Live Poll Decypher Blog: Day Before Election (and Election Contest) (updated 4x)

This is a live blog to update any and all polls that come out in the last day before the election in the USA. I will help calculate out the undecided vote so we get to see what the 'final' opinion of that pollster is without the fudging of the undecided vote. I will also average the polls only for those out on the last two days (from Sunday), ie the 'final' polls (the RCP average looks also at older polls).


UDPDATED - 11:30PM EST (Reuters/Ipsos)

NOTE WE NOW HAVE A CONTEST. Free to enter, you can win copy of the book with 1001 jokes about Trump (see below)


In this blog I will give you the three scenarios of how the undecided vote may go, so you can pick which way to interpret the data (and I will average those three). As some pollsters have been more accurate than others this season, I will indicate which pollsters have a bias one way or the other (R is Republican bias, D is Democratic bias, W is wild fluctuations on both extremes). And with that, I will include two sets of the data, one for the 'good pollsters only' and the other with 'all pollsters including bad'. I will update this blog as new polls are released up to Tuesday if any are released then. So the numbers you will see, may change as more polls are added. When I first publish this at 2 PM EST (New York time) ie 14:00 for us rest-of-worlders, we had 9 pollsters out. I will include them in alphabetical order. Note the top set is the 'best pollsters' set. The bottom set is the 'all pollsters set'.

BEST POLLSTERS OPINION

I measured all 166 polls released of the 4-way race this season, against each other. There were some pollsters who systematically had a bias (like notorious Rasmussen, who sneakily grew their bias for Trump from 0% in July to 5% by October in gradual increments). Others had occasional nutty polls like CNN/ORC who once found a 5% lead for Trump when all rival pollsters in the past 10 previous or next polls all had the race for Hillary - etc. But there are pollsters who were never off the consensus view by 3% or more. They, the 'best set' of pollsters this season were:

CBS
CBS/NY Times
CNBC
Economist/YouGov
Fox
Guardian/Survey USA
GWU/Battleground
Pew
PPP

This is the 'gold standard'. These pollsters were never out with a poll that was a clear outlier. Out of this group we've had only 2 three pollsters out with polls yet. But almost certainly we will get more of them before the election. But lets take those two and see how they see the race:

GOLD STANDARD POLLSTER VIEW TO ELECTION 2016

Pollster . . . . . . . . Hillary . . Trump . . Johnson . . Stein . . Undecided . . . Raw Lead
CBS . . . . . . . . . . 45% . . . . 41% . . . 5% . . . . . . 2% . . . . 7% . . . . . . . . . Hillary 4%
Economist/YG . . 45% . . . . 41% . . . 5% . . . . . . 2% . . . . 7% . . . . . . . . . Hillary 4%
Fox . . . . . . . . . . . 48% . . . . 44% . . . 3% . . . . . . 2% . . . . 3% . . . . . . . . . Hillary 4%
AVERAGE . . . . . 46.0% . . . 42.0%. . 4.3%  . . . . 2.0% . . . 5.7% . . . . . . . Hillary 4.0%

UNDECIDED VOTER ALLOCATION MODEL:
Most pro-Trump scenario: 4.2% Race for Hillary
Most pro-Hillary scenario: 9.0% Race for Hillary
Most pro-Others scenario: 5.9% Race for Hillary
Average of 3 scenarios: 6.4% Race for Hillary

Table by TomiAhonen Consulting 7 November 2016, based on raw data at RealClearPolitics
This table may be freely shared

You can allocate the undecideds based on which model you think applies. In any normal year the most reasonable way to allocate undecideds is by proportional allocation (exact percent of what they earned, applied to the last undecided voter). This gives us the most-Trump Scenario. It assumes there is no bias against any voters picking Trump and they are seriously considering the undecided choice between all four remaining candidates.

