Michael Graham is making an argument in the Boston Herald that not only will Mitt Romney win the election, he’ll win it easily. “Maybe by a mile”, even. And he does it by citing polls and says the objective evidence points to a solid Romney victory that we should see “well before bedtime”.
His evidence, I’m sorry to say for Mitt Romney supporters, is not entirely persuasive. Graham cites the two most Romney friendly pollsters out there with big name recognition, Gallup and Rasmussen, to support his argument. But both pollsters have significant baggage that saps their credibility. Rasmussen did reasonably well in 2008 but took a horrible turn in 2010 when they missed the final margin in all the races they polled by an average of 5.8 points. 13 polls missed the final margin by at least 10 points, and they blew a Senate contest in Hawaii by 40 points.
There’s some evidence that Rasmussen’s bias and inaccuracy has continued again this year. From August 22nd through October 5th, Rasmussen was the only pollster in the nation to show Mitt Romney with a lead. Three polls showed a tied race and 33 showed Barack Obama with the lead, at times by as many as nine points.
Gallup switched from a registered voter model showing a consistent 2-3 point lead for Obama to a likely voter model that has shown a 5-7 point lead for Mitt Romney, the largest lead Romney has had from any poll this year. Gallup’s LV model had Obama leading John McCain by 11 points on election eve in 2008, missing the actual margin by four points. Gallup had Republicans ahead in the generic ballot (“which party would you like to see control Congress?”) in 2010 by 15 points, when the actual result was +7.
Then there’s this:
Apart from Gallup’s final poll not having been especially accurate in recent years, it has often been a wild ride to get there. Their polls, for whatever reason, have often found implausibly large swings in the race.
In 2000, for example, Gallup had George W. Bush 16 points ahead among likely voters in polling it conducted in early August. By Sept. 20, about six weeks later, they had Al Gore up by 10 points instead: a 26-point swing toward Mr. Gore over the course of a month and a half. No other polling firm showed a swing remotely that large.
Then in October 2000, Gallup showed a 14-point swing toward Mr. Bush over the course of a few days, and had him ahead by 13 points on Oct. 27 — just 10 days before an election that ended in a virtual tie.
In 1996, Gallup had Bill Clinton’s margin over Bob Dole increasing to 25 points from nine points over the course of four days.
That poor history weighs against taking either Gallup or Rasmussen seriously, yet these are the only pollsters that Michael Graham cites to support his belief that Mitt Romney will win the election. If you stop to think about it, that’s actually necessary, because even if you include Gallup and Rasmussen in an average of polls, you still find the data indicating a relatively comfortable Obama victory. You have to not just focus on Gallup and Rasmussen, you have to ignore everything else to get even remotely close to a set of data favorable to Mitt Romney’s election chances.
Graham also plays some word games in his story that don’t make a lot of sense without more data, if you stop and think carefully about what he’s saying:
And let’s start with the big one: Before Gallup suspended polling due to Hurricane Sandy, Mitt Romney was at or above 50 percent among likely voters for 14 consecutive days. No candidate above 50 percent at this point has ever lost the presidential race.
Graham provides no evidence to support that claim, but more importantly, it would be really easy to mistake that paragraph for saying that no candidate above 50% at this point with Gallup has ever lost the race. But Graham didn’t say that, and that’s problematic. Other pollsters have at times had Obama up by 50% or more with likely voters at the same time that Gallup has had Romney at or above 50%. What happens to this uncited, unproven rule of thumb if Gallup has Mitt Romney at 50%, and Ipsos has Obama at 51% on election eve?
That also wraps back around to the problem if cheery picking favorable polls, which is what Graham is guilty of doing. Only Gallup has either candidate at or above 50% with likely voters (other than RAND which I’m going to exclude because their poll format is experimental and therefore not directly comparable). There are at least seven other firms doing trackers that I know of and none of them has either candidate at 50%. The most recent Hurricane-delayed information I have has Obama between 47 and 49 points, and Romney between 46 and 49.
Graham literally ignores the existence of the other seven tracking polls and the non-trackers as well. My records show several polls in October where both candidates reached the 50% mark, the most recent being a 50-47 lead for Obama from JZ Analytics/The Washington Times 10/18 – 10/20. (You may be more familiar with JZ A when it was called Zogby.) PPP had Romney leading 50-46 on 10/12 – 10/14. Before that, CNN had Obama leading 50-47 on 9/28 – 9/30.
Given the closeness of national polling and Graham’s cherry picking two Romney-favorable polling firms, it’s hard to take any of his evidence seriously at this point.
The story doesn’t really change when you move to polls of battleground states. The first cited by Graham is Rasmussen’s average of several battleground states, a metric that is beyond useless. Averaging national polling is acceptable and useful because the polls all ask roughly the same question which would decide the same contest, and averaging their results will smooth out noise and decrease the collective margin of error. You can do that — need to — with state polls too, but only polls from the same state. It accomplishes nothing to average polls across North Carolina, Virginia, and Florida, because each contest can have a different outcome. That average will tell you nothing about who will win each state.
So when Graham says that Rasmussen’s poll model shows Obama losing 47-52 in battleground states, you have to remember that Rasmussen has a history of blowing polls by an average of 5.8 points and up to 40 points maximum, and that battleground averages are pointless.
