Wednesday, October 31, 2012

November 1, 2012 Analysis (posted early)

I'm sure with RCP moving into a tie and the National Journal poll being released, people are nervous about what is going on.  I figured I would post early and hopefully allay some fears.

Polls included:

Tracking Polls
R+2 - Rasmussen Daily Track
R+5 - Gallup
O+1.3 - IBD/Tipp
Even - ABC/WaPo

Others
O+1 - Battleground
Even - Pew
R+1 - NPR
O+1 - CBS/NYT
Even - Fox
O+5 - National Journal

O+0.03% - Current RCP Average
O+0.56% - Average using the 2008 turnout model
R+2.23% - Average using the D+3 turnout model
R+4.46% - Average using the 2010 turnout model
R+4.51% - Average using the 2004 turnout model
R+6.24% - Average using the Rasmussen Party ID turnout model

With the bizarre National Journal poll entering the averages, we see the averages again move slightly toward Obama.  Romney has just about given back all of his gains from a week ago, bringing the race back to the steady state we were at for the prior two weeks.

While it would be easy to blame the National Journal poll for bringing down the numbers (and to be clear, in this run of the models, that poll is the sole cause for the drop), over the last 3 days overall averages have dropped.  I was expecting Rasmussen to move back to 50-47 today, as the Saturday sample dropped out, but that did not happen.  Also, ABC/WaPo showed a one point move toward Obama, without changing their sample.  We may well be seeing a slight movement in Obama's direction due to the Hurricane Sandy response.

Having said that, Romney is still far ahead in this race.  Obama needs about a D+5 or D+6 turnout to make this a close race.  There are certainly plenty of polls out there that are predicting that exact result, but all of the early vote results, partisan enthusiasm measurements, and Independent support (which is still averaging 8 points, unchanged over the last 4 days) do not make such a result very likely.

With Gallup and other polls on hiatus due to the storm, we will need to wait to see if other tracking polls confirm this movement toward Obama, or if they remain stable. 

National Journal Poll - October 31, 2012


Likely Voter, 713 sample, Obama leads 50-45, 5% undecided.

I really hate these guys.  They post an article about a poll, provide no internals, then RCP includes them in their average.  I don't feel right leaving this poll out of the average, this late in the election, but they are giving me very little work with.  Consider the following quote from the article (bolding added by me):

In its likely-voter model, the Congressional Connection Poll projected that the 2012 electorate will be virtually unchanged from 2008, with Democrats holding an 8 percentage-point advantage among voters (compared with 7 points last time) and whites representing 73 percent of voters (compared to 74 percent last time).

Other recent polls, such as the Pew Research Center survey released on Monday showing a tied race, have found a narrower, or nonexistent, Democratic identification advantage for 2012. Romney’s prospects, obviously, will increase the more the partisan gap declines. The biggest message from the survey is that even small changes in the electorate’s composition next week could have huge ramifications in a campaign that is dividing the nation so closely.
Way to hedge your bets there, guys.

With this set of clues, I am going on the assumption that they are still very close to their last D/R/I of 36/29/30.  I am using 37/29/30.  I am also assuming that Independents still favor Romney by 8 points, like they did a month ago.  It's probably higher since the last poll was taken before the first debate.

Which is an important point. This last poll was released on October 2nd, the day before the first debate.  We are supposed to believe that there has been a 5 point move toward Obama since before Denver?  How does that make any sense what so ever?

The turnout models give the following results:

O+5 - Current result
O+3.0 - 2008 turnout
R+0.1 - D+3 turnout
R+2.4 - 2010 turnout
R+2.5 - 2004 turnout
R+4.3 - Rasmussen Party ID

Obviously, this poll shifted strongly toward Obama by about 3 points in all models.  If turnout is D+3 then this poll is showing a tie.  In all other models, Romney leads comfortably.

Fox News Poll - October 31, 2012

Likely Voter, 1128 sample size, Tied 46-46, 6% undecided.

This poll was last run on October 10th.  Despite the headline that "Romney's lead has vanished and it is all tied up", that is not true.  The last time this roll was run it had a D/R/I of 40/39/20, or D+1 with Independents favoring Romney by 16.  This time the D/R/I has moved to D+5, 42/37/18.  However, Independent preference for Romney has dropped to 7 points.

Putting these numbers through the models, we get the following:

Even - Current result
R+1.4 - 2008 turnout
R+5.8 - 2010 turnout
R+5.8 - 2004 turnout
R+7.4 - Rasmussen Party ID

Across EVERY model, Romney's support has increased by 2 points in this poll.  This shows very strong movement toward Romney in the last two weeks.

Senate Update - Panic in the Disco edition

I hate when Fox does those stupid "look at the Senate" pieces during the panel, because they always end up with me dealing with nervous people thinking we are on the verge of losing the Senate.  I actually like Kevin Dujan's term for what Fox is doing, "Fear Porn".

Anyway, not much changed in the Senate polling since last time, but I'll make some pessimistic assumptions just to address everyone's fears.

Here is my opinion on all 16 of the current Senate races I am following.
  • Nebraska 85% -  Charles Lane tried to argue that Fisher is dropping back in this race.  Yes there is a new poll showing Fisher only up by 3 at 49-46.  But the same poll says that Kerry wins Independents by 24 points.  Given Romney's lead in Nebraska, this is just not credible.  But I'll drop this to 85% anyway, just to appease the nervous folks.
  • Indiana 50% - Today the panic is that Mourdock says his internal polls put him in a tie 44-44, and other polls show Donnelly leading.  I think all this discussion of ticket splitting is crap, myself.  Everyone knows that if you don't elect Republican Senators, then Obamacare remains law forever.  But I'll go ahead and drop this down to 50/50.
  • Nevada 100% - Poll released yesterday has Heller up 6.  I still think this race is over.
  • Wisconsin 80% - Rasmussen is still showing Thompson with a 1 point lead, and is polling better than Romney.  As hard as Romney is pushing to win this state, I don't see Thompson losing.  Increasing these odds to 80%
  • Montana 60% - The polling still hasn't changed.  Last night someone on the Fox panel tried to argue that a 3rd party candidate will stop the GOP.  Not this year.  Tester is an incumbent and stuck at 48%.  Undecideds will break toward Rehberg, and Romney will have coattails.
  • Massachusetts 50% - I'm moving this up to 50/50.  The polling still favors Warren, but the final debate was cancelled.  Warren wanted to reschedule, but Brown said no.  Losing campaigns are the ones that want to reschedule a debate.
  • Virginia 75% - Latest poll is Quinnipiac with Kaine leading by 4.  In a D+7 poll.  If Virginia turns out D+7, it doesn't matter who controls the Senate, because Obama will be President.
  • Florida 25% - The polling continues to be bad for Mack.  He is losing to Nelson by 13 in a D+7 poll.  While Nelson is still under 50%, I'm not seeing the movement needed to overtake him.  I've been watching the Mack ads for months, and he is really a poor candidate.
  • Ohio 60% - Pulling this number back a bit.  Mandel is under performing Romney in Ohio.  If Romney wins by the 4-5% I think he is likely to, then Mandel will also win.  If it is 2%, then Mandel probably loses barely.  I do think Mandel's polling is also being affected by the "Lying Democrat" issue.
  • New Mexico 10% - I don't think anyone is bother to even poll here any longer.  Pretty sure the Dems will win this race.
  • North Dakota 85% - Berg is up 2 points in a recent Mason Dixon poll, but Romney will win by double digits.  I'll pull this one back to 85% just to appease the nervous folks.
  • Missouri 50% - Mason Dixon giveth, Mason Dixon taketh away.  Latest poll puts Akin within 2, and McCaskill stuck at 45%.  They REALLY don't like her in Missouri.  I'm moving this up to 50/50.  I don't see how Akin loses this right now.
  • Michigan 10% -  No one is even polling this race.
  • Maine 10% - The Independent will win this race.  He could decide to caucus with the GOP, especially if it means being in the majority.
  • Pennsylvania 40% - Casey's top line number is stuck in the mid to high 40s.  Some polls are showing Smith getting very close.  This would be my upset special.
  • Connecticut 25% - Rasmussen showing Murphy back out to a 6 point lead.  This race will depend on how much the people of Connecticut want to save their insurance industry.
So again, this is the pessimistic version of the odds.

Odds of at least a tie in Senate = 90.8%
Odds of winning the Senate = 79.5%

October 31, 2012 Ohio Update

Two new polls come into the RCP average from Survey USA and Quinnipiac.  The Survey USA poll showing Obama leading by 3 is unchanged from last week.  The D/R/I remains 38/32/26 with Independents favoring Romney by 11 now (up from 9 last week).  Quinnipiac shows Obama leading by 5 using a D/R/I of 37/29/30 and Independents favoring Romney by 6.  This poll in particular suffers from the Lying Democrat problem of the other Ohio polls.  It has too many people claiming they have already voted, and that they favor Obama by 2:1.  I discussed how absurd this is yesterday.