But as Trump is so toxic, the most disliked politician ever to run, its possible that he has reached a ceiling and has loyal fans, but is unable to convert those who remain undecided. Then its possible that this year we have an unusual allocation of the undecided vote, only among the three other candidates. A kind of 'never Trump' choice between Hillary, Johnson and Stein. If that is the case, we then still have two alternatives. Do they go NOW proportionately among those three (ie Hillary takes the lion's share of the undecideds) - and I call this the most-Hillary scenario; or do the undecideds also strongly dislike Hillary and the undecided vote splits evenly three-ways between Hillary, Johnson and Stein. This I call the most Pro-Others scenario.

You can pick whichever scenario you think is most suitable but note, there are still undecided voters measured by the pollsters, and they HAVE to go somewhere, either pick one of these four candidates or not vote or vote for a write-in candidate, and whatever they do, it impacts the race. So you can pick the model you think is most likely. And its possible some voters are willing to consider Trump, most not; and some who don't want Trump are willing to consider Hillary but others can't stomach her either, and then the average of the three models is a possible outcome too. I have thus included the average of the three as your last scenario option. Note that a poll that says the race is a 4% race, but has 5% undecided - is not a 4% race. Those undecideds have to go somewhere. And thus, pick the sceneario you want.

Now. There have been 27 pollsters who have done a poll of the 4-way race this year. We should get well more than a dozen of them out this year. And to be fair, lets also do the full pollster view, all polls published on Sunday or after, of the race, including those with a bias one way or the other. So here is the first second edition with the first 9 10 11 first 12 polls included (including the two in the above). The 'all pollster' average view is this:

ALL 4-WAY POLLSTER VIEW TO ELECTION 2016

Pollster . . . . . . . . Hillary . . Trump . . Johnson . . Stein . . Undecided . . . Raw Lead
ABC/WaPo (W) . . 47% . . . . 43% . . . 4% . . . . . 2% . . . . 4% . . . . . . . . . Hillary 4%
Bloomberg (R) . . 44% . . . . 41% . . . 4% . . . . . . 2% . . . . 9% . . . . . . . . . Hillary 3%
CBS . . . . . . . . . . 45% . . . . 41% . . . 5% . . . . . . 2% . . . . 7% . . . . . . . . . Hillary 4%
Economist/YG . . 45% . . . . 41% . . . 5% . . . . . . 2% . . . . 7% . . . . . . . . . Hillary 4%
Fox . . . . . . . . . . . 48% . . . . 44% . . . 3% . . . . . . 2% . . . . 3% . . . . . . . . . Hillary 4%
Gravis (R) . . . . . . 47% . . . . 43% . . . 3% . . . . . . 2% . . . . 5% . . . . . . . . . Hillary 4%
IBD/TIPP (R) . . . 41% . . . . . 43% . . . 6% . . . . . . 2% . . . . 8% . . . . . . . . . Trump 2%
Monmouth (D) . . 50% . . . . . 44% . . . 4% . . . . . . 1% . . . . 1% . . . . . . . . . Hillary 6%
NBC/SM (D) . . . . 47% . . . . . 41% . . . 6% . . . . . . 3% . . . . 3% . . . . . . . . . Hillary 6%
NBC/WSJ (D) . . . 44% . . . . . 40%  . . 6% . . . . . . 2% . . . . 8% . . . . . . . . . Hillary 4%
Rasmussen (R) . 45% . . . . . 43% . . . 4% . . . . . . 2% . .  . 6% . . . . . . . . . Hillary 2%
Reuters/Ipsos(D) 42% . . . . . 39% . . . 6% . . . . . . 3%. . . 10% . . . . . . . . . Hillary 3%
AVERAGE . . . . . 45.4% . . . 41.9%. . 4.7%  . . . . 2.0% . . . 6.0% . . . . . . . Hillary 3.5%
Note: (R) means Republican-leaning pollster in 2016 polls; (D) means Democrat-leaning 
pollster and (W) means Wildly swinging to both sides

