Friday, November 04, 2016

Trump should win Florida

++Addition++Cicatrizatic on why Trump should narrowly win Nevada and North Carolina as well.

---

Through Friday, Republicans held a slim lead in early voting in Florida:


In 2012, Democrats had a 3%-5% advantage (depending on the source) at the same point in the cycle.

So Republicans are up a couple thousand votes now and Democrats were up 50,000 at the same point four years ago. Big deal, there are more outstanding Democrat ballots than Republican ones. Democrats will probably narrowly reclaim the early voting advantage by election day.

That's not why Trump looks like the better bet here. Independents and unaffiliated voters are the reason he does.

In 2012, Obama beat Romney 50%-47% among independents in Florida. Polls show Trump with an 8-15 point lead among independents nationwide. Among the most recent RCP polls out of Florida, Trump leads among independents by 6 points and by 13 points.

If we take the pessimistic end of the range and assume Trump is up on Hillary by 6 points among those without a major party affiliation, say 53%-47%, we go from a Republican advantage of less than 2,000 votes to a Republican advantage of nearly 70,000, with polls in Florida showing Trump with a double-digit lead among those who plan to vote on election day.

22 comments:

IHTG said...

Help make this go viral:

https://www.youtube.com/watch?v=vST61W4bGm8

Mil-Tech Bard said...

AE

Two caviots in this.

1. There is a media report that Hillary is getting a 28% crossover Republican vote in Florida

2. There are also reports of massive new voting by recently naturalized hispanic voters.

The former I doubt, the latter is Democratic fraud SOP, and should be accounted for in the Democratic vote count as stated.

Cicatrizatic said...

Agree. I think Trump is set to win Florida by a 3-4% margin. This morning, Dems have taken a narrow lead of 0.1%, still nowhere near enough to cancel out election day. NBC/WSJ has FL election-day voters at Trump +11.

I am amazed at the number of early vote "analysts" (including many professors) who are making final state predictions based purely on comparison of D ballots and R ballots in the early voting numbers, WITHOUT allocating Independents (and without considering whether one candidate may have a cross-over advantage). I find it remarkable that these self-described experts are incapable of identifying the major variables.

In states like NC and FL, the polls don't show a huge cross-over vote for any candidate, but Trump does consistently get somewhat more Ds than Hillary gets Rs, which is especially significant in NC since there are more Ds.

If you compare NC and FL early voting to 2012, then allocate the cross-over vote and Independents according to the latest polls, it looks like Trump is set to win each state. Of course, there are always variables. Dems could have unexpectedly high turnout on election day, or the polling cross-tabs on Independents could be wrong. But based on the weight of evidence, FL and NC look good.

I think it's going to come down to NV, CO, PA, NH, and MI. NV and one other state can do it. NH is already red, and PA and MI are narrowing. CO is within the margin of error. I think VA and WI will be within a few points (2-4%), but still blue.


Audacious Epigone said...

IHTG,

Soros and Yellen back to back, with narration that could've almost been out of Bernie Sanders' mouth during the primaries. That has to be running in states like PA and MI.

Mil-Tech Bard,

Good point. That would be the only plausible explanation that meshes with the polling data, but it'd require either the partisan polling in the state to be incorrect (they're showing Trump's Dem support slightly better than Hillary's Rep support), or a much higher likelihood of early Republican voters to be crossover voters. My naive assumption would be that party crossovers would wait until election day because their minds could still be changed but maybe not.

Cicatrizatic,

That is inexplicable given that 20% of the early votes so far are from independents/unaffiliated. When the margin is razor thin, discounting one-fifth of the vote is ridiculous.

Of course you're right about the polls simply being incorrect. My instinct was to title the post "Trump will win Florida", but I hedged because of exactly what you point out.

Cicatrizatic said...

The media is declaring victory for Hillary in Nevada based on early vote numbers. As usual, they are ignoring the above-mentioned variables.

I just ran the numbers. 767,000 Nevadans have voted so far. Among the ballots returned: 42% Democrat, 37% Republican, 21% Independent.

To allocate the votes to specific candidates, I averaged the cross-tabs on Democrats, Republicans, and Independents from the last 3 Nevada polls. I did not include the previous poll before that (Emerson) as it was taken 10 days ago. The last 3 polls are all within the last week.