Survey USA:
O+3 - Current result
R+7.49% - 2004 turnout model
O+2.12% - 2008 turnout model
R+4.23% - 2010 turnout model
R+3.48% - 2012 registration model
R+1.31% - D+3 turnout model

Quinnipiac:
O+5 - Current result
R+6.76% - 2004 turnout model
O+2.50% - 2008 turnout model
R+3.25% - 2010 turnout model
R+2.50% - 2012 registration model
R+0.07% - D+3 turnout model

The Time, Suffolk, and Cincinnati Enquirer polls were dropped from the average.  Note that two of those polls showed Romney tied with Obama.  I still can't figure out RCP, since the C.E. poll in particular was more recent than the CNN poll that remained in the average.  Looking at the new averages across 8 polls:

Polls included:
O+3 - Survey USA
R+2 - Rasmussen
O+4 - PPP
O+5 -Quinnipiac
O+2 - Purple Strategies
O+4 - CNN/ORG
O+2 - ARG
O+1 - Gravis

O+2.38% - Current RCP Average
R+7.45% - Average using the 2004 turnout model
O+2.16% - Average using the 2008 turnout model
R+4.19% - Average using the 2010 turnout model
R+3.46% - Average using the 2012 registration model
R+1.25% - Average using the D+3 turnout model

Again, slight movement toward Obama across all the models, which is unsurprising given the change of 5 polls in the RCP average.  If Democrats are able to get a D+3 turnout in Ohio, they will lose, but it will be close.  Anything under D+3 and Romney will declared the winner in Ohio early.

I've said this before, but let me reiterate.  Either the Ohio polls or the National polls are wrong.  If Obama wins Ohio then he will also win Nationally and we will be all talking about how Gallup and Rasmussen blew their polling this year.  However, if the National polls are correct, then the Ohio polls are wrong and we will be discussing why.

I am confident that the later will be the case.  The reason why will be identified as the over sampling of Democrats claiming to have early voted for Obama, who were lying about it.  The way to correct this is the pollsters can not include an unweighted sample of early voters, they must weight their sample to the actual number of early votes received.  Otherwise, campaigns will continue to game the system, like the Obama campaign is doing this year.

October 31, 2012 Analysis

Polls included:

Tracking Polls
R+2 - Rasmussen Daily Track
R+5 - Gallup
O+1.3 - IBD/Tipp
R+1 - ABC/WaPo

Others
O+1 - Battleground
Even - Pew
R+1 - NPR
O+1 - CBS/NYT

R+0.71% - Current RCP Average
O+0.42% - Average using the 2008 turnout model
R+2.37% - Average using the D+3 turnout model
R+4.63% - Average using the 2010 turnout model
R+4.67% - Average using the 2004 turnout model
R+6.43% - Average using the Rasmussen Party ID turnout model

RCP dropped brought the CBS/NYT poll back into the average.  Rasmussen and Gallup did not change yesterday, while IBD/Tipp suspended polling due to the hurricane.  ABC/WaPo updated their tracking poll, showing a 2 point move toward Romney, but this was accomplished through a sample shift from D+7 to D+5, so the averages in the models did not change due to this particular poll.  The average support from Independents remains Romney +8.

Overall, Romney gave back some of the ground he has gained over the last few days due to the CBS/NYT poll pulling the averages down.  We will see what happens with Rasmussen today as the Saturday sample drops out of his tracking poll.

Tuesday, October 30, 2012

CBS/NYT Poll - October 30, 2012

Likely Voter, 563 sample size, Obama leads 48-47, 3% undecided.

Let's get the obvious out of the way first.  This is a horrible poll.  A sample size of 563 is bordering on malpractice, and if RCP had any respect they would reject this poll and not include it.  They are claiming to be able to measure the preferences of the Republicans across the country from the responses of 175 people.

Compared to the last time this poll was run, this poll is excellent for Romney.  Previously this poll had Obama leading by 2 with a sample that was dead even with Democrats and Republicans both at 32% and Independents favoring Obama by 6.  While the top line poll results didn't change much, the sample did.  It is now D+5 with a D/R/I of 36/31/33.  Additionally, Independents now favor Romney by 12 points, an 18 point swing in one week.  All of this data points to the previous poll being an outlier, since it did not match any of the other polls at the time.

Putting the results into the reweighting models we get the following:
O+1 - Current result
O+1.8 - 2008 turnout model
R+1.1 - D+3 turnout model
R+3.5 - 2010 turnout model
R+3.5 - 2004 turnout model
R+5.3 - Rasmussen Party ID

This poll is in alignment with the results of the other polls, but trends towards Obama compared to the others.  Bottom line, this is about as good as it gets for Obama.

Some trends over the last few days

Let's take another look at the poll model trend lines over the last few days.  I've shown this graph a couple times so far, and I want to point out the last 6 days.






On October 25th, the models broke from the steady state they have been holding for about two weeks.  This occurred right after the first of the new emails broke that the Administration knew Benghazi was a terror attack within the first 2 hours of the incident.

Since then, we are seeing a slow increase in support for Romney across all of the models.  They have all moved about 0.6 of point in the last 6 days.  During the same period we have seen significant erosion of Obama job approval rating.

Note that the two things that are not tracking this movement are the RCP average itself and Rasmussen's daily track.  In the case of RCP, the new polls moving into the average have offset the Romney gains by increasing their Democrat sample.  Rasmussen has a similar problem, but that is caused by his typical sampling issue over weekends.




What is really going on with early voting?

I noted a problem yesterday with the Rasmussen Ohio poll that I think needs further discussion.  As you recall, this Ohio poll showed that Romney is leading by 2 in Ohio, 50-48.  However, there is a big head scratcher in that poll regarding early voting.  33% of the sample claim to have already voted in Ohio, and of those voting early, they favor Obama by about 2:1. 

This situation is replicated in many of the Ohio polls.  The most recent PPP poll shows early voters at 36% again with a 2:1 preference for Obama.  I have been noting this problem all month, with Ohio polls from ARG and Survey USA also showing very high early vote totals heavily favoring Obama.

But is it true?

Let's quickly do the math.  Ohio has about 7.9 million registered voters right now.  For these polls to be correct then they are reporting the following:

  • Rasmussen - 33% - 2,500,000 already voted
  • PPP - 36% - 2,800,000 already voted

The Election Project at George Mason University is maintaining a count of the number of early votes actually received by each of the states.   Take a look at their Ohio totals as of 10/26:




So far slightly over 1 million early votes have been received in Ohio, and 161,000 from Cuyahoga county.  Cuyahoga is, of course, the major Democrat bastion and the intended source for most of the Democrat votes during the election.  In 2008 Cuyahoga county had 266,000 early votes, so as of 10/26 the Democrats have only booked 60% of their 2008 total from this county.  Overall, the GOP analysis offered is that Democrats are 8% off their early voting pace, while Republicans are 5% ahead of their 2008 pace.

In any case, in Ohio, the total number of actual early votes is only 12% of the registered voters, or 17% of the 5.7 million actual voters in 2008.  Not 33% or 36%.

The result of this discrepancy in the polls is to drive up the likely voters in the sample, and to over sample Democrats, since the 2:1 preference ratio favors Obama to such a high degree.  Regardless of whether these voters who are lying about early voting are really likely voters, it still represents a significant over sample of Obama support, given the results in the rest of the poll.

Gallup released a survey yesterday that provides a better look at the state of early voting nationwide.  This survey is gathered from their likely voter pool that they are using in their tracking poll that is showing Romney leading 51-46.  This survey fits the results we saw from the Ohio early voting returns.  15% of the respondents have already voted, and 18% plan to vote before election day.


Or do they?  There is a very curious item if you look at those who have voted and those who plan to vote.  Those who have actually voted favor Romney 52-46, closely matching the overall poll results.  Those who plan to vote on election day also favor Romney by almost the same amount, 51-45.  But those who say they will vote early, but who haven't yet voted give equal support 49-49.


To me, this is just a continuation of the lying Democrat phenomena.  They say they have/will voted early for Obama, but they actually haven't/won't.  This behavior is what is driving the Ohio polls to overstate Obama's support.  Support that will not result in actual votes on election day.

Overall, we are seeing the results of an excellent strategy by the GOP.  Their plan all along was to neutralize the Obama early voting advantage, and turnout their base on Election Day.  In 2008, Obama drove up huge early voting totals in many states like Florida and North Carolina, allowing them to take those states, even though McCain received more votes than he did on Election Day (in those states).  Neutralizing this strategy has been the goal of the Romney campaign, and especially the RNC for the last year.  This is why so much effort has been put into voter contact.  Look at one final graph from the Gallup article.


Between already voted and plan to vote (including those lying Democrats) the totals are running even.  Romney is slightly ahead of Obama 34-33.  This neutralizes the Obama ground operation and primary strategy.  On election day, the remaining 2/3rd of the electorate will vote, and they will favor Romney 51-45.

This is how you win the ground game.

October 30, 2012 Analysis

Polls included:

Tracking Polls
R+2 - Rasmussen Daily Track
R+5 - Gallup
O+1.3 - IBD/Tipp
Even - ABC/WaPo

Others
O+1 - Battleground
Even - Pew
R+1 - NPR

R+0.81% - Current RCP Average
O+0.13% - Average using the 2008 turnout model
R+2.63% - Average using the D+3 turnout model
R+4.86% - Average using the 2010 turnout model
R+4.91% - Average using the 2004 turnout model
R+6.65% - Average using the Rasmussen Party ID turnout model

RCP dropped the NBC/WSJ, Monmouth, CBS, and AP/GfK polls over night, and included the Pew and NPR polls.  Despite all of this movement, the overall averages didn't change too much.

Rasmussen dropped to R+2 as he typically does with the weekend polling.  He will start to move back out to R+4 or R+5 as the Sat/Sun samples drop out of his poll.

