538 Poll Predictions

[Ed W]

Nate Silver and 538 have the first aggregated predictive poll up for this election season. Scroll down to near the bottom if you're statistically inclined and want to know the methodology.

Please remember that just like last time, one candidate is whining that the polls are biased against him.

http://projects.fivethirtyeight.com/2016-election-forecast/

[cannon_fodder]

When The Donald figures out there is an electoral college, he's gonna be so mad.  Even if the Libertarians were not siphoning off votes, its a tough sell based on the electoral college map.

Lets pretend The Donald doesn't lose any of the states for him that are in play (if we are counting a sub 80% mark as "in play," The Donald could lose Missouri, Georgia, SC, Mississippi (!), Kansas, Indiana, Texas (!), Utah, South Dakota, Montana, North Dakota, and Alaska), lets start counting off the States Hillary would have to give up to for The Donald to win the election:

Arizona - 11
North Carolina -15
Colorado - 9
Ohio -18
Iowa -6
Virginia - 13

That is nearly EVERY single state that with a less than 80% confident interval for Hillary winning. If we throw in Florida (where The Donald has never polled above the margin of error) and The Donald doesn't lose any of his swing states, then we have President The Donald.

Obviously things can change and polls can be wrong. But expect to hear a ton of whining about the system being rigged by those pesky Founding Fathers...

[davideinstein]

I just am not counting him out after he just won the nomination. It's a trap.

[Townsend]

I wonder how many non-trumpers would have to skip the election for him to win.

Because complacency in this case frightens me.

As John Oliver stated...sort of, "There are no turducken do-overs."

[Conan71]

Two days before the election Nate Silver was still saying Hillary had a 74% chance of winning.  It's pretty crazy how wrong the polls were, but it's been repeated constantly that Carter was ahead of Reagan by 9% in the week prior to the 1980 election.

Any guesses as to what went so wrong on the predictions?  The unspoken Trump supporters who felt too intimidated to admit they would vote for him?  Did Democrats just not show up because they felt defeated after the primaries?

[AquaMan]

26% is what happened. The polls were based on assumptions that made them predict a 3 in 4 chance she would win. Run that election 3 more times and see if she doesn't win 3 times. Then criticize Silver.

As far as why, I think the minority populations of smaller states did not break her way in the same numbers as they did the last two elections. There was a 1% decline in voting numbers as well and of course the efforts by states like North Carolina to suppress voter engagement.

[cannon_fodder]

Many of the polls were within the margin of error, but leaned towards Hillary. Florida, Pennsylvania, Michigan, Wisconsin.  Polls are an exercise in probability.

The other factor was that many people who would not be considered "likely voters" actually turned up. An angry man who hasn't voted in the last 3 elections isn't counted in the polls as a "likely voter." This year, those guys showed up. If you try to factor things like that in, you are no longer engaged in probability, you are trying to prognosticate.

Finally, we had the FBI announcement 3 days before the election that Hillary's investigation would be reopened. Over the weekend we had Fox News announce that an indictment was "eminent." Few polls could take that impact into account, the aggregate polls could have been destroyed by that. No way to really measure the impact.

Hard to see what could be "fixed" in the polls. Other than people who voted this election that usually don't will now be notched closer to "likely voters."

[BKDotCom]

The other factor was that many people who would not be considered "likely voters" actually turned up. An angry man who hasn't voted in the last 3 elections isn't counted in the polls as a "likely voter." This year, those guys showed up. If you try to factor things like that in, you are no longer engaged in probability, you are trying to prognosticate.

Poll all eligible voters.
Ask them when they last voted... ask them if they plan on voting in this election.    It's still probability.

Personally, I wish they did away with all the reporting of poll results.   It affects the very thing they're trying to measure.

[heironymouspasparagus]

The results are an artifact of the electoral college.   Fewer voters is a sad commentary, but since she is still 220,000 votes ahead, it just shows how divided we are.  At least some of Johnsons votes would go to her, but I suspect most would have gone to Trump, so it may well have put him in the lead by vote.

Not likely to change anytime soon, either.  And the reactions will continue.  World keeps on turning.

When Obama won...

http://www.nydailynews.com/news/politics/republicans-react-obama-win-anger-gloom-calls-fight-article-1.1198334


And now for a little mood music;

https://www.youtube.com/watch?v=InRDF_0lfHk&list=RDInRDF_0lfHk&index=1

[erfalf]

The results are an artifact of the electoral college.   Fewer voters is a sad commentary, but since she is still 220,000 votes ahead, it just shows how divided we are.  At least some of Johnsons votes would go to her, but I suspect most would have gone to Trump, so it may well have put him in the lead by vote.

Not likely to change anytime soon, either.  And the reactions will continue.  World keeps on turning.

When Obama won...

http://www.nydailynews.com/news/politics/republicans-react-obama-win-anger-gloom-calls-fight-article-1.1198334


And now for a little mood music;

https://www.youtube.com/watch?v=InRDF_0lfHk&list=RDInRDF_0lfHk&index=1

You see the difference though right?

Trump is "controversial". Celebrities and such commenting on Trump today, not so much.

[erfalf]

I have come to the conclusion that Trump was elected by people who did not feel they were being heard. Apparently they are right, as the pollsters didn't hear them either.

[dbacksfan 2.0]

I have come to the conclusion that Trump was elected by people who did not feel they were being heard. Apparently they are right, as the pollsters didn't hear them either.

And that's why Hillary lost. She relied on previous election data and believed that the urban centers that voted Democrat would out vote the rural areas that in the past had low turnouts.

Call it what you want, the rural population believe in Trump because he spoke to them and their feelings. Hillary couldn't care less, she believed that she didn't need them.

[heironymouspasparagus]

I have come to the conclusion that Trump was elected by people who did not feel they were being heard. Apparently they are right, as the pollsters didn't hear them either.

THAT is the single biggest factor in this election!  One of the reasons I found the siren call to vote for Trump compelling....that plus the feeling that we should just go ahead and get him in there with all Republican Congress so it will speed up the collapse and at least have a chance of 'enough' left over so we can rebuild.

[cannon_fodder]

Poll all eligible voters.
Ask them when they last voted... ask them if they plan on voting in this election.    It's still probability.

The used to do that. What they found was if you ask someone point blank if they intend to vote they overwhelmingly say yes, in spite of evidence to the contrary. So they use the metric "likely voter," which is often based on voter rolls of people who have actually voted in X% of the last elections.

[erfalf]

Even more irony in this election. I believe Clinton had more votes in the primary of '08 as well. My oh my.