Based on the last 3 Nevada polls, the cross-tabs are:

Democrats: 82% Clinton, 14% Trump
Republicans: 85% Trump, 9% Clinton
Independents: 46% Trump, 34% Clinton

As you can see, Trump has a significant cross-over advantage and handily wins Independents.

Applying these cross-tabs to the early vote numbers, you get:

Trump - 360,413 votes
Clinton - 344,459 votes

That puts Trump at a 2% lead.

Obviously this is all contingent on the poll cross-tabs being accurate. But based on the available data, Trump is likely narrowly winning Nevada.

You can find no shortage of articles today saying "It's over in Nevada, Hillary has won."

In 2012, Obama won the Nevada cross-over vote - he got 7% of Republicans, while Romney only got 4% of Democrats. Romney only won Independents by 7%, whereas Trump looks to win them by double-digits.


Random Dude on the Internet said...

http://www.thedenverchannel.com/news/local-news/early-voting-lead-for-colorado-democrats-dwindles-as-republican-voting-surges
http://projects.fivethirtyeight.com/2016-election-forecast/nevada/

Lots of positive news about Nevada and Colorado. I'd say these states are likely going for Trump. If Nevada and Colorado are Trump victories, New Mexico may not be far behind. If he is doing well in those states, he likely does well with voters in places like Wisconsin, New Hampshire, and the white voters in Michigan. If he wins those states, Pennsylvania should be in the bag and although it's a stretch, maybe Virginia as well.

Audacious Epigone said...

Random Dude,

That's the path for a Trumpslide. 3% of the popular vote and at least a 50 EV margin qualifies as such, I'd say.

Cicatrizatic said...

I went ahead and crunched the numbers for North Carolina.

As of the end of yesterday, 2,892,088 people have voted in NC: 42% Democrats, 32% Republicans, and 26% Independents. As with FL, Democrats are far behind their 2012 pace. In 2012 NC early voting, Democrats were 48% to Republicans at 31%.

Here is an average of the cross-tabs for the last 3 NC polls (all taken within the last 7-10 days, after the FBI news):

Democrats: 80% Clinton, 17% Trump
Republicans: 86% Trump, 8% Clinton
Independents: 46% Trump, 37% Clinton

Huge cross-over vote for Trump, and he wins Independents by 9, which is consistent with his national average margin with Independents.

When applied to early votes you get:

Trump: 1,363,300 votes
Clinton: 1,323,997 votes

That would mean Trump actually leads early voting by 1.4%.

As a caveat, I don't really think Trump will get 17% of Democrats. It will probably more like 10-12%. But still, it's clear from the NC polls that he will win the cross-over battle.

NBC/WSJ has NC election-day voters at Trump +2, which, when combined with the above early voting numbers, Trump would ultimately win by 1.5%; however, that poll was taken two weeks ago, at Trump's nadir, and the top-line of that poll (Clinton +6) was one of Trump's worst.

To get a better indicator of election-day performance, we can use the 2012 exit polls for a Party ID breakdown, and apply the above cross-tabs. That is likely too favorable to Clinton, as Democrats have declined in both party registration and Party ID in NC in the past 4 years, but we'll use it as a conservative estimate.

It is estimated that about 1.5 million votes will be logged on election day in NC. In 2012, NC exit polls showed a Party ID breakdown of 39% Democrats, 33% Republicans, 28% Independents.

Applying the above cross-tabs to that sample, you get:

Trump - 726,750 votes
Clinton - 663,000 votes

Combined with the early votes, that would give Trump an overall 2.3% margin of victory, almost identical to Romney.

Obviously, a lot of caveats. The polling cross-tabs may overstate Trump's cross-over vote and his lead with Independents. Then again, I have used a 2012 Party ID to estimate election day voting sampling. In light of the collapse of Democratic turnout (relative to Rs and Is) in early voting, that's a relatively generous assumption for Clinton's purposes.

I think Trump's chances in NC are good.

Cicatrizatic said...

For comparison, in 2012, 2.8 million early votes were logged in NC. 48% from Ds, 31% from Rs, 21% from Is.

When you apply the 2012 exit poll cross-tabs to the 2012 NC early vote, Obama probably led Romney by about 229,000 votes at the conclusion of the early voting process.

On election day, Romney (based on the estimates) beat Obama by 321,000 votes, just to give you a sample of how lop-sided actual election day is in favor of Republicans.