Gallup increased back to R+5 and is looking to be on track for winning the award for most accurate when everything is done.  I'm pretty surprised by this, and am gaining new respect for their polling.  I just wish I could see their internals.

IBD/Tipp is not polling because of Hurricane Sandy, so no change.

ABC/WaPo is acting weird.  They keep increasing their partisan sample to make it more Democrat.  In the last 3 or 4 days they have moved from D+4 to D+7.  Despite finding Independents preferring Romney by 10 points, they show a tie.

Independents continue to favor Romney by an average of 8% across all 7 polls.

The important thing is that this is the 3rd day in a row that support for Romney has increased across all models, this time 0.1%.  Note that again, my models are all showing movement toward Romney, while the RCP remains flat.  I plan to post about this trend line later today, but there is clear movement in all polls toward Romney in the last 3 days.

  NPR Poll - October 30, 2012


Likely Voter,  1000 sample size, Romney leads 48-47, 5% undecided.

A new poll moved into the RCP average.  NPR's poll uses a D/R/I of 37/31/31 or D+6 with Independents favoring Romney by 12 points, 51-39 including leaners.  If you look at strong support among Independents, Romney has even better support with 42% of Independents saying they will definitely vote for Romney and 29% definitely voting for Obama.  Unlike a lot of recent polls, this one gets the early voting sample correct.  They report 15% of the voters have already voted, which matches what Gallup found.  Among those early voters, 14% of the Republicans, 15% of the Independents, and 16% of the Democrats have already voted.  This is very good for the GOP since the Democrat strategy is to dominate early voting to make up for losing on Election Day.

Putting these results through the models, we get the following:

R+1 - Current result
R+2.0 - 2008 turnout
R+4.9 - D+3 turnout
R+7.2 - 2010 turnout
R+7.3 - 2004 turnout
R+9.1 - Rasmussen Party ID

This is a very good poll for Romney, showing him with a healthy lead in all turnout models.  Even a 2008 turnout would not save Obama.

Monday, October 29, 2012

Pew Poll - October 29, 2012


Likely Voter, 1495 sample size, Tied 47-47, 6% undecided.

October 8th was the last time the Pew poll was published.  At that time they were showing a Romney lead of 4, which has now become a tie.  Before you get worried about Obama suddenly gaining support, the D/R/I of that 10/8 poll was 33.4/36.3/30.3 or R+3.  The D/R/I of this poll is 35/34/30 or D+1.  Their sample shifted 4 point toward the Democrats and their results also shifted 4 points toward Obama.  Independents favor Romney by 8 points.

Putting these results through the models, we get the following:

Even - Current result
O+2.4 - 2008 turnout
R+0.4 - D+3 turnout
R+2.7 - 2010 turnout
R+2.7 - 2004 turnout
R+4.6 - Rasmussen Party ID

This poll is in line with the Battleground poll.  Just like that poll, this poll seems off.  An 8 point lead among 30% of the respondents should translate into a 2.4% lead, not a tie.  Note that my 2010 and 2004 turnout models produce almost exactly that expected result, 2.7%.

October 29, 2012 Ohio Update

Scott Rasmussen, come on down!  You are the big winner!

I asked yesterday who would be the first to show an Ohio lead, and it was Rasmussen.  His new Ohio poll is conducted completely on a Sunday (traditionally a bad day for polling Republicans) and it finally breaks through the wall, and shows Romney ahead 50-48.  The D/R/I is 38/38/24 which at an Even split is a little surprising.  Independents favor Romney by 4.  Even so, we still get a silly result where 33% of voters claim to have already voted.  As of 10/27 1,006,398 ballots have been cast, or 17.4% of the registration numbers.  For 33% to be correct, 2 million votes would need to have already been cast.  Since these early vote samples are clearly wrong, their 2 to 1 preference for Obama has to be questioned.

We also have new polls from PPP (O+4) and Gravis (O+1) that are horribly over sampled with Democrats, D+8 in both cases.  The PPP D/R/I is 43/35/21, while the Gravis D/R/I is 40/32/28.  Independents in the PPP poll favors Romney by 4, while Gravis shows preference to be 10.  I'm going to give the model results for all three polls.  You will see that Rasmussen and PPP are actually very similar, while Gravis is a strong poll for Romney.

Rasmussen:
R+2 - Current result
R+6.40% - 2004 turnout model
O+2.30% - 2008 turnout model
R+3.44% - 2010 turnout model
R+2.72% - 2012 registration model
R+0.81% - D+3 turnout model

PPP:
O+4 - Current result
R+6.36% - 2004 turnout model
O+2.47% - 2008 turnout model
R+3.32% - 2010 turnout model
R+2.70% - 2012 registration model
R+0.62% - D+3 turnout model

Gravis:
O+1 - Current result
R+9.42% - 2004 turnout model
O+0.03% - 2008 turnout model
R+6.20% - 2010 turnout model
R+5.49% - 2012 registration model
R+3.32% - D+3 turnout model

Looking at the new averages across 10 polls:

Polls included:
O+3 - Survey USA
R+2 - Rasmussen
O+4 - PPP
Even - Suffolk
O+5 - Time
O+2 - Purple Strategies
O+4 - CNN/ORG
O+2 - ARG
Even - Cincinnati Enquirer
O+1 - Gravis

O+1.90% - Current RCP Average
R+7.67% - Average using the 2004 turnout model
O+1.55% - Average using the 2008 turnout model
R+4.53% - Average using the 2010 turnout model
R+3.83% - Average using the 2012 registration model
R+1.72% - Average using the D+3 turnout model

A slight movement in Obama's direction across all of the models.  The CNN/ORG poll remains the big outlier, being almost 5 points off the averages.  Interestingly, both PPP and Rasmussen are also slight 1% outliers pulling the averages in Obama's direction.  Independents favor Romney by an average of 8% across all polls.

October 29, 2012 Analysis

Polls included:

Tracking Polls
R+3 - Rasmussen Daily Track
R+4 - Gallup
O+1.3 - IBD/Tipp
R+1 - ABC/WaPo

Others
O+1 - Battleground
Even - NBC/WSJ
R+3 - Monmouth
O+2 - CBS
R+2 - AP/GfK

R+0.97% - Current RCP Average
O+0.23% - Average using the 2008 turnout model
R+2.54% - Average using the D+3 turnout model
R+4.78% - Average using the 2010 turnout model
R+4.81% - Average using the 2004 turnout model
R+6.56% - Average using the Rasmussen Party ID turnout model

A lot of shifts in the individual polls that added up to a pretty strong shift in Romney's direction in the models.  First of all, both Rasmussen and Gallup dropped a point as usual with weekend polling.  This is typical for Sundays.  IBD/Tipp showed significant movement in Romney's direction, and also an increase in the undecideds.  Also the ABC/WaPo poll stayed the same, but their sample moved from D+4 to D+6 and Independent support for Romney increased to 20%.

Independents are favoring Romney by an average of 8% across all 9 polls.

The new Battleground poll enters the average, with improved results for Obama.  However, the Zogby poll leaves the average with it's very very good numbers for Obama.  Of the 9 polls, only the CBS poll is very good for Obama, and the Battleground poll is ok for him.  Those two polls alone are depressing Romney's support by about 1.2% in the D+3 model.

Battleground Poll - October 29, 2012


Likely Voter, 1000 sample size, Obama leads 49-48, 3% undecided.

This poll is surprising a of people, since Brit Hume said on the air yesterday that it would report Romney +5 today.  Let's look at the details and the reweights, then I'll make a comment or two.  First of all, the D/R/I of this poll remains the same as last time at D+2 or 43/41/18.  There was also a significant shift in Independent preference from Obama +1 to Romney +10.  However, Obama picked up 2 points of support, and Romney lost 1.

Putting these results through the models, we get the following:

O+1 - Current result
O+2.3 - 2008 turnout
O+0.2 - D+3 turnout
R+1.7 - 2010 turnout
R+1.7 - 2004 turnout
R+3.2 - Rasmussen Party ID

This poll is showing solid gains for Obama in the last week across all turnout models.

Now, having said that, I'm very skeptical of this poll.  73% of Romney supporters say they are "extremely likely" to vote while only 60% of Obama supporters say the same.  Additionally, the preference of Independents moved from a tiny advantage for Obama to a 10 point advantage for Romney.  Yet Romney is trailing?

@numbersmuncher mentioned this yesterday, so I'm going to say it flat out.  I think Democrats are lying to the pollsters that they have already voted.  They are doing this to game the polls, because if they say they have voted, they go straight through the likely voter screens.  You don't lose elections with a 13% enthusiasm edge and 10% edge in Independents.  That is demographically impossible.  One set of numbers or the other are wrong in this poll.

Sunday, October 28, 2012

Really Smart Analysis from Josh Jordan (@numbersmuncher)


I'm not going to try to paraphrase what he is saying.  It is a similar argument to what I have been saying about Ohio, and is backed up by the Ohio reweights I've been doing.

Just go read it.

I will quote his final point though.  I fully agree with it.

We’ll know in a little over a week which polls are right between national and Ohio, but history would suggest that come Election Day we’ll be looking back at the Ohio polls and trying to figure out how they could be so wrong.

Well, I won't be trying to figure out why they were so wrong, I'll be able to tell you.

