Don’t Blame the Messenger — Obama Got a Decent Convention Bounce
Are the pollsters really pulling one over on conservatives to diminish turnout?
September 10, 2012 - 1:57 pm
The blogosphere is alive with conspiracy theories — that the pollsters, all aligned with the Obama campaign, or succumbing to pressure from its henchmen (e.g. David Axelrod), are spinning out poll results designed to dispirit conservatives and Republicans, and insure Obama’s election by depressing the turnout for Romney.
I have seen many bad polls this year. Quinnipiac, a respected pollster in years past and now aligned with the New York Times and CBS, has had a series of very favorable polls for Obama over the last few months in some key battleground states, including Ohio and Florida. The samples of voters who were polled in both states included a higher number of Democrats than Republicans. More significantly, the samples included voters who stated they voted for Obama over McCain in 2008 by margins 10% greater than Obama received in each state. The Quinnipiac director of polling admitted that he had not adjusted the poll sample in each state to fit a fixed ratio with allegiance to each political party or based on prior presidential voting. This is not evidence of deliberate bias, but suggests these particular polls are not to be believed — the samples are heavily skewed towards favorable results for Obama.
Public Policy Polling (PPP), a polling firm that does work for the hard-left Service Employees International Union, has had a solid bias towards Obama in pretty much every survey this year. However, there was one notable exception. On the day after Todd Akin put his foot in his mouth on the subjects of rape and the female reproductive system (two topics about which he obviously knew not a thing), many Republicans were demanding that Akin bow out of the race so that a more mainstream, sensible Republican could replace him and win the very winnable Missouri Senate seat held by Claire McCaskill, the highly vulnerable incumbent.
On the day after Akin’s interview became big news, PPP released an instant poll that showed Akin ahead by one point. Oddly enough, after oversampling Democrats in pretty much every other state so far this year, PPP polled many more Republicans than Democrats for this poll, making it far more likely the result would be favorable to Akin. Akin decided to stay in the race, and without a doubt the PPP poll firmed up his resolve to fight on. On the other hand, a poll in Florida by a group previously unknown to me showed Romney up by 15%. I think it unlikely that either candidate can win Florida by more than 3%, so this one seems to be off-the-charts wrong.
Given this abbreviated history of bad polls over the last few months, it is not surprising that some on the right are convinced the fix is in with the new numbers out the last few days showing Obama with a 4-5 point lead over Romney following the Democratic National Convention. Sarah Hoyt chides the naysayers on the right who have noticed the Obama poll bump:
And then you look at the polls that caused this “sky is falling” fit. What are those polls? Those polls are, in fact, the same old sh*te. They poll all adults, which you KNOW skews Democrat. They poll registered voters, which you know skews Democrat. They poll with a skew of 4% Democrats over Republicans as though this were still 2008, as if 2010 had never happened.
Much as I am rooting for the poll numbers to turn around and for the Obama poll bump to dissipate, here are the facts to date:
1. There are four tracking polls that conduct interviews every day and average the results of 3-, 4-, or 7-day samples in their daily poll releases. The 7-day samples tend to be more stable, and the results reported each day move less than in the 3-day samples.
In each of the four tracking polls — Gallup, Reuters-Ipsos, Rand Corporation, and Rasmussen — Obama has received a bigger boost from his convention than Romney did from his. All of them now show Obama with a lead, ranging from 3.5 to 5 points.
The biggest poll bounce for Obama since the convention is in the Rasmussen survey, a poll that uses a likely voter screen, adjusts its sample to fit a model with slightly more Republicans than Democrats, and is judged by Nate Silver to have a small GOP lean this year compared to the average of all polls, even after Silver adjusts his model to take account of different results for likely-voter versus registered-voter polls. In the current Rasmussen poll released today, Obama leads by 50-45, his biggest lead in this poll since March 17. His approval rating, meanwhile, is at 52%, his highest in this poll since January 2011.
One can not know at this point if the Obama poll surge will fall off quickly, fall off slowly, or remain where it is (with a decent-sized Obama lead heading into the first debate on October 3). But it is simply wrong to say these polls show nothing more than that the fix is in and the pollsters are pulling one over on conservatives to diminish turnout. That is nonsense.
I do not take PPP polls seriously, but I think Gallup and Rasmussen are very serious polling organizations with a long track record — better in some years, worse in others (no different in that regard from Nate Silver’s elaborate models). The Rand tracking poll is a different species entirely — interviewing the same 3,500 voters once each week (500 a day) and allowing the respondents to give a percentage answer for their chances of voting (e.g 70%) and for which candidate (e.g., Obama 80%, Romney 20%). Certainly this poll does not include a sample more favorable for Obama than it did a week or two weeks ago. Reuters-Ipsos polls have been strong ones for Obama much of the year, and their tracking has been more volatile than the other polls, which have tended to be stable or move in one direction for several days at a time. I discount the results of this one a bit because of this.
2. There are some signs that the Obama surge began with Bill Clinton’s well-received speech at the convention. The terrible jobs report on Friday does not seem so far to have dimmed the boost Obama has received, though there are some indications that the surge has peaked. A few state poll numbers out today in Ohio and North Carolina by PPP are not particularly strong for Obama, and one in New Mexico shows a much tighter race (Obama up 5) than prior surveys in the state.
A week from now, my best guess is that the tracking polls will show a smaller Obama lead of perhaps 2-3 points, a bit more than he held before the conventions, but a tighter race than at this same point in 2008. I do not agree with Nate Silver that we may be seeing a decisive change in the race or that Obama’s current chances of winning are over 80%. A poll bump for several days right after a convention is normal. McCain led by 3% after the GOP convention in 2008 (and lost by 7.26%), and Silver’s model made McCain the favorite to win at that point. A week ago, Rasmussen had Romney up by 4% after the GOP convention.
The polls have been reasonably stable for most of the year and have been more volatile over the last two weeks — again, not surprising since more people are now engaged in the process. In retrospect, conservatives can rue the fact that more people did not watch the Giants-Cowboys game Wednesday night during Clinton’s speech.
If there is something to fight about, it is not the polls, but the mainstream media screen that obscures what comes out of the conventions or, for that matter, the daily campaign reports. Fewer than 15% of adults watched the biggest nights of each convention. For the rest, they received reports on what happened — online, on the evening news, on radio, and in newspapers and magazines.
For the GOP convention, the story line that came through was that Paul Ryan told a bunch of fibs and the fact-checkers nailed him. By and large, the charges against Ryan were nonsense, nothing more than talking points from a website called TalkingPointsMemo.com distributed to the media each morning. The charges against Ryan, as reported by TPM, wound up in pretty much every major newspaper. Romney’s speech was little reported. Instead, the focus of the final night of the GOP convention was on Clint Eastwood’s odd performance. The main story from the Democrats’ convention was that Bill Clinton made the greatest speech in convention history.
Is it any wonder that poll numbers after the Democrats’ convention moved the polling needle more for Obama than the Republicans’ convention did for Romney?