The US Election Polling / Data Thread

[octothorp]

So, as we've had in years past for political events, here's a thread dedicated to polls, data, predictions and related discussion.

This is not the place for discussing candidates' or their policies' merits; except as how those policies might relate to polling. These threads actually have a great record of staying on topic, considering how intense political debate can get.

Here's some helpful links:

Prediction sites:

http://projects.fivethirtyeight.com/...tion-forecast/
http://www.nytimes.com/interactive/2...-forecast.html
http://cookpolitical.com/presidential/charts/scorecard
http://predictwise.com/politics/
http://election.princeton.edu/

Poll aggregators

http://elections.huffingtonpost.com/...ump-vs-clinton
http://www.realclearpolitics.com/epo...lls/elections/

Analysis tools

http://www.270towin.com/

If you've got a favourite website for electoral data, post it and I'll add it to this list.

[GGG]

I really like 538s 3 model way of predicting the election. Through the Conventions the polls plus model has stayed at 60/40 as it accounts for 4% bumps from the conventions. Whereas the Now cast and polls only models have swung wildly.

So I generally look at the 538 polls plus model as an accurate state of the race and probably about a month out I will switch to their polls only forecast.

[Iowa_Flames_Fan]

That is likely what I will do as well, though I like to see what all of the aggregators are doing.

It's a republican-leaning site, but the poll aggregator at realclearpolitics.com is very good, and I very much like the format. I like the graphs on the Huffington post aggregator but the format is less transparent to me.

My one complaint about 538 this year is that it is hard to tell exactly what new data is going into the model day by day. I would like to see a "report" that tells us each time the forecast is updated exactly what the "new" data points are.

[octothorp]

Yeah, I tend to favour 538, but I find it frustrating trying to get new polling data. I understand the logic in ranking polls by their weight in the aggregate, but don't understand why they don't at least have it in a table sortable by date.

[GGG]

The Clinton bounce looks to be in the 5 point range though it's hard to isolate the tailing off of the trump bounce and the rise of the Clinton bounce. I think tomorrow's poll released will be pretty interesting to watch.

[dobbles]

Quick question... In hockey people always like to throw around that idea that the standings by American thanksgiving are often representative of the teams that end up making it.

Are there any polls that are accurate enough to have a deadline like that where there is a relatively high percentage that things are locket in?

[Street Pharmacist]

The android (maybe iOS too?) game Campaign Manager is just awesome if you like this sort of election math. Essentially, you pick a side and invest in various forms of advertising strategically by state, issue, tone, etc. Really neat. Played it in 2012 too

[octothorp]

Originally Posted by dobbles

Quick question... In hockey people always like to throw around that idea that the standings by American thanksgiving are often representative of the teams that end up making it.

Are there any polls that are accurate enough to have a deadline like that where there is a relatively high percentage that things are locket in?

Well, it already really is locked in for about 40+ states or so, and of the remaining states, only half-a-dozen are large enough to really matter except in extremely unlikely scenarios. But that means that predicting the election hinges on polling small, regional areas that can be really difficult to poll.

[Senator Clay Davis]

Post-convention polls finally getting us to numbers we're more likely to see in a general and not both candidates around or under 40%. PPP has Clinton up 50-45 in a straight up poll, while CBS has her ahead 46-39. Neither of these polls include Trump's weekend meltdows so he might even be a few ticks further down. Basically his only hope appears to try and make her even more unlikable than him, but that also assumes he can stay out of his own way long enough for her to actually decline.

[CorsiHockeyLeague]

I guess it's time to bring this back.

[GGG]

3 weeks after the convention polls become much more certain and difficult to move. Though the conventions were a month early this time so that may make a difference.

[Iowa_Flames_Fan]

Yeah the early conventions are a bit of a wild card. My own "date" is labour day--after that, I suspect it gets harder to move the needle. However, with two historically unpopular candidates, who knows?

[Tinordi]

I'm biased but I think we'll see a large bump for Clinton. The DNC was a powerful, moving event while the RNC was a complete circus. Add to that the shenanigans from Trump post-DNC (which the strategic plan should have been to try to move on as quickly as possible from the DNC) and I think he's really opened a space where it's okay for moderate republicans and lean republicans to develop a personal narrative where it's okay to hold their nose this election and patriotically either not vote or vote against Trump.