October 28, 2012 Ohio Update

RCP dropped Gravis, Quinnipiac, and Fox from the Ohio average and added a new poll from the Cinncinati Enquirer.  This new poll shows the race tied at 49-49, but like all the rest it has some significant problems.  The D/R/I for this poll is 47/44/10.  How any legitimate pollster can publish a poll that has almost half of the sample as Democrats is beyond me.  It is also significant that they do not mention the preference of Independents in the poll.  They are able to tell you the percentage of respondents who are scared if the other candidate wins, but not this basic metric that has proven to be critical to every race in history.  In order to run my model, I am assigning Independent preference to be equal to the current average across all polls of 8%.  I suspect it is probably much higher, but we can't have a poll showing Romney leading in Ohio, can we?

Reweighting this poll we get the following:

Even - Current result
R+6.10% - 2004 turnout model
O+0.59% - 2008 turnout model
R+3.76% - 2010 turnout model
R+3.18% - 2012 registration model
R+1.76% - D+3 turnout model

Looking at the new averages across 9 polls:

Polls included:
O+3 - Survey USA
Even - Rasmussen
O+1 - PPP
Even - Suffolk
O+5 - Time
O+2 - Purple Strategies
O+4 - CNN/ORG
O+2 - ARG
Even - Cincinnati Enquirer

O+1.89% - Current RCP Average
R+7.85% - Average using the 2004 turnout model
O+1.41% - Average using the 2008 turnout model
R+4.70% - Average using the 2010 turnout model
R+3.99% - Average using the 2012 registration model
R+1.88% - Average using the D+3 turnout model

Several polls that had good internals (D+9 and D+11) for Romney dropped out of the average, bringing the numbers down across all models.  However, Romney retains his lead in Ohio.  Who will be the first pollster to break the moratorium and actually show Romney leading?

October 28, 2012 Analysis

Polls included:

Tracking Polls
R+4 - Rasmussen Daily Track
R+5 - Gallup
O+2.1 - IBD/Tipp
R+1 - ABC/WaPo

Others
R+2 - Battleground
Even - NBC/WSJ
R+3 - Monmouth
O+3 - Zogby
O+2 - CBS
R+2 - AP/GfK

R+0.99% - Current RCP Average
O+0.51% - Average using the 2008 turnout model
R+2.23% - Average using the D+3 turnout model
R+4.47% - Average using the 2010 turnout model
R+4.50% - Average using the 2004 turnout model
R+6.25% - Average using the Rasmussen Party ID turnout model

The meme over the last two days has been the "Romney's momentum has stalled" with various people, like Nate Silver trying to prove that with dubious mathematics.  Romney's momentum has only stalled in the sense that there is no one left to convince.  Rasmussen moved out to a 4 point lead, with a 5 point lead in his swing state tracking poll.  Gallup remained at 5.  Even the horribly over sampled IBD/Tipp poll moved toward Romney as some undecideds decided to vote for him.  Independents are favoring Romney by insane amounts.  Rasmussen is seeing Romney lead them by 23 points.  On the Economy sub samples, Romney is simply crushing Obama.

This has led to the second meme that is developing.  There really needs to be a name for this one, similar to Godwin's Law (the first person to call the other a Nazi loses).  Losing campaigns always fall back on "we might lose the popular vote, but still win the Electoral College".  Several media outlets started floating this one yesterday pointing to "Ohio's stubborn refusal to follow national trends and support Romney".  Rubbish.  If Romney wins nationally by over 2 points, he will win the Electoral College too.  And considering where the Rasmussen and Gallup party identification numbers are sitting right now, a 5 or 6 point Romney win is looking very possible.

And Romney is leading in Ohio.  I will offer further proof in my next post.

Saturday, October 27, 2012

October 27, 2012 Ohio Update

Three new polls entered the RCP average yesterday, bringing the total number of polls up to a mind numbing 11.  The Purple Strategies, and ARG polls were very good for Romney, while the CNN/ORG poll was good for Obama:

Purple Strategies: Obama +2, D/R/I 34/27/39, Ind +2 for Romney
CNN/ORG : Obama +4, D/R/I 35/32/33, Ind +5 for Obama
ARG: Obama +2, D/R/I 43/34/23, Ind +21 for Romney

The ARG one is particularly amusing, with a 21 point lead for Romney with Independents and a D+9 sample.  There sure appears to be a concerted effort to not report a poll in Ohio that has Romney ahead.  This is the kind of thing that is leading to Ace's angst.

Anyway, with these new polls in the average, things don't change very much:


Polls included:
O+3 - Survey USA
Even - Rasmussen
Even - Gravis
O+5 -Quinnipiac
O+1 - PPP
O+3 - Fox
Even - Suffolk
O+5 - Time
O+2 - Purple Strategies
O+4 - CNN/ORG
O+2 - ARG

O+2.27% - Current RCP Average
R+8.53% - Average using the 2004 turnout model
O+1.08% - Average using the 2008 turnout model
R+5.27% - Average using the 2010 turnout model
R+4.54% - Average using the 2012 registration model
R+2.33% - Average using the D+3 turnout model

About a half point drop for Romney across all the models, however, Romney continues to lead comfortably.  Keep in mind the early voting information that was published yesterday by the Romney campaign.  Democrats are under performing their 2008 early vote totals by 8.05% while Republicans are over performing their 2008 totals by 5.82%.  Obama can not reach the 2008 vote totals he had with this level of early voting, and especially when losing Independents.  Across all 11 polls, Romney is leading with Independents by an average of 10 points. The best result Obama can hope for is a D+3 result, which puts him over 2 points behind, given the preference Independents have for Romney.

October 27, 2012 Analysis

Polls included:

Tracking Polls
R+3 - Rasmussen Daily Track
R+5 - Gallup
O+2.3 - IBD/Tipp
R+1 - ABC/WaPo

Others
R+2 - Battleground
Even - NBC/WSJ
R+3 - Monmouth
O+3 - Zogby
O+2 - CBS
R+2 - AP/GfK

R+0.87% - Current RCP Average
O+0.63% - Average using the 2008 turnout model
R+2.12% - Average using the D+3 turnout model
R+4.35% - Average using the 2010 turnout model
R+4.39% - Average using the 2004 turnout model
R+6.13% - Average using the Rasmussen Party ID turnout model

There was very little movement in the turnout averages yesterday, with all of the averages moving a miniscule 1/100th of a percent.  Romney retains his lead in all turnout models that are likely to occur on November 6th.

There are some odd things going on that make me question what is really happening though.  The ABC/WaPo tracking poll is showing Romney ahead by 1.  However, the D/R/I on the poll is only D+4 at 34/30/32.  They are also reporting Independents favoring Romney by an incredible 20 points.  It seems strange that Romney would only lead by 1, even with a D+4 sample, when he is drawing that level of support from non-partisans.  The only way that is possible is if a lot of respondants are saying they are Republicans voting for Obama, or they are reporting the 20 point lead on a much smaller subsample.  Their wording of the statement tends to make me think the latter.  They say "true independents favor Romney by 20 points".  My thought is that the 32% sample is made up of a lot of "Democrat leaning Independents".  But I don' know, they don't give a break down in the internals.

Ramussen has a similar problem.  He is finding Independents favoring Romney by 17 points, but only showing Romney up 3.  There was a rumor floating yesterday that you can only get to his results with a D+6 sample, but I was never able to find the source of the rumor.

Finally Gallup posted an analysis of the 2012 electorate.  Their headline was bizarrely titled "2012 electorate very similar to 2008", but when you actually look at the partisan split in the results you see a shift from a 54-42 lead by Democrats  to a 49-46 lead by Republicans (with leaners).  This is a 15 point shift in party identification, and matches the monthly Rasmussen Party ID poll that is showing a 3 month average of R+2.6.  Rasmussen does not factor this information into his daily tracking poll.  Even without leaners, Gallup is finding an 11 point shift in hard partisan identification toward the Republicans.

Three polls.  Three sets of data that make you scratch your head, because things aren't adding up.  The trends in the underlying information are pointing to much stronger Romney support than we are seeing.  What if my Rasmussen Party ID model above is the most correct?  What if Romney really is leading by 6.13%?  That would be similar to 1988, when Bush beat Dukakis 53.4% to 45.7%.

Friday, October 26, 2012

Trends

Since I've now been posting these averages for 30 days now, I thought it would be worthwhile to take another look at the polling trends over the last month.  The chart below shows the trends for the RCP Average, Rasmussen's daily track, and my 2008, 2010, D+3, and Rasmussen Party ID models.





There are a couple interesting things I'd like to point out.  First of all, notice that the tracks for my models are remarkably stable, especially from 10/13, through 10/22.  The tracks had a blip on the 22nd and 23rd, because I added Gallup into the mix on the 22nd, and RCP cycled 10 polls during those 2 days.  But note how quickly the tracks returned to their steady state.

The two lines that aren't steady during these periods are Rasmussen's daily track, and the RCP average itself.  The RCP average is wandering all over the place, while the reweight models stay on very steady horizontals.

Second, Obama's campaign was doing very well on October 1st.  Romney's support had been degrading up to that point, and the trend line was looking bad.  The race changed for good on October 6th.  That is when the debate performance became a factor in the polls.  Romney's support has never tapered off since.  Romney had a slight increase in his lead after the VP debate, but that settled back to his level as of the 6th.

Finally, you can see from the graph that the RCP is averaging out to about a D+5 advantage across all of the polls.  That is 2 points better for Obama than the best case scenario discussed by any pundit during this election season.  There are going to be a lot of shocked people on November 7th.