If you want to know who will win the battleground states, you should just look at the battleground states. Here are those states with the result as an average of the most recent five polls of each state:
Colorado: Obama +1.6
Florida: Romney +0.4
Iowa: Obama +2.2
Nevada: Obama +3.6
New Hampshire: Obama +3.6
North Carolina: Romney +2.4
Ohio: Obama +2.6
Virginia: Obama +1.4
According to Nate Silver, an average lead of 2.6 points about four days away from an election should result in a win roughly 80% of the time. By all rights, Obama should win Iowa, Nevada, New Hampshire, and Ohio without much sweat. He obviously has better odds to win Virginia and Colorado, and Romney should easily win North Carolina. Florida has replaced Virginia and New Hampshire as a coin flip.
Actual polls of these battleground states doesn’t show anything like what Michael Graham and Rasmussen are saying, and that’s because averaging the most recent five polls (which in the case of Ohio could mean only having 1-2 days of polling data) will smooth out biased and inaccurate outliers like Rasmussen, and partisan Wenzel Strategies.
Here are the last ten polls of Ohio, newest to oldest:
Obama +4 (PPP)
Obama +3 (Pharos Research)
Romney +2 (Rasmussen)
Obama +3 (SurveyUSA)
Obama +3 (Grove Insight)
Obama +2 (U of Cincinnati)
Obama +6 (JZ Analytics/Newsmax)
Obama +5 (PPP)
Obama +3 (Ipsos)
Romney +3 (Wenzel)
Romney’s leads are outliers from Rasmussen and Wenzel, a poll they conducted on behalf of Citizens United. Yes, *that* Citizens United.
You’d have to seriously cook the data to come to the conclusion that Mitt Romney is leading in Ohio. Obama’s average lead there is 2.4 points over the last ten polls, and 2.2 points over the last five. But because Rasmussen only looks at Rasmussen polls, as does Graham, he sees what he wants to see because he’s cherry picking favorable polls.
If the election were held today, Obama would win Colorado, Iowa, Nevada, New Hampshire, Ohio, and Virginia. Those states carry enough electoral votes to push Obama to nearly 300, far more than the 270 he needs to win the election. And it’s quite possible that with Romney leading Florida by 0.4 points, Obama could win that state as well. That outcome is far more likely than Romney winning Ohio when confronted with a 2.4-2.2 point deficit, pushing Obama to 329 electoral votes.
Worse for Romney, Obama’s default lead with safe states means he could afford to lose Colorado, New Hampshire, Virginia — all states where he currently leads by at least 1.4 points, larger than Romney’s lead in must-win Florida — and Florida too, and still win the election 277-261.
Then there is this falsehood:
But more important, all the polls show Obama sliding or stuck. None show any upward movement.
First, even if it’s true that Obama is stuck, that’s just fine for him because Barack Obama is leading in enough states to win as things stand today (and have been standing, despite changes in places, all year long.) But it’s not true. Not only is Obama not sliding, he’s not stuck either. Not if you look at all the polls rather than just the favorable ones.
Between October 27th and November 1st, Obama has expanded his lead in Iowa, Nevada, New Hampshire, and Virginia, held steady (until late this evening) in Ohio, while cutting into Romney’s lead in Florida and North Carolina, and only giving up ground in Colorado, a state that Obama can afford to lose. That’s six gains, one loss, and a hold in eight states.
Not only has Obama showed upward movement, he’s shown it in 75% of battleground states including two that matter a great deal: Virginia and Florida, while Romney’s only real gain came in throw-away Colorado, a state Obama can lose and that doesn’t really help Romney due to it only having 9 electoral votes.
OK. But what is a swing state? Forget Virginia and Ohio. Obama’s lost so much ground he’s been forced to send Joe Biden to Pennsylvania and Bill Clinton to Minnesota — a state so blue Ronald Reagan never carried it.
I think the reason that Graham wants you to forget Virginia and Ohio is because Obama’s lead has been remarkably steady in Ohio over the past week and a half, and Obama has literally taken the lead away from Mitt Romney in Virginia just this past week. Romney can’t win the election without both states while Obama can lose one, both probably not both. That’s what makes that paragraph nothing but spin.
Just to prove the point, here are the last ten polls of Pennsylvania: Obama +4, +7, +4, +10, +5, +3, +3, +5, +6, +4.
No change since October 12th.
And Minnesota: Obama +8, +8, +10, +4, +8, +10, +8, +5, +3, +7.
No chance since September 7th.
This is no different than the Romney campaign sending its 100+ surrogates and Paul Ryan around to states that Romney isn’t personally campaigning in.
The president, on the other hand, is only up by 6 among the loony-left granola-crunchers of Oregon.
First, I’d note that if a Democrat or liberal were talking this way about people in Arizona, the right would be talking nonstop about how much the left hates America. Second, Obama’s average lead in the last ten polls of Oregon is identical to Romney’s lead in aforementioned Arizona: 7.1 points.
If Obama only leading by 7.1 points in Oregon is a cause of concern and mockery, what does that say about Romney’s 7.1 point lead in Arizona?
Between desperate, last-second proposals for a “Secretary of Business” and embarrassing ads comparing voting for Obama to a girl losing her virginity, you can smell the desperation from the Obama camp.
Is it desperate to issue a proposal that is backed by the U.S. Chamber of Commerce, and staunch pro-business Republicans like Tom Coburn? Because they both support Obama’s proposal to create a Secretary of Business. And it’s not new, either. Obama made the proposal at his 2011 State of the Union address almost two years ago.
If I’m wrong, I’m counting on you to mock me for it mercilessly next Wednesday. But I’m not wrong.
As much as I’d like to take Michael up on his offer, I don’t have time to mock fools. I’ll leave that for others to do.