538 releases initial Presidential election forecast

Post Reply New Topic

Headline figures:

Based on polls, which currently give Clinton a 4% lead, Clinton has a 63.5% chance of winning. Trump has a 36.5% chance. The gap narrows by a few points using their polls-plus forecast. In other words, there's a 2 in 3 chance of Clinton winning.

Gary Johnson may get around 8% of the vote, but 0 or 1 electoral votes. No other parties will perform that well.

The closest state is likely to be Florida.

If the election were held today, Clinton's advantage would be compounded. However, she'd get a smaller share of the popular vote. Seems like 538 expect Johnson support to drop away throughout the country and become focused on a few battleground states, with his supporters going to Clinton if anywhere.

Must admit, I'm amazed how well Trump's holding up considering how terrible his policy platforms are, but 538 have a good track record in presidential elections so I'll trust them on this.

538 has also been completely wrong this election season, their formula only works if the polls are working which they clearly are not on both sides of the Atlantic. Nate Silver can't stop himself from editorializing, he can't stop projecting his own aspirations about the campaign. Polling this far out I think it mostly useless, Hillary is not a strong candidate and Democrats have been repeating 'structural advantage' and this idea that Trump can't win without X amount of minority voters more as a mantra than the actual truth since Trump has a clear path the victory given his strength with independents and in strategic battleground states. If Trump wins Ohio, Pennsylvania, Michigan, and Florida then he wins the election simple as that. You can substitute Michigan with Virginia or North Carolina too, multiple paths. Hillary is going to regret promising to put coal miners and the coal industry out of the job big time.

The polls are slightly misleading, because Clinton's appeal is to youth, black and Hispanic, who historically have poor voting habits.

They may pick up a phone and say, "I like Clinton", but will they actually vote?

Nooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo

If there's a good turnout by black and Hispanic voters, Trump has no chance.

This is factored into the model.



(Edited for length so I can get past the Captcha)

538's forecasts were right about Trump and Clinton getting the nominations and they generally called the States correctly. Silver's editorial pieces generally disregarded the models, which he's admitted, but the models were a good guide throughout. These models are largely the same as in 2008 and 2012. There are no "it's Trump!" variables, he's treated exactly the same as if he were Romney or Superman. Same with Clinton.

Clinton might not be as strong as Obama but she doesn't have to be, she just has to be stronger than Trump, and she is, both objectively (in terms of how likely she is to become president) and subjectively (the value call of whether it would be better to have a politician or a racist blowhard buffoon as president).

Clinton is currently forecast to win in all of those states except North Carolina.

Is that what they're saying now? Their 'formula' can't work if the polls are wrong and this election doesn't follow past trends, it's not magical. Remember the polls in your country when it came to the last general election and Brexit? Polling while sometimes accurate is almost always agenda based and is increasingly dependent on dinosaured methods, clearly voters do not stay acting one way or another progressive bloc voting be damned. It's all based on snapshots from sampling and the assumptions made by those administering the poll. There is no 'formula' to politics and you cannot model human behavior, that is the epitome of progressive stupidity. Nate Silver is a loser!

http://i.imgur.com/lWD5XbD.jpg

I seem to remember you saying the same things in 2012...

It's difficult to model the behaviour of an individual, but very possible to do it to a large population. Indeed, the continuing existence of most businesses depends on it.

The Brexit result was well within the margin of error for polls, nobody who was paying attention was surprised by it. We don't have the same sort of data for General Elections that Silver has access to, and our system is much noisier than the Presidential race.

I very much doubt that, Nate Silver was no less a dork then but that was a conventional and very predictable election. I did not support Mitt Romney and never thought he ever had any chance of beating Obama, he was always the designated loser.


This is what I said in response to Nate Silver's projection in September 2012:

viewtopic.php?f=20&t=210763

Also go look at the Brexit polls 3-4 months out if you want to compare

There are three months between now and November's election...and both candidates enjoy solid bases of support, and suffer from trecherous standings. If any of these allegations stick to Clinton to any degree, and she's not through all of them, she'll go from a reasonable favorite to not having a chance in hell. Likewise, Trump's base is fervent and fanatical...but Trump seems to have gotten that base by luck rather then skill; all it takes is one big mistake and he could lose a lot of his zealots...and he's already on the back foot for the election overall.

Projections are all well and good, but they're just something to pass the time with; there is still a lot of stuff in the air.

That was before campaigning had even begun. It would be like asking people who they want to win in 2020, in early 2018.

Ugh no, it would be like asking 3-4 months out. I don't think 3-4 months in American time is somehow 2 years in British time, we live on the same planet right? The general election campaign officially begins after the nominating conventions with the first debate happening at the end of September.

I thought we were on this forum because we *weren't* on the same planet.

In all seriousness, while time is objective, I think saying that polls pre-campaign are going to be different from poll mid-campaign is a relatively logical statement to make. Not years of difference, certainly...but pointing out the difference between political climates is valid.