October 26, 2012 Analysis

Polls included:

Tracking Polls
R+3 - Rasmussen Daily Track
R+3 - Gallup
O+2 - IBD/Tipp
R+3 - ABC/WaPo

Others
R+2 - Battleground
Even - NBC/WSJ
R+3 - Monmouth
O+3 - Zogby
O+2 - CBS
R+2 - AP/GfK

R+0.90% - Current RCP Average
O+0.62% - Average using the 2008 turnout model
R+2.13% - Average using the D+3 turnout model
R+4.36% - Average using the 2010 turnout model
R+4.40% - Average using the 2004 turnout model
R+6.13% - Average using the Rasmussen Party ID turnout model

An unusual day yesterday when three of the four daily tracking polls showed the same results, Romney leading 50-47 (why is that 47% number so familiar?).  After reweighting, the IBD/Tipp poll is pretty close to the others, so it should also be close to 50-47, if it wasn't for their weird sample.

Overall the averages remained stable, with about .25% more support for Romney across the board.  Just a reminder, Romney continues to outperform the national averages in Ohio by about .5%.  He is ahead by 2.13% nationally with a D+3 turnout, but ahead by 2.72% in Ohio using the same turnout. 

Thursday, October 25, 2012

October 25, 2012 Ohio Update

Rasmussen updated his Ohio poll to reflect movement toward Romney, and a tie.  Also, Survey USA released a new poll which did not change from their last Ohio poll.

The big news last night was the Time Obama +5 poll, that everyone was concerned about.  Don't be.  It is a D+10 poll where Independents favor Romney by 14 points.  Also, someone made a math error in the poll, because when I add up the respondents, I get 694 likely voters/already voted.  Not the 742 that they are touting.  In any case, putting their numbers through my Ohio model, we get:

R+8.64% - Using the 2004 turnout model
O+1.28% - Using the 2008 turnout model
R+5.27% - Using the 2010 turnout model
R+4.54% - Using the 2012 registration model
R+2.24% - Using a D+3 turnout model

In other words, it is exactly in line with the other Ohio polls, off by a few tenths of a percent in all turnout models.  The Ohio averages now include 8 polls, so the fidelity is getting to be pretty good.

Polls included:
O+3 - Survey USA
Even - Rasmussen
Even - Gravis
O+5 -Quinnipiac
O+1 - PPP
O+3 - Fox
Even - Suffolk
O+5 - Time

O+2.13% - Current RCP Average
R+8.83% - Average using the 2004 turnout model
O+0.64% - Average using the 2008 turnout model
R+5.61% - Average using the 2010 turnout model
R+4.89% - Average using the 2012 registration model
R+2.72% - Average using the D+3 turnout model

The results shifted less than 0.05% in all of the reweighted models.  Romney continues to lead Obama by a comfortable margin.

Stepping away from the polls for a second

I had the privilege of meeting Mitt Romney in 2008 on a cruise, a week after Obama had won the election.  This was long before anyone knew what the Obama presidency would be like.  We feared the coming disaster, but hoped we were wrong.

I was struck at how humble he seemed.  He was articulate and clearly smart.  He understood economic issues and could articulate them well.  He was a little goofy in how he would talk.  Not like he was forcing it, just that he wasn't a professional at public speaking.  He came across as genuine.

Meeting people, and shaking their hands, he was warm and personable.  Happy to accept well wishes, while being humble about why he lost to McCain.

It turns out that I look a lot like Mitt, and spent the entire week being asked if I was his son.  It was pretty funny.  I show pictures of the two of us together and everyone laughs at the resemblance.

The point is that this man is a genuine man.  Yes, he is wealthy, but that never came across.  You don't get the impression that he considers himself to be an exceptional man who to whom being President is a birthright.

Beyond the polls, you know Mitt Romney will win this election.  His rally at Red Rocks in Colorado showed me that he has grown into what the country wants and needs in this time of turmoil.  This election is no longer about his desire to be President.  He has now become the vessel through which the will of the voters is being expressed.





But this isn't the same as the adoration that Obama received in 2008, where he was being elevated to a God-King.  That was a man born of hubris, convinced that the adulation was for him.  That he was the Rock Star, that he was deserving of the love showered upon him, because of his own special magnificence.

With Romney is it different.  He understands that he is just a man, a man put in this place, at this time, to express the will of We the People.  It is a rare individual who can separate his ego from the roar of the crowds, and see a higher purpose.




If you haven't, go watch the videos of Red Rock.  It is the picture of a man who is stunned by what is happening.  A man who is realizing that he is going to be the next President, and the enormous responsibility that will hold.

I read a report from Ulsterman's Insider a few minutes ago, that brought tears to my eyes.  Let me quote one small passage.

And enough people notice that the area gets a lot more quiet, and they are trying to watch the governor without looking like they are watching the governor.  They can all kind of tell something is happening right then.  It was described as something very peaceful and powerful that came over that backstage area for a moment.  And the governor, he lowers his head and his eyes shut tight and you could see him take a slow deep breath and then he lets it out and says quietly, but just loud enough for some to hear, “Lord, if this is your will, please help to make me worthy.  Please give me the strength Lord.”  And then his eyes open up, and he’s back to smiling and laughing and shaking hands and being the candidate once again.
I don't need a poll to tell me this man will win.

Update:  By request, the picture of Mitt and me.

October 25, 2012 Analysis

Polls included:

Tracking Polls
R+4 - Rasmussen Daily Track
R+3 - Gallup
O+3 - IBD/Tipp
R+1 - ABC/WaPo

Others
R+2 - Battleground
Even - NBC/WSJ
R+3 - Monmouth
O+3 - Zogby
O+2 - CBS
R+2 - AP/GfK

R+0.70% - Current RCP Average
O+0.79% - Average using the 2008 turnout model
R+1.97% - Average using the D+3 turnout model
R+4.19% - Average using the 2010 turnout model
R+4.24% - Average using the 2004 turnout model
R+5.96% - Average using the Rasmussen Party ID turnout model

The very large Romney samples that had been included in the Gallup tracking poll finally dropped out of their average yesterday, bringing their results down to a Romney +3 lead.  Importantly, now both Gallup and Rasmussen are lining up, showing Romney at 50%.  The IBD/Tipp track picked up a bit of support for Obama, but remains over sampled with a heavy D+7 skew.  ABC/WaPo remained about the same.

The new poll entering the average is the monthly AP/GfK poll that shows a decent lead for Romney.  This is one of the first polls I've noticed that is correcting it's sample to reflect a more reasonable assumption regarding turnout.  It is now showing D+4, where a month ago it was D+13.

Overall the race remains stable after the Monday debate with very little movement for either candidate.  Romney's averages dropped slightly due to the Gallup drop, but he retains the momentum.

AP/GfK Poll - October 25, 2012

Likely Voter, 839 sample size, Romney leads 47-45, 4% undecided.

Update:  It was pointed out to me that I looked at the wrong table in the internals, so I've corrected.

The last time I looked at this poll, I used it as an example in my methodology explanation.  Back in September, it had a D+13 sample differential, with Democrats making up 50% of the overall partisan sample.  Now we have an example of a poll starting to reflect reality.  Using a likely voter screen, they end up with a D+4 sample, which has a D/R/I of 34/30/36.  They also show a slight preference of Independents for Romney by 2%.

Putting the results into the reweighting models we get the following:

R+2 - Current result
R+0.7 - 2008 turnout model
R+3.8 - D+3 turnout model
R+6.2 - 2010 turnout model
R+6.3 - 2004 turnout model
R+8.1 - Rasmussen Party ID

Another poll to include in the average showing a steady lead for Romney.

Wednesday, October 24, 2012

Senate Update - October 24, 2012

With the Mourdock kerfuffle last night, I figured I would update my Senate odds.  I'm going to call this the anti-Silver prediction, since I'm going to make some optimistic assumptions and discount a few polls in the process.  Hell, if he can do it and land a gig and the NYT, why can't I?

Here is my opinion on all 16 of the current Senate races I am following.
  • Nebraska 100% -  This is Ben Nelson's seat, and Deb Fisher locked it up ages ago.
  • Indiana 100% - Mourdock is up by 5 in this race, with 2 weeks to go and Romney leading by 15 in the state.  He will win this, no matter how much the media tries to twist the rape comment.
  • Nevada 100% - Heller is up 7 and over the 50% mark.  This race is over.
  • Wisconsin 75% - Rasmussen is showing Thompson with a 2 point lead, and is polling better than Romney.  As Romney's lead grows, so will Thompson's.
  • Montana 60% - The polling didn't change, but my opinion did.  Tester is an incumbent and stuck at 48%.  Undecideds will break toward Berg, and Romney will have coattails.
  • Massachusetts 40% - Warren is holding her lead, but I distrust some of this fly-by-night polling.  I think she really leads by 2 in this race.  Brown will need to win the turnout battle.
  • Virginia 60% - Kaine is ahead 49-48 in Rasmussen.  I think Obama's Navy gaffe is going to sink his chances, and Allen will pick up the 2 points he needs.
  • Florida 40% - Mack continues to trail, but Nelson seems to have a ceiling of 47%.  If Mack was closer, I would increase my confidence in this race.
  • Ohio 75% - Mandell is catching up to Brown in all the "official" polls.  However, they are all the same ones that are over sampling Democrats from D+6 to D+11 (7 polls like this now in the RCP average).  I don't trust these polls.  If you look inside them and adjust to a normal Ohio turnout, Mandell leads.  Besides, even in the skewed versions of the polls, Brown can't get to 50%.
  • New Mexico 10% - I don't think anyone is bother to even poll here any longer.  Pretty sure the Dems will win this race.
  • Missouri 35% - No new polling here.  It is a weird race, but I like Akin's chances with Romney leading in the state by 12.
  • Michigan 10% -  No one is even polling this race.
  • Pennsylvania 40% - Casey's top line number is stuck in the mid 40s, just like OH, MT, and FL.  Some polls are showing Smith getting very close.  This would be my upset special.
  • Connecticut 40% - McMahon is getting close again, and Murphy still can't break 50%.  One recent poll puts her within 1.
Odds of at least a tie in Senate = 95.8%
Odds of winning the Senate = 89.1%

Update:  Fortunately, Optimizer is checking my math.  I left two races off the list:

  • North Dakota 100% - Berg is now up 5 points in the race, almost to 50%, and Romney will win by double digits.
  • Maine 10% - The Independent will win this race.  He could decide to caucus with the GOP, especially if it means being in the majority.