UNDECIDED VOTER ALLOCATION MODEL:
Most pro-Trump scenario: 3.7% Race for Hillary
Most pro-Hillary scenario: 8.4% Race for Hillary
Most pro-Others scenario: 5.5% Race for Hillary
Average of 3 scenarios: 5.9% Race for Hillary

Table by TomiAhonen Consulting 7 November 2016, based on raw data at RealClearPolitics
This table may be freely shared

The above tables will be updated as more polls come in. Note while the mathematical average of the polls say 3.5% race that is BEFORE the undecided voters are allocated. When undecided voters are allocated these pollster average views suggest a race between 3.7% Hillary victory up to 8.4% Hillary victory and very likely it would be around 5.9% based on these polls.

Now one note of caution. This is the consensus pollster average view of the race as of Monday night. There are five factors very difficult to measure in polling, that are likely in play in 2016. These 5 factors are likely to influence the election BEYOND what those polls say.

FIVE FACTORS NOT CAUGHT BY POLLING (AT ALL, OR ONLY PARTIALLY MEASURED)

Female voter surge. Many of the pollsters have used a standard voter turnout model, of 53% women, 47% men as its been in many election cycles very consistently for more than a decade. If the voter turnout does end up with 53% women, that is fine. There are strong signs, however, out of voter registration data and early voting data, plus anecdotal evidence, that women's vote has surged this year. More women are voting than normal, partly because a woman is on top of the ticket for the first time, and perhaps also driven by the strongly sexist rhetoric and behavior of the rival candidate, Mr Trump. If women turn out more, and the women's vote on Tuesday ends up at 54% or 55% of the vote, that means Hillary (and Jill Stein) are likely to outperform what the polls measured. IF however, a given pollster has used a female/male mix of more women than men, greater than 53/47, that factor might be already measured. I will try to discover this and indicate it.

Hispanic voter surge. Many of the pollsters have had a hard time contacting Hispanic voter or they have deliberately used polling samples to count Hispanic vote at a modest level, similar to the 2012 Hispanic voter proportion of the total vote. There is strong evidence both from voter registrations and from early voter data, and a lot of Hispanic-specific polling, that there is actually a surge in Hispanic vote. If the pollster has not increased their Hispanic voter ratio to reflect this, that would suggest their model under-counts Hillary support. The historic level of HIspanic vote in 2012 was 10%. This year its likely to be 12% to 14%. I will try to discover this and indicate it.

The Democratic party's Get-Out-The-Vote machine was built on a radical bleeding-edge technology type in 2012, by the Obama re-election team to great cost and with great effort. That machine has not been matched by the Republicans and it is now being used by Hillary's team. In 2012 this blog calculated that the GOTV effort advantage by Obama was worth 4% of the final election, and turned a 1% nail-biter election as measured on election-eve polling, into a 5% blowout by Obama on election day. This GOTV effort is likely to play a big part again this year and would favor Hillary; it could favor her GREATLY. I believe in rough terms its advantage this year would yield 4% ABOVE WHAT THE POLLING is indicating. BUT note, there is overlap with the effect of GOTV and the above female voter surge, and the Hispanic voter surge.

There is likely a voter enthusiasm/turnout DECREASE from 2012, in two significant demographics: Youth vote will likely be down, and Black vote will likely be down. These are already fully measured in the polls and thus they should not impact the MEASUREMENT and any error between polling and the final outcome. So these two factors are 'baked in'.

On the Republican side, there is a hope/wish by the Trump supporters of a secret Trump voter demographic, called the 'Shy Trump Supporter' who hides from his Trump support but will vote for Trump. They believe/hope that these Shy Trump voters have lied to pollsters and therefore, there will be a greater Trump turnout than measured in the polls. This would be a variation of the famous Bradley Effect measured in California (which however, went into the opposite direction, they said they were FOR Bradley and then voted against him) but there is some measurement of this phenomenon for the general election suggesting its likely a myth. We will not know until election day, but there is no real evidence this is happening. It may be. It could cause the Trump vote to be under-counted in the polling.

