Lies, Damn Lies and Polls

“Pay no attention to that man behind the curtain”, the voice of the great and powerful Wizard of Oz tells Dorothy and her companions amidst the clap of thunder, clouds of smoke and frightening images of power and omniscience.

Yet all things come to an end, and the wizard’s unmasking soon follows as Toto (a dog) pulls aside a curtain, revealing the great and powerful wizard is but a humbug. “I’m a good man,” he tells Dorothy, “but a very bad wizard.”

Fox News commentator and political analyst Dick Morris has a story out that pulls aside a curtain, telling his readers not to pay any attention to a slew of recent battleground state polls that show President Barack Obama leading challenger Gov. Mitt Romney. His story “Swing State Polls Are Rigged”  states that:

From noted Republican pollster John McLaughlin comes a clear and convincing exposé of the bias of media polls in the swing states of Florida, Ohio, and Virginia.

Citing recent polls in Virginia, Florida and Ohio, Mr. Morris explains the pollsters’ error (bias?)

McLaughlin reviewed exit polls in each state for the past four elections. From this data about who actually voted, he found that the party divisions manifest on election day have little to do with the samples upon which the media is basing its polling. And, coincidentally, it is always the Republican vote that tends to be undercounted. … Things are no better in Ohio. Here, McLaughlin finds a 2 point Democratic edge in the past four elections (38% Dem, 36% Rep). But the media polls show vastly more Democrats and fewer Republicans in their samples:

9-26: CBS/NY Times = 35% Dem / 26% Rep; 9-23: Wash Post = 35% Dem / 27% Rep; 9-11: NBC/Wall St Journal = 38% Dem / 28% Rep

Interesting stuff this, but why, you might ask, is this a GetReligion story? Because, the man conducting the NBC/Wall Street Journal polls, Dr. Lee Miringoff, director of the Marist Institute for Public Opinion, along with his colleague Dr. Barbara Carvalho gave a briefing to the lunchtime session of the Religion Newswriters Association meeting.

This is my first RNA meeting and to my surprise I have enjoyed it greatly. My experience of mass gatherings of reporters has been filtered through my work for English and other overseas newspapers. British press “culture” is very different from that found at the RNA. More yoghurt, less alcohol — there is a reason why they call Fleet Street reporters “hacks”.

Meeting at a hotel in suburban Washington, over a hundred religion reporters (among whom are all the GetReligion team members) are participating in an assortment of briefings, seminar discussions and presentations on religion and the media. In a fascinating talk sponsored by the Knights of Columbus, Drs. Carvalho and Miringoff presented some of their research on the current campaign. Among their findings was a disgust for the negative politicking among the electorate, a sense of alienation between the issues the candidates are discussing and issues of concern to voters, along with data showing that attempts to describe the moral issues (abortion, gay marriage, etc.) in black and white terms — forcing the answer to are you against or for abortion to be a yes or no — does not faithfully paint a true picture of America.

The presentation, which was recorded by C-Span along with a panel discussion entitled “Whose First Amendment?”, are well worth watching for you political junkies out there. And it will also give you a glimpse of me asking Dr. Miringoff to respond to Dick Morris’ criticism of his polling.

When I was asking the question, Dr. Miringoff began to smile at me. Now I am a likeable fellow, all told, but his smile was not of bonhomie nor of Freemasons exchanging secret signs, but that of a cat who has a canary in his sights. Dr. Miringoff pounced on Dick Morris’ story — see the video for his full comments unfiltered by my note taking skills — and argued that 1) this was sour grapes on the part of a Republican strategist whose man was lagging in the polls. In 2004 the Democrats charged the polls were under reporting their voters — and now it was the Republicans’ turn. 2) Party identification is not a fixed component akin to age, race, or economic status as party affiliation changes over time and also at the whim of the voter. Comparing the exit poll data from discrete points of time with something as amorphous as party identification was not as straight forward as Dick Morris had claimed. And 3) new polls will be released this Sunday in the wake of the first presidential debate that will likely cause the Democrats to complain.

Dr. Miringoff and Dr. Carvalho discussed the methodology of their polling, but also spoke to some of the unique circumstances pollsters are facing in this election. Youth turnout is likely to decline relative to the 2008 election. Will the absence of young people at the ballot box next month spell defeat for Barack Obama? One can speculate, but Dr. Miringoff suggested it would be wise for the president if he spent some more time on college campuses.

So, should we pay no attention to the man behind the curtain as Dick Morris suggests, or are the polls showing the president leading in the battleground states a fair approximation of voter sentiment a month out from election day? What say you, GetReligion readers?



Browse Our Archives

Follow Us!

What Are Your Thoughts?leave a comment
  • Jerry

    The first comment I have about the polls is economic. If a pollster wants to get an account measuring what we think of Chevy or Coke, the need to be able to show that their results are accurate. Being way off in a political poll is totally against their economic self-interest. That does not mean that a pollster will get it right, of course, just that if they fail and fail badly their bottom line will be impacted. Too many forget that polling is a business (at least for most polls).

    The second point is that of course polling is an uncertain business. There can be “house effects” as Nate Silver has pointed out – polling organizations whose result tilts Democratic or Republican compared to the final outcome. Then there’s the issue of registered versus likely voters, the impact of ignoring cell-phone only voters. And there is the question of the sample size and a host of other factors.

    But when there’s a consensus amongst pollsters using different methods, then the odds of that being the correct result are very high.

    • Charles

      Are not their customers the very liberal-leaning media institutions with a stake in manipulating the enthusiasm levels of voters and donors? The bottom line is the price of complicity.

      • Charles, there probably are “very liberal-leaning media institutions with a stake in manipulating the enthusiasm levels of voters and donors”. But then there’s Fox News, which might meet all of those criteria, except the word ‘liberal’.

        Two sites for people who are interested in the polls: and The first leans liberal, and the second conservative. But they both buy their polling data from the various pollsters and are open about how they process that data. (They’re also usually in close agreement regarding the actual numbers.)

  • deiseach

    Re: the habits of Fleet Street’s finest, I refer you to the Wikipedia article on “Private Eye” magazine:

    “Lunchtime O’Booze has been among the magazine’s resident journalists since the early days. The name is a comment on journalists’ supposed traditional fondness for alcohol, their prandial habits, and the suspicion that they pick up many of their stories in public houses.”

  • Darren Blair

    When I was doing my undergraduate work, I had to take a course on statistics and another on marketing research.

    One of the things made abundantly clear by both classes is just how easy it is for surveys to wind up skewed.
    Name a portion in the process from start to finish, and there’s a chance for someone to introduce bias or even error… usually without meaning to do so.
    Everything from “poor wording on the questions