A Small, Yet Important Point

A lot of people have noticed that historically, Ohio tends to vote slightly more Republican than the national results.  Going back a number of years, Ohio votes consistently about .5% more toward the GOP than the national share the GOP gets.

I'm also seeing a lot of angst about the Ohio RCP.  "I'd be more comfortable if that would stop showing Romney tied!".

This is where the reweighting I am doing shows the truth of the matter.  Look at the D+3 turnout model.

D+3 national average: Romney +2.17%
D+3 Ohio average: Romney +2.79%

If Ohio has a similar turnout to what is seen nationally, then Ohio will show .62% more support for Romney than his national share.

Quit stressing out about Ohio.

October 24, 2012 Analysis

I didn't realize this, but apparently the ABC/WaPo poll will be a tracking poll from now until the election.  All of the polls from yesterday remained in the average, but there was movement in all four of the tracking polls.

Polls included:

Tracking Polls
R+4 - Rasmussen Daily Track
R+5 - Gallup
O+2 - IBD/Tipp
R+1 - ABC/WaPo

Others
R+2 - Battleground
Even - NBC/WSJ
R+3 - Monmouth
O+3 - Zogby
O+2 - CBS

R+0.89% - Current RCP Average
O+0.57% - Average using the 2008 turnout model
R+2.17% - Average using the D+3 turnout model
R+4.38% - Average using the 2010 turnout model
R+4.43% - Average using the 2004 turnout model
R+6.14% - Average using the Rasmussen Party ID turnout model

The drop we saw yesterday was erased today due to improved support from Romney in the tracking polls.  Obama received an additional point of support in Gallup, while the Romney top line number of 51 remained unchanged.  Meanwhile, the other 3 tracking polls all showed a 2 point move toward Romney.  Romney leads by at least 2 points in the race, nationally.  Baring some very unlikely occurrence, such as a major gaffe, this race will now remain stable through the election, and Romney will win.

Tuesday, October 23, 2012

October 22, 2012 Ohio Update

We have one new Ohio poll from Suffolk University to include in the mix.  Their D/R/I is 39/35/27, and they do not break down support from Independents individually.  It is in line with the other Ohio polls:

R+8.50% - Using the 2004 turnout model
O+0.78% - Using the 2008 turnout model
R+5.34% - Using the 2010 turnout model
R+4.66% - Using the 2012 registration model
R+2.50% - Using a D+3 turnout model

When included with the other Ohio polls, the numbers don't change very much:


Polls included:
O+3 - Survey USA
O+1 - Rasmussen
Even - Gravis
O+5 -Quinnipiac
O+1 - PPP
O+3 - Fox
Even - Suffolk

O+1.86% - Current RCP Average
R+8.86% - Average using the 2004 turnout model
O+0.78% - Average using the 2008 turnout model
R+5.66% - Average using the 2010 turnout model
R+4.94% - Average using the 2012 registration model
R+2.79% - Average using the D+3 turnout model

Romney continues to lead in all reasonable turnout models.

Is my Methodology Valid?

Last night I had a brief Twitter discussion with Ace regarding whether or not what I am doing has any validity.  To be fair, he is being reasonably skeptical of whether or not poll rebalance is viable, and can even be done.  Do we have enough information to even come up with valuable conclusions?  Here is his final tweet from last night:

I believe most polls are tilted too Democrat.  What I doubt is that is is possible to post-facto rebalanced missing data.  - Ace

Let me make my argument why I believe this is a valid methodology.  The be sure, we won't know if I am right until the election.  This is when we will know the actual vote totals and the partisan split of the electorate.  However, I think we have some data that gives indication that this methodology  might be valid.

First of all, my basic belief is that elections are based on two factors only.
  1. How many partisans of each side can the parties get to the polls?
  2. What is the opinion of the non-partisans regarding the two candidates?
Democrats vote for Democrats or don't vote.  Republicans vote for Republicans or don't vote.  Independents will vote for one or the other candidate, based on the persuasiveness of the arguments and the facts on the ground (e.g. I hate the war, the economy sucks, I want my free stuff).

I fundamentally do not believe that Democrats voting for Romney, or Whites supporting Obama are of any practical value in predicting voter behavior.  People self identify as either partisan (which does not mean they are necessarily registered in their preferred party), or non-partisan and willing to change their vote election by election.

Partisans move from one party to another slowly, and once they move it is a personal decision based on core philosophy.  They will not change their minds based on the candidate and vote for the other team.  But they might choose not to vote, which is reflected by enthusiasm.

Enthusiasm is ultimately determined through turnout.  If Republicans are more enthusiastic than Democrats, then they will show up at the polls in higher numbers.  This was demonstrated in 2008.  As I've mentioned before, most people misunderstand the 2008 election.  They think 7% more Democrats showed up to vote for Obama.  This isn't true.  Democrats were able to get their base to enthusiastically show up at 2% higher rate than in 2004 (which was also a good turnout for them).  The real difference is that 5% of Republicans didn't bother to vote, because McCain didn't enthuse them, and McCain did not run an effective GOTV operation.  The final factor is that non-partisans supported Obama overwhelmingly.

The other day, Rush discussed his belief that people don't change their minds as quickly as the polls seem to indicate.  He believes that core voting decisions are established, and that the voters do not swing wildly between the candidates.  I think he is right on this.  Poll fluctuations we are seeing are based on the sampling variations, and the fact that RCP does not correct for these fluctuations and skews.  They are comparing apples to oranges, and trying to tell us that we really were expecting a fruit salad.

I also have a problem with RCP in that it can be easily "gamed".  A few polls that are purposely biased can skew the average and show momentum for one of the candidates.  We have proof that the Obama campaign is purposely disabling credit card validation in their online fundraising operations.  Why is it so hard to believe that some polls are being purposely manipulated to affect the RCP average?

My argument to Ace is that the poll internals do provide all of the information needed to get a good view of what the election results will be given a specific turnout model.  They provide the partisan split in the votes, and they (usually) provide the non-partisan preference between the candidates.  Using this information, it is possible to adjust the poll results such that all polls averaged can be compared using the same turnout baseline.  We are comparing apples to apples, and getting applesauce.

Consider the following chart:


This chart tracks the RCP average, the Rasmussen Daily tracking poll, and my 2010 turnout average from Sept 26th to October 23rd.  Notice a couple trends.  From the 26th to the 1st, Romney was gradually sliding, until the first debate.  After the debate, his support began to move up, and then on October 13th we hit a very stable spot in the race.  In the 2010 turnout model, the Romney lead stayed very constant for 9 days.  During that period, there were no real changes in the state of the race, no gaffes, nothing that was really changing the momentum.  However, during that same period, Rasmussen stayed pretty even, bouncing between R+1 and R+2, while RCP began to drop down to an Obama lead.

My point is that there was no reason for this drop, other than the new introduction of polls with widely divergent samples.  The reweighted polls showed stability in the race, that reflected the our general sense of where the race stood during that time.  The RCP average did not.

Now I am not saying that the election will resemble 2004 in turnout.  I am only saying that rebalancing to potential turnout scenarios gives a better picture of the state of the race over time.  This is why I'm offering 5 different turnout models.  Voter enthusiasm and GOTV operations will determine which of the models ends up being the valid view of the election.

What my models do provide is a good view of what results we can expect if the electorate resembles a specific model.  Since we know that 2008 level turnout is highly unlikely, we can then say "If we see a D+3, then the polls say this" and "If the GOP repeats the 2004 turnout, then the polls say that".  I find this more useful that looking at a D+9 poll that says Obama is ahead by 5, calling it BS and ignoring it.  It also helps in places like Ohio, when we get a series of 7 out of 10 polls that sample Democrats between D+6 and D+11.  We are able to get value out of the polls anyway, rather than ignore them and getting a badly skewed view of what is really happening in Ohio.

October 23, 2012 Analysis

RCP dropped the Hartford Courant poll from their average, and included 4 new polls: Monmouth, Zogby, ABC/WaPo, and CBS.  This is one of those odd days where the RCP average didn't change much at all, but the underlying dynamics of the race did.  Most of the new polls added were very good for Obama, especially Zogby and CBS.  As a result, the overall averages under the different turnout models moved significantly in Obama's direction.