Lastly there is the phenomenon of the Secret Hillary voting Female Republican. There is a clear gender gap measured in polling, and Hillary steals more Republican voters than Trump steals Democrat voters (it happens in all elections but in most elections this is rougly balanced out; 2008 was an exception when many Republicans voted against Sarah Palin on the ticket). But apart from the measured gender gap, there may be secretive Hillary voters hiding among wives and daughters of Republican strong male household heads - husbands and fathers. The sexism and sexual abuse that Trump has brought to the front, has introduced a new rift in the Republican party among its core rural religious and very 'loyal' voter base. The wives of many Republicans are protesting. Some are in households where the husband or father has 'iron rule' and usually gets to boss the family around. And the husband/father may THINK that his wife/daugher will vote for Trump. She may SAY she will but because of the sexism or sexual abuse that Trump has exposed (that often also underlies the Republican party and even some religious leaders) may drive such female voters to secretly take revenge in the ballot box and vote for Hillary instead. Anecdotally there is some evidence of such voters but no measurement of their size, and it would be impossible to measure by polling. Note - its likely that this will not be even measure in the Exit Poll. If there is a significant discrepancy with the Exit Poll result and Hillary final vote being somehow more than Exit Polls said - this is the most likely reason why. It might be a meaningful percent of total vote, could be 1% or 2% that were Republican women, secretly voting for Hillary even as they never admit this to anyone, perhaps fearing a violent reaction by the man heading the household.

The first 3 of these 5 factors are real, they have been measured, and most pollsters do not factor the top two into their polls; the third item - the GOTV effort cannot be measured in polling, it only can be witnessed after it has been deployed on Tuesday. The last two items are far more wishful thinking and with very scant if any evidence and may be far smaller in their total impact to the race. Four of the five factors help Hillary (one was measured to add 4% to the final outcome, in 2012) and only one would work in the opposite direction and the only published estimate of its possible size said it is so slight, it would mean less than 1% effect on the final outcome.

I personally believe that these last 5 factors will contribute so much to the election that it will be a double-digit landslide for Hillary. I am still awaiting all the final polls to come in. It may end up around 12% as the final election result as my guess, when these 'unmeasurable' factors are added to the polling consensus view. I will let you know. I will have my final guess before midnight USA on Monday (EST/NY Time East Coast time) based on what all polls had come in by then.


UPDATE - I have now made my final-final forecast out at 11PM (EST) and yes, I said 12% result. See full final forecast here.



PARTISAN POLLSTER VIEWS

Do we want the partisan pollster views? Of course we do! So lets take those pollsters who saw this race always with red-colored glasses, so far three four pollsters with a Republican-leaning bias have given their final count. Lets take their average:

REPUBLICAN-LEANING 4-WAY POLLSTER VIEW TO ELECTION 2016

Pollster . . . . . . . . Hillary . . Trump . . Johnson . . Stein . . Undecided . . . Raw Lead
Bloomberg (R) . . 44% . . . . 41% . . . 4% . . . . . . 2% . . . . 9% . . . . . . . . . Hillary 3%
Gravis (R) . . . . . . 47% . . . . 43% . . . 3% . . . . . . 2% . . . . 5% . . . . . . . . . Hillary 4%
IBD/TIPP (R) . . . 41% . . . . . 43% . . . 6% . . . . . . 2% . . . . 8% . . . . . . . . . Trump 2%
Rasmussen (R) . 45% . . . . . 43% . . . 4% . . . . . . 2% . .  . 6% . . . . . . . . . Hillary 2%
AVERAGE . . . . . 44.3% . . . 42.5%. . 4.3%  . . . . 2.0% . . . 7.0% . . . . . . . Hillary 1.8%
Note: (R) means Republican-leaning pollster in 2016 polls; (D) means Democrat-leaning 
pollster and (W) means Wildly swinging to both sides

