RSS
The polls are not rigged — they're just nuanced
Conservative howling about supposedly nefarious surveying practices is entirely misplaced
 
Paul Brandus
Paul Brandus

There they go again. Those pesky liberals in the left-wing media showing their bias by picking on Mitt Romney.

The Romney campaign is a "rolling calamity," says former Reagan speechwriter Peggy Noonan in the Rupert Murdoch-owned Wall Street Journal.  

Mitt Romney's comments on the 47 percent of Americans who pay no federal income tax were "arrogant and stupid," huffed neoconservative Bill Kristol of the Weekly Standard. Romney "seems to have contempt not just for the Democrats who oppose him, but for tens of millions who intend to vote for him."

"How can anyone support a candidate with this kind of a vision of the country?" asked Mark McKinnon, a senior aide to President George W. Bush. "Isn't a divided America under Obama what folks on the Right rail against?"

Those lefties sure are mean. 

And what of the polls that show Romney falling even further behind? Clearly fraudulent. Participating in this conspiracy of fraud: Fox News, which says President Obama leads Romney 48 percent to 43 percent nationally. Fox says he's down by five in must-win Ohio and seven in Florida.

Damn liberal polling and left-wing media stooges.

One major complaint I get from conservatives about polling is that it is biased in favor of Democrats. The other day I tweeted the Fox poll and a new Gallup poll (the granddaddy of all polls), which currently shows President Obama leading Romney by six on a national basis.

"BS," says a Twitter follower.

A conservative saying Fox News is BS? Give me a break. Conservatives don't typically say Gallup's methodology is skewed or biased, or at least they didn't as recently as two weeks ago, when Obama and Romney were tied at 47, or in mid-August, when Romney was ahead by two points. But now that their candidate has clearly fallen behind, why of course Gallup is full of it.

Another complaint: Polling has been clearly skewed or biased against Republicans, because of "oversampling" of Democrats and independents in recent surveys. The question that should be asked is: Do these polls sample more independents and Democrats than Republicans for nefarious reasons — as some conservatives suggest — or because more people identify as independents and Democrats than Republicans?

This chart from Talking Points Memo sheds light on this question. In terms of political identification, adults, registered voters, and likely voters describe themselves this way:

·      34.5 percent call themselves independents

·      33.3 percent call themselves Democrats

·      22.1 percent call themselves Republicans

These numbers, TPM's Josh Marshall says, reflect averages from a variety of polls. Over long periods of time, the numbers smooth out statistical volatility that can occur over shorter sampling periods, and with polls that may over or undersample. So here's the question: If two-thirds of voters do not describe themselves as Republicans, is a poll oversampling when it surveys fewer Republicans — or is it accurate? Let's compare Marshall's numbers to one recent example:  

AP/GfK survey: Sept. 13-17

Affiliation

    Total Voters

    Registered Voters

    Likely Voters

Democrat

   31 percent

   32 percent

   31 percent

Independent

   29 percent

   30 percent

   30 percent

Republican

   22 percent

   25 percent

   30 percent

This sample shows that fewer voters — 22 percent — identify themselves as Republicans. Yet it also shows Republicans more likely, on a proportional basis, to be registered voters and more likely to actually vote, which is the only thing that counts. So what did the AP-GfK survey say overall?  

·      Obama 47 percent

·      Romney 46 percent

This suggests (as of mid-September) that the Republican base is more revved up than the Democratic base. Fewer voters say they're GOP, yet Romney is statistically even with Obama. This doesn't strike me as cause for celebration in the Obama camp.

But since this poll was taken, Romney's position has deteriorated, thanks largely to the "my job is not to worry about those (47 percent of the) people" comment. What had been a very close race all of a sudden has become less so, and with five weeks to go — and early voting already underway in some states — Romney has clearly lost ground. Again, even Fox News polling says this.

Many Republicans don't buy it. One reason? They claim the media is relying on 2008 turnout data, which was repudiated by the GOP's wave election two years later. The problem with this is that just as 2008 party allegiances changed quickly, so, perhaps, have 2010's. Don't take it from me. Just ask one of Fox News' biggest personalities: Chris Wallace. Here's what the host of Fox News Sunday told an unbelieving conservative radio host, Mike Gallagher, after Wallace told him that Romney's in trouble:

GALLAGHER: "What do you mean, Romney's in trouble? What, because of this poll manipulation crap?"

WALLACE: "Let me just say something. This criticism of the polls is craziness. I've done some research on this today, which is more than you've done. No self-respecting pollster in the country — including Fox, I might add — when they poll, they're trying to find out things about people and to weigh it, they will weigh how many men, how many women, how many blacks, how many Hispanics, because that is immutable. But to ask someone what your political opinion is do you consider yourself a Republican or a Democrat, that changes all the time. So they don't weight it to that. And the fact of the matter is, in 2010, when they asked what do you consider yourself, more people said they were Republicans. Now, more people are saying that they're Democrats."  

But what about one poll — Rasmussen — that Republicans often cite to show that Romney's not doing as badly as the other polls claim? Rasmussen's daily tracking poll, after all, still says Obama and Romney are neck and neck. But Rasmussen weights its polling samples to reflect what it says is a balance between Republicans and Democrats — not to reflect always changing demographic trends. Rasmussen also only polls people who have old-fashioned land lines whose numbers are published — a practice which itself is biased against young voters, few of whom identify as Republicans. And Rasmussen itself now shows Obama ahead by 2, suggesting that perhaps even Rasmussen has become part of the vast left-wing conspiracy.

 

THE WEEK'S AUDIOPHILE PODCASTS: LISTEN SMARTER

Subscribe to the Week