, because it was pretty clear there was widespread delusion going on.

iriche 10y ago

Which news outlet didn't say that?

swang 10y ago
seriously. fox news. cnn. msnbc. EVERYONE said it. 538 said it was 66/33 or 75/25 or something similar (i forget, but it wasn't 90/10 in favor of Clinton) and they got roasted over it when they don't even control any of the polls.
dragonwriter 10y ago
The most well-regarded poll-based predictor, 538, gave Clinton a much lower than 90% chance and pointed out the key error of those predictors giving that high of a chance -- they used models which assumed, contrary to evidence, that deviations from polling results in different states in the same Presidential election are independent, when in fact they are strongly correlated.

khuey 10y ago

Trump winning doesn't prove they were wrong. That's not how p-values work.

joekrill 10y ago

OK, and your point? They said 90 percent. Not 100.

alexc05 10y ago
Exactly - I've played enough Dungeons & Dragons to know that it is possible to roll 1 or 2 on a D20. If there was a 90% chance of winning, it is the 10% which happened. That's pretty clear by how crazy close it wound up being to. Just 80,000 specific votes in specific states are outweighing more than 3,000,000 across the rest of the United States. That really feels like something that had about a 10% chance of happening (or less really).

akvadrako 10y ago

You got my up-vote; thanks for reminding us. Clearly lots of people still want to think Trump's winning was a fluke. But the LA Times polls, which had a unique methodology, predicted it. And Scott Adams gave quite reasonable explanations of why. Any news outlet that didn't give him at least close to a 50% chance wasn't doing their job. Personally, I made a few grand betting on https://www.predictit.org https://www.predictit.org, because it was pretty clear there was widespread delusion going on.