UNDECIDED VOTER ALLOCATION MODEL:
Most pro-Trump scenario: 1.9% Race for Hillary
Most pro-Hillary scenario: 7.6% Race for Hillary
Most pro-Others scenario: 4.1% Race for Hillary
Average of 3 scenarios: 4.5% Race for Hillary

Table by TomiAhonen Consulting 7 November 2016, based on raw data at RealClearPolitics
This table may be freely shared

Note that they find Trump the same as all others, but these three Republican-leaning pollsters find Hillary's support far lower. They then throw those voters into the undecided pile which to this group is far higher. The outcome after undecideds are allocated, by the average of these pro-Republican pollsters, sees the race from anywhere at 2% for Hillary (bigger win than Bush-Gore election of 2000), up to 8% for Hillary (better than Obama-McCain election of 2012). Then lets do the Democratic-leaning pollsters to the 'other side' of the story

DEMOCRAT-LEANING 4-WAY POLLSTER VIEW TO ELECTION 2016

Pollster . . . . . . . . Hillary . . Trump . . Johnson . . Stein . . Undecided . . . Raw Lead
Monmouth (D) . . 50% . . . . . 44% . . . 4% . . . . . . 1% . . . . 1% . . . . . . . . . Hillary 6%
NBC/SM (D) . . . . 47% . . . . . 41% . . . 6% . . . . . . 3% . . . . 3% . . . . . . . . . Hillary 6%
NBC/WSJ (D) . . . 44% . . . . . 40%  . . 6% . . . . . . 2% . . . . 8% . . . . . . . . . Hillary 4%
Reuters/Ipsos(D) 42% . . . . . 39% . . . 6% . . . . . . 3%. . . 10% . . . . . . . . . Hillary 3%
AVERAGE . . . . . 45.8% . . . 41.0%. . 5.5%  . . . . 2.3% . . . 5.5% . . . . . . . Hillary 4.8%
Note: (R) means Republican-leaning pollster in 2016 polls; (D) means Democrat-leaning 
pollster and (W) means Wildly swinging to both sides

UNDECIDED VOTER ALLOCATION MODEL:
Most pro-Trump scenario: 5.0% Race for Hillary
Most pro-Hillary scenario: 9.1% Race for Hillary
Most pro-Others scenario: 6.6% Race for Hillary
Average of 3 scenarios: 6.9% Race for Hillary

Table by TomiAhonen Consulting 7 November 2016, based on raw data at RealClearPolitics
This table may be freely shared

The pro-D pollsters see far less undecided voters and are allocating those who recently made up their minds, now to Hillary but Trump gets no gains vs all pollster average view. The range of outcomes after undecideds are allocated is far more narrow, as a clear Hillary win from from 5% (same as Obama vs Romney of 2012) to 9% (better than Obama vs McCain of 2008) by the average of these pro-D pollsters.

CONTEST - WIN THE JOKE BOOK

So we have a free contest. Guess the final outcome of the election. You are allowed one entry but you may change it up to the moment polls close on Tuesday at 7pm Eastern Standard Time (New York Time) ie when the first polls close. You have to enter via Twitter. To enter, just make your guess - what percent do you think the final election outcome will be. Do this to precision of one percent, so no decimal points. Will the election be say 48% for Hillary and 44% for Trump. PLUS also, enter the Electoral College vote, how many states will they win. the EV vote totals 538. So for example right now the Real Clear Politics average suggests 301EV for Hillary and 237EV for Trump. Send an open Tweet to me - so start the Tweet with @tomiahonen - and then include those two sets of numbers. And include a link to THIS BLOG ARTICLE. So you can cut-and-paste this link: https://t.co/BmuEWfUlva

Thats it. I will have two winning results. Whoever gets the win percentage correctly - wins the ebook with 1001 Trump jokes. And whoever wins the EV vote correctly also wins the ebook. I will allow multiple winners, so don't worry if someone else has picked 'your' number, enter anyway, you both can win. And you don't have to guess BOTH correctly (the EV vote is difficult to figure out if you haven't played with the math) but you can win on either. If you want to play with the states, to see what your EV vote opinion could be, here is the link to RCP page where you can build your own scenario, to see, what if Hillary wins Pennsylvania, and Trump wins Ohio... 