Polls included:
R+2 - Rasmussen Daily Track
R+6 - Gallup
O+4 - IBD/Tipp
R+2 - Battleground
Even - NBC/WSJ
R+3 - Monmouth
O+3 - Zogby
O+1 - ABC/WaPo
O+2 - CBS

R+0.33% - Current RCP Average
O+1.16% - Average using the 2008 turnout model
R+1.57% - Average using the D+3 turnout model
R+3.76% - Average using the 2010 turnout model
R+3.81% - Average using the 2004 turnout model
R+5.51% - Average using the Rasmussen Party ID turnout model

In the last two days we have seen 6 new polls enter the average, and 4 polls drop out.  We now have 9 polls included which should be giving a decent statistical view on the current race.  Romney continues to lead in all turnout models that we can reasonably expect (D+3 or better), but his margin has shrunk by about .5% from where he was 2 or 3 days ago.

ABC/WaPo Poll - October 23, 2012

Likely Voter, 1376 sample size, Obama leads 49-48, 1% undecided.

Unlike the CBS poll, the ABC poll is an OK one for Romney, though not as strong as some of the others we have seen recently, like Battleground, NBC, and Monmouth.  The D/R/I of the poll is a supportable D+5 at 34/29/37.  There is at least an argument to be made for a turnout similar to this.  ABC provides no information in their internals regarding the preference of Independents.  As I result, I am having to assume it is 50/50.

Putting the results into the reweighting models we get the following:

O+1 - Current result
O+2.6 - 2008 turnout model
R+0.5 - D+3 turnout model
R+2.9 - 2010 turnout model
R+3.0 - 2004 turnout model
R+4.9 - Rasmussen Party ID

This poll supports the conventional wisdom that we will have a D+3 turnout, and a close race.

CBS Poll - October 23, 2012

Likely Voter, 790 sample size, Obama leads 48-46, 4% undecided.

I'm not going to try to spin this one.  This poll is very good for Obama, and his campaign is hoping that it is representative of the election.  It is an outlier, in that it doesn't track with other polling results.  First of all the D/R/I on this poll is dead even 32/32/38.  Additionally, Independents in the poll favor Obama by 6 points.  Showing an Obama lead with this sample is a very good result for the Democrats.

Putting the results into the reweighting models we get the following:

O+2 - Current result
O+6.0 - 2008 turnout model
O+3.0 - D+3 turnout model
O+0.5 - 2010 turnout model
O+0.5 - 2004 turnout model
R+1.5 - Rasmussen Party ID

These results are fairly similar to the Zogby results.  Again, there is no way to spin this, other than to note that the internals don't match the results of most other polls.

Monday, October 22, 2012

Wash Times/Zogby Poll - October 22, 2012

Likely Voter, 800 sample size, Obama leads 50-47, 3% undecided.

Unlike all of the other polls recently, Zogby provides some good news for Obama.  The D/R/I sample for this poll is a not terrible 37/35/28.  Independents favor Romney by 7 points.  Which makes you really wonder about these results.  The only way you can get to these results is if significantly more Republicans support Obama than Democrats support Romney.

What can I say, it's Zogby.

Putting the results into the reweighting models we get the following:

O+3 - Current result
O+5.4 - 2008 turnout model
O+2.8 - D+3 turnout model
O+0.5 - 2010 turnout model
O+0.5 - 2004 turnout model
R+1.2 - Rasmussen Party ID

Well, at least is gives the Democrats some hope.  I'm sure they were getting pretty depressed.  Of course, pinning your hopes on Zogby being right is not where I would want to be.

Monmouth Poll - October 22, 2012


Likely Voter, 1402 sample size, Romney leads 48-45, 5% undecided.

Another new poll, this time from Monmouth University.  This is a very large sample size, double what we see in other polls.  Typically, this means a lower margin of error.  The D/R/I for this poll is 35/31/34, which at D+4 isn't too bad, considering some of the other polls out there (Tipp I'm looking at you).  Independents in this poll favor Romney by a huge margin, 19%.

Putting these results through the models, we get the following:

R+3 - Current result
R+2.0 - 2008 turnout
R+4.9 - D+3 turnout
R+7.3 - 2010 turnout
R+7.3 - 2004 turnout
R+9.3 - Rasmussen Party ID

What can I say?  There is no good news for Obama in this poll, no matter how you slice it.

October 22, 2012 Ohio Update

I was hoping not to have to do a daily update on Ohio, but the pace of new polling coming out of Ohio is making that impossible.

Today CBS/Quinnipiac (I've actually learned to spell that now) released a new Ohio poll showing Obama ahead by 5, 50-45.  It continues the tradition in Ohio, now being the 7th poll in a row to over sample Democrats by an unreasonable amount, in this case by 9 points.  The D/R/I of their sample is 35/26/38.  Amusingly, the Republican's have a 12 point enthusiasm advantage and Independents favor Romney by 7 points.  Yet Obama still leads by 5.  Anyway, putting this poll into the Ohio model, we get the following:

R+6.79% - Using the 2004 turnout model
O+4.31% - Using the 2008 turnout model
R+3.05% - Using the 2010 turnout model
R+2.27% - Using the 2012 registration model
O+0.36% - Using a D+3 turnout model

For once, Obama does actually lead in the D+3 turnout model, making this poll slightly better for him, than others.  However, I question how he gets to a D+3 turnout when he is 12 points under water in enthusiasm.

Here are the results of averaging all of the current polls in the RCP average:

Polls included:
O+3 - Survey USA
O+1 - Rasmussen
Even - Gravis
O+5 -Quinnipiac
O+1 - PPP
O+3 - Fox

O+2.17% - Current RCP Average
R+8.92% - Average using the 2004 turnout model
O+0.51% - Average using the 2008 turnout model
R+5.72% - Average using the 2010 turnout model
R+4.99% - Average using the 2012 registration model
R+2.84% - Average using the D+3 turnout model

This update to the RCP gives Romney about another .25% support across all turnout models, and he retains his lead in all possible turnout scenarios.

October 22, 2012 Analysis

Welcoming Gallup to the averages

 Thanks to all of the criticism that Gallup has been getting, they revealed a bit of their internals over the weekend.  Enough that I think I can estimate their poll enough to include it in my averages.  Since RCP just uses top line results, I don't think I can be criticized for lack of integrity.  I am using a D/R/I for Gallup of 36/37/27 and assuming no preference by Independents.  This might be a little conservative, but it lets me get them into the averages.  If that ridiculous Tipp poll can be included, Gallup should be too.

Polls included:
R+2 - Rasmussen Daily Track
R+7 - Gallup
O+6 - IBD/Tipp
R+2 - Battleground
Even - NBC/WSJ
O+3 - Hartford Courant/UConn

R+0.38% - Current RCP Average
R+0.37% - Average using the 2008 turnout model
R+2.87% - Average using the D+3 turnout model
R+4.86% - Average using the 2010 turnout model
R+4.94% - Average using the 2004 turnout model
R+6.42% - Average using the Rasmussen Party ID turnout model

Several polls were dropped over night, and the new Battleground and NBC/WSJ polls were added.  People hoping for an Obama victory should be worried, since the only two polls that are good news for him are Tipp and UConn, where both firms do not have a reputation for accuracy.

With this update, every average moves into an advantage for Romney, even the 2008 turnout model.

Battleground Poll - October 22, 2012


Likely Voter, 1000 sample size, Romney leads 49-47, 4% undecided.

Last night I was wondering whether the polls would continue with their Democrat over samples, or start to move toward a rational baseline.  Battleground has chosen the later, and looks to be the first to start adjusting their results to better reflect the likely turnout in 2012.  The last time this poll was run, it used a D+4 sample of 44/40/18.  This poll moves to a D+2 of 43/41/18, which is much closer to a likely partisan split.  As usual with this poll, they undersample Independents by pushing respondents to indicate party preference, even if they aren't registered with that party.  They are also reporting a very small preference for Obama of 1 point by the miniscule number of Independents they have left.

Putting these results through the models, we get the following:

R+2 - Current result
R+0.7 - 2008 turnout
R+2.9 - D+3 turnout
R+4.8 - 2010 turnout
R+4.8 - 2004 turnout
R+6.2 - Rasmussen Party ID

Solid results for Romney, very much in line with what Rasmussen and Gallup are showing.  Again, note that with an R+1 sample like Gallup is using, the Battleground results would be Romney at 53%.

October 22 21 (night), 2012 Analysis

[note: I wrote this up before going to bed last night, then RCP went and tossed out a bunch of polls over night.  I thought the analysis at the bottom was important, so I'm posting this anyway]

Polls included:
R+2 - Rasmussen Daily Track
R+1 - Fox News
O+6 - IBD/Tipp
O+1 -Battleground
O+3 - ABC/WaPo
R+1 - Survey USA
O+3 - Hartford Courant/UConn
Even - NBC/WSJ

O+1.09% - Current RCP Average
O+0.57% - Average using the 2008 turnout model
R+2.05% - Average using the D+3 turnout model
R+4.13% - Average using the 2010 turnout model
R+4.20% - Average using the 2004 turnout model
R+5.76% - Average using the Rasmussen Party ID turnout model

Rasmussen moved to Romney +2 today. IBD/Tipp poll however is showing itself to be total garbage.  It moved from O+3 to O+6 is a single day in a multiday tracking poll.  Interestingly, it didn't do that through increasing Obama's support.  That stays stuck at 47%.  Instead support for Romney switched to Undecided.  This one poll is dragging the RCP average down to an unrealistic level.  However, note that the overall averages did not move much since yesterday.  This is because once the Tipp poll is reweighted, it comes close to the averages.