I just don't see any more upside to Trump. I know that people have been saying that for a year but where does he go to get another 5 points? He has no policies, he has no credibility, all he has is the aggrieved. He has all of them. The next marginal voter is going to be very difficult.

Whereas for Clinton, picking off marginal Trump voters should be a much easier endeavour.

[Tinordi]

Oh and people are really underrating the gelling action that's going to happen now post convention which is what happened with Obama in 2008. The Democrat party will start consolidating firmly now behind Clinton. Trump's actions again, dig his own grave with this regard. He's Hillary's best campaigner to bring the aloof and the scorned Bernie supporters back in the tent.

If Bernie will agree to campaign and be deployed to well to white working class voting districts then he'll be an effective weapon as well.

[octothorp]

Originally Posted by Tinordi

I just don't see any more upside to Trump. I know that people have been saying that for a year but where does he go to get another 5 points? He has no policies, he has no credibility, all he has is the aggrieved. He has all of them. The next marginal voter is going to be very difficult.

In addition to that other 5%, it's also hard to find many maps for Trump to get to 270. It was talked about in the other thread that he almost absolutely needs to win Florida to have any shot.
But the other option is that he puts another large state not currently considered a swing state into play. With a rust-belt blue collar jobs focus, I could see Michigan being the best candidate for the latter, especially given how unexpectedly poorly Clinton did there in the primaries. Clinton is leading in polls there consistently, but not by an insurmountable margin. If he can dominate that region and flip all of Michigan, Ohio, and Pennsylvania red, that opens up a lot of maps for him. I honestly think trying to expand the map is the best play for him. The only other option is to play defence across the whole country with very little margin of error.

[GGG]

He needs Florida, Ohio. He can get there without Pensylvania. I think if he gets Pensylvania he gets Florida as well and it's moot.

So picking up the Nevadas and other close states works with Florida, Ohio.

The other thing that's interesting is that trump underperforms in the mid west relative to his primary polling.

[Iowa_Flames_Fan]

Per Nate Silver, Clinton's "bounce" is a little bigger than Trump's.

http://fivethirtyeight.com/features/...r-than-trumps/

Bounces can be temporary, but my own view (based only on intuition) is that the top end of a bounce might be an indication of a candidate's "ceiling" -- that is, the percentage of voters that in an ideal week might vote a certain way. In that regard, Clinton's ceiling appears a little higher than Trump's, though both are pretty low.

[Senator Clay Davis]

And that's really how I view post convention polls, as an indication of what a candidate's ceiling for votes is. If we get some debates then those will be the most important polls, but that's two months out. I do think as I've seen mentioned the big advantage Hillary has is the Olympics are right after and that will take away a lot of focus, especially with the favorable time zone of Rio. So she can go kinda dark and just let the bigger event take focus away from the race for a while. We'll see what Trump has in mind to try and steal attention.

[Iowa_Flames_Fan]

Originally Posted by GGG

He needs Florida, Ohio. He can get there without Pensylvania. I think if he gets Pensylvania he gets Florida as well and it's moot.

So picking up the Nevadas and other close states works with Florida, Ohio.

The other thing that's interesting is that trump underperforms in the mid west relative to his primary polling.

I think that's right in the sense that if Trump wins Pennsylvania he is likely also polling well enough nationally that Florida and Ohio (and North Carolina) fall into his column.

In a close race, I think we see Trump's demographic disadvantage become a real problem. Nevada and Colorado are tough for him in that regard, and it speaks volumes that Clinton's campaign has stopped airing ads in Colorado, viewing that state as no longer really in play.

[Bunk]

Originally Posted by octothorp

In addition to that other 5%, it's also hard to find many maps for Trump to get to 270. It was talked about in the other thread that he almost absolutely needs to win Florida to have any shot.
But the other option is that he puts another large state not currently considered a swing state into play. With a rust-belt blue collar jobs focus, I could see Michigan being the best candidate for the latter, especially given how unexpectedly poorly Clinton did there in the primaries. Clinton is leading in polls there consistently, but not by an insurmountable margin. If he can dominate that region and flip all of Michigan, Ohio, and Pennsylvania red, that opens up a lot of maps for him. I honestly think trying to expand the map is the best play for him. The only other option is to play defence across the whole country with very little margin of error.

He's really wooing those rust belt swing states

@woodruffbets
Trump on PA and OH: "They call it a rust belt for a reason: because everything's rusting and rotting"