Now, I need you to enter via Twitter, sorry. So while you can discuss it also here in the comments, we won't accept entries from this blog because those on Twitter will not see those entries here from the blog. Sorry. So please do join via Twitter. But send in your guess, what will the result be. The winner gets the ebook. And you have a little over 24 hours to guess. This contest ends when the first polls close in the USA on Tuesday evening.

IF YOU WOULD RATHER LAUGH THAN DO MATH

And finally a plug. I released my joke book. 1001 jokes about Donald Trump. I call it TRUMP vs The Seventh Steve. It runs 253 pages, has actually 1,052 jokes about Trump and 1,434 jokes in total including other, non-Trump jokes like those of other Republicans. If you would rather laugh at Trump than see polling about him, why not buy the ebook, it only costs $4.99. And yes, here is the page for TRUMP vs The Seventh Steve.

Comments

[Wayne Borean]

Heh. My investment in popcorn futures would pay dividends, if I didn't keep on eating all of it.

[Wayne Borean]

Forgot. If you need some chuckles read this article.

https://www.buzzfeed.com/craigsilverman/how-macedonia-became-a-global-hub-for-pro-trump-misinfo?utm_term=.tuRqWnJvE#.puMDnGl8B

[Tester]

Ouch! It just shows how unbelievably stupid Trump's hard core followers really are.

The best bit in that article was that trying the same thing with Bernie Sanders didn't work. Not really surprising in my book, intelligent people do not fall for such blatant scams.

[steve]

HRC 47
DJT 41 %
GJ 5%
JS 2 %

[Tomi T Ahonen]

Hi Steve

Pls send those to me via Twitter, include the link to this blog page, then you're entered into the contest (also if you want, include the EV guess as well - a separate book awarded for that winner)

Tomi Ahonen :-)

[davebarnes]

"you the three scenarios"
Uh, no.
three scenari
the plural

[Wayne Borean]

This is why counting Registered Democrats and Registered Republicans can be really misleading.

http://www.cnn.com/2016/11/07/opinions/navarro-republican-voting-for-clinton/index.html

[Tomi T Ahonen]

Hi Everybody

First off, the count of pollsters is now up to 12, if any of you feel like re-reading the math and analysis.

Secondly, I did my final-final forecast after all. Yeah, down from 16%, I'm going with 12%. I think the 3.5% average of polls has propagandists mixed in. I took them out. Then of the 4% nominal result, I allocated the undecideds. That got me to 7.5%. Then as I said its a 4% gain for Hillary out of the GOTV effort, I am persuaded by the secret women's vote to add the half percent to bring Hillary to 12%. So I'm saying 51% Hillary, 39% Trump, 6% Johnson, 3% Stein and 1% McMullin and others. Texas is her last state won, and she also picks up Georgia, Arizona and North Carolina over what Obama won. Plus she carries all states Obama won in 2012 including Ohio and Iowa where she is behind in in-state polling but the GOTV effort will save her there. Oh, and Trump will lose Utah to McMullin

Tomi Ahonen :-)

[Winter]

Trump is trying to demoralize his voters to the very end:

Optimism From Hillary Clinton and Darkness From Donald Trump at Campaign’s End
http://www.nytimes.com/2016/11/08/us/politics/hillary-clinton-donald-trump.html?_r=0

[Wayne Borean]

I'll stick with my earlier forecast.

[Wayne Borean]

Having fun tweeting about CPAC. This confuses Americans, because the real CPAC is Canadian...

[yallo]

the computers have been hacked