The new NBC/WSJ poll enters the average, bringing us up to 8 polls in this compilation.  I'm becoming quite confident in my model, since all of the polls are showing similar results once normalized.  For example, in the D+3 model we get a range of R+1.26 to R+3.82 (ignoring the Tipp O+0.28).  This is about a 2.5% range over 7 polls.  That is statistically very stable.

Sunday, October 21, 2012

NBC/WSJ Poll - October 21, 2012


Likely Voter, 816 sample size, Tied 47-47, 4% undecided.

The release of this poll was greeted with media cries of amazement, "Look it's tied!".  The only problem is that it isn't.  This poll continues to show Romney's polling lead.  The DRI of this poll is 44/38/18, of D+6.  They don't provide any information regarding independent preference between Obama and Romney.

When we adjust this result to the different models, we get the following results:

Even - Current result
R+1.4 - 2008 turnout
R+3.8 - D+3 turnout
R+5.7 - 2010 turnout
R+5.8 - 2004 turnout
R+7.1 - Rasmussen Party ID

Notice that this poll confirms the Gallup results.  Reportedly, Gallup is using an R+1 sample.  In the 2010 model, which is close to R+1, this poll would give Romney 52%, exactly the same as Gallup.

There is a lot of foolish criticism being leveled at Gallup, calling it an outlier.  People who should know better, like Rove and Cadell are saying this.  What they fail to see (or report) is that most of these polls are showing the Romney lead at about 52%, if they also used an R+1 sample.

It is going to be very interesting to see if all of these pollsters are going to ride their D+7 samples all the way to election day, or if they will finally come back to reality.

October 21, 2012 Ohio Update

Two new polls in the RCP Ohio average, PPP and Gravis.  It is really hard to keep from suspecting some form of conspiracy, when both polls reweighted their samples to be more favorable to Obama from the last time they were run.  The result is that neither poll shows Romney leading, when he clearly is if you examine the internals.

PPP Obama leads 49-48.  They have broadened from the D+4 they used last time to D+8, D/R/I of 42/35/23.  Independents favor Romney by 7%. Reweighting this poll using the various turnout models we get the following:

R+9.79% - Using the 2004 turnout model
R+0.77% - Using the 2008 turnout model
R+6.70% - Using the 2010 turnout model
R+6.04% - Using the 2012 registration model
R+3.95% - Using a D+3 turnout model

Again, Romney leads in all turnout models, including if the electorate matches 2008. I'm noticing a pattern.

With Gravis it is a 47-47 tie.  This one is even worse than PPP, if you can believe it.  It is D+9 with a D/R/I of 41/32/27.  Independents favor Romney by an incredible 19% (52-33), yet somehow Romney can only eek out a tie. Reweighting this poll using the various turnout models we get the following:

R+12.48% - Using the 2004 turnout model
R+3.27% - Using the 2008 turnout model
R+9.34% - Using the 2010 turnout model
R+8.62% - Using the 2012 registration model
R+6.55% - Using a D+3 turnout model

If the electorate in Ohio matched 2008, Romney would be ahead by over 3 points.  This makes 6 of the 8 polls used in the RCP average that over sample to match or exceed the 2008 turnout, yet even with this insane artificial boost, Obama can only manage a 2 point lead.

Here are the results of averaging all of the current polls in the RCP average:

Polls included:
O+3 - Survey USA
O+4 - CNN/ORG
R+1 - ARG
O+1 - Rasmussen
Even - Gravis
O+6 - NBC/WSJ
O+1 - PPP
O+3 - Fox

O+2.13% - Current RCP Average
R+8.61% - Average using the 2004 turnout model
O+0.74% - Average using the 2008 turnout model
R+5.43% - Average using the 2010 turnout model
R+6.04% - Average using the 2012 registration model
R+2.58% - Average using the D+3 turnout model

Ohio is slipping away from Obama, since even in their most optimistic model for this year (D+3) they are close to 3 points behind in Ohio.  This matches the activities of the Obama campaign, as they begin to stop campaigning in Ohio and start trying to shore up everything that is left to keep Romney from getting those last 4 EVs he needs.  This is why you are seeing campaigning in NH, CO, IA,WI, and NV.  If Obama loses any of them, he loses.

October 21, 2012 Analysis

Polls included:
R+1 - Rasmussen Daily Track
R+1 - Fox News
O+3 - IBD/Tipp
O+1 -Battleground
O+3 - ABC/WaPo
R+1 - Survey USA
O+3 - Hartford Courant/UConn

O+0.94% - Current RCP Average
O+0.58% - Average using the 2008 turnout model
R+2.09% - Average using the D+3 turnout model
R+4.22% - Average using the 2010 turnout model
R+4.29% - Average using the 2004 turnout model
R+5.90% - Average using the Rasmussen Party ID turnout model

As expected, Rasmussen moved to Romney +1 today. This averaged out with the IBD/Tipp poll which lost a point of support for Romney to "Undecided".  Tipp is on track to being a laughing stock at the end of this.  Their internals are nonsense, with Independents favoring Romney by 10, yet Romney continuing to slide.

Overall not much change from yesterday, with a small movement toward Romney in all the averages.

Saturday, October 20, 2012

October 20, 2012 Ohio Update

The new Fox Ohio poll was causing considerable discussion last night.  I went ahead and updated my Ohio analysis using this poll.

First of all, this poll is yet another poll with a ridiculous Democrat over sample of D+8.  There are a lot of polling firms that will have a lot of egg on their faces come November, because they have been assuming turnout that matches 2008.  This poll in particular finds a preference by Independents for Romney of an incredible 24%.   With only 23% Independents in the sample, this preference is severely dampened.

Reweighting this poll using the various turnout models we get the following:

R+9.95% - Using the 2004 turnout model
R+1.35% - Using the 2008 turnout model
R+7.02% - Using the 2010 turnout model
R+6.28% - Using the 2012 registration model
R+4.43% - Using a D+3 turnout model

Romney has a large, healthy lead in all Ohio turnout models, including if the electorate matches 2008.

Here are the results of averaging all of the current polls in the RCP average:

Polls included:
O+3 - Survey USA
O+4 - CNN/ORG
R+1 - ARG
O+1 - Rasmussen
R+1 - Gravis
O+6 - NBC/WSJ
O+5 - PPP
O+3 - Fox

O+2.50% - Current RCP Average
R+7.75% - Average using the 2004 turnout model
O+1.72% - Average using the 2008 turnout model
R+4.53% - Average using the 2010 turnout model
R+3.82% - Average using the 2012 registration model
R+1.65% - Average using the D+3 turnout model

With this new Fox poll, Romney extends his lead in Ohio by about a half a point across all turnout models.

October 20, 2012 Analysis

Polls included:
Even - Rasmussen Daily Track
R+1 - Fox News
O+2 - IBD/Tipp
O+1 -Battleground
O+3 - ABC/WaPo
R+1 - Survey USA
O+3 - Hartford Courant/UConn

O+0.96% - Current RCP Average
O+0.65% - Average using the 2008 turnout model
R+2.02% - Average using the D+3 turnout model
R+4.15% - Average using the 2010 turnout model
R+4.22% - Average using the 2004 turnout model
R+5.84% - Average using the Rasmussen Party ID turnout model

Rasmussen dropped Monday from his average, which was a good day for Romney.  However, he is warning that Obama's increase is coming solely from good polling on Tuesday prior to the debate.  Polling since the debate has been showing small increased support for Romney.  Depending on how polling went yesterday, the Rasmussen track should move to Romney +1 or +2 today.

The IBD/Tipp poll continues to defy logic.  Their insistence on a D+7 sample is going to be their hill to die on.  They like to talk about their accuracy last cycle, but if this is their model, that accuracy is based on being lucky to have a sample that matched the turnout.  Obama is pulled down in this poll thanks to an 11 point preference by Independents for Romney.

The Hartford Courant in now included in the average, which I posted about yesterday.  They need a 47% Democrat sample to get to Obama +3.

The numbers tightened slightly due to these changes.  Romney still leads in the D+3 model by 2 point, which puts him right under the 50% mark.  I consider this to be the worst case turnout model.  I think 2010 is the most likely model, which puts Romney at 51%, matching what Gallup is showing.  Regardless, Romney continues to lead nationally.

At this moment, Romney will win the Electoral College by at least 285 to 254.

Friday, October 19, 2012

Hartford Courant/UConn Poll - October 19, 2012


Likely Voter, 1023 sample size, Obama leads 48-45, 6% undecided.

It's been a few days since a national poll has been released, other than the tracking polls.  I've been curious to see if a new poll will confirm Romney's lead. This poll has not appeared in the RCP average for a long time, and it is good news for Romney.  The D/R/I sample for this poll is 47/39/15 with a very small (2%) Independent edge for Romney.

Yes, you read that right, 47% of the likely voter sample is Democrats.

Let the games begin.

When we adjust this result to the different models, unsurprisingly, Romney looks pretty good:

O+3 - Current result
R+0.7 - 2008 turnout
R+3.0 - D+3 turnout
R+4.7 - 2010 turnout
R+4.9 - 2004 turnout
R+6.0 - Rasmussen Party ID

Romney leads in this poll, even with a national turnout equal to 2008.