Cicatrizatic said...

In FL, as of this morning Democrats have taken a 0.5% lead in combined early votes and absentee ballots. Still 2.5% off of their final lead from 2012.

For some FL counties, yesterday was the last day of early voting. Some FL counties still have early voting today.

In any event, the polls do not reflect any appreciable cross-over vote in FL for either candidate, and Trump wins Independents by an average of 7-12 points in each FL poll. That is a big difference from 2012, where Obama won FL Independents by 3. Given that cross-tab, Trump is likely already leading the number of votes and is set to increase his lead significantly on election day. Still looks like he'll win Florida by about 4%.

Audacious Epigone said...

RCP's no toss up map has Trump a Florida flip away from winning, 270-268, but Florida is in the Clinton column now. The markets still have it 60%-40% in Hillary's favor.

Cicatrizatic said...

YouGov poll of Florida voters who plan to vote on election-day: Trump +16

The Z Blog said...

Everyone keeps forgetting that Maine splits its two electors by congressional district. Trump will probably win ME 2. That could very well be the difference.

That said, I suspect if he does win, it is going to be by a lot. We'll see states like PA and MI go Trump.

Cicatrizatic said...

Axiom Strategies finally battleground polls:

CO: Clinton +1
Florida: Trump +3
Nevada: Trump +1
North Carolina: Trump +3
Ohio: Trump +1
Pennsylvania: Clinton +1
Virginia: Clinton +2
Wisconsin: Clinton +8

CO, PA, and VA all within the MOE.

Audacious Epigone said...

Yikes, but so is Ohio. With Kasich's cuckery there, that's worrisome.

Audacious Epigone said...

Z,

I was thinking today that anything from Trump +5 to Clinton +5--across that 10-point range--in the popular vote wouldn't really feel surprising, nor would a 100 EV MOE in either direction.

Anonymous said...

It's 297-241 for Clinton with Florida to Clinton (polled at +1%).
Florida is decisive (Florida to Trump -> Clinton 268 Trump 270)

Cicatrizatic said...

Final numbers in from Florida: with the Democratic surge from yesterday, their final lead is 1.35% in the combined early vote + absentee ballots. Most Republican counties were closed yesterday, and most Democratic counties open.

Still, with Independents breaking +7 for Trump for Quinnipiac and CNN, the two candidates are likely currently tied in votes, and Trump will easily overtake her on election day. YouGov has Trump at +16 with election-day voters. There will probably be about 2-2.5 million votes on election-day in Florida.

Still looks like Trump by 4 in FL.

The Z Blog said...

The Colorado numbers are one reason to be optimistic right now. As Steve Sailer has pointed out, high Hispanic numbers can push the white vote in the GOP column. Texas is the most obvious example. Nevada is another. Colorado has had a lot of immigration. It is too close to call according to the polls. If Trump wins Colorado, he wins the election.

I'm always cautiously pessimistic, but if I'm forced to bet, I'd wager on Trump winning tomorrow. I suspect many of those people saying they plan to vote Goofy Weed Man or Based Yenta will break Trump in the privacy of the polling booth.

The Z Blog said...

I see New Mexico is now moving from Clinton to toss-up. Maybe it means nothing, but there are no examples of states moving the opposite direction.

Audacious Epigone said...

Still frustrating to see Florida at 2-1 Dems' favor in the markets.

Where is the 20% Rep-for-Hillary coming from? The last Quinnipiac poll from Florida showed early voters going 48-40 for Hillary at at time when more GOP ballots had been cast than Dem ballots in Florida, even though Reps and Dems were equally likely to vote Trump and Hillary, respectively, in the same poll.

Either the early voters are slanted Hillary across party lines, or the polling is off and Trump is going to outperform the state's polls significantly.

Cicatrizatic said...

Two PA polls today with Trump ahead, only one of them is included in the RCP average. The PA polling average has been a steady descent down into the margin of error.

There is one MI poll today with Trump winning by 2. Not sure if I believe it. There simply has not been much quality polling in Michigan. After the polling debacle in the Michigan Democratic primary (a polling miss of 23 points), a lot of pollsters probably decided to sit it out.

Based on the fact that Obama made a campaign appearance in Detroit today, I'm betting the margin in Michigan will be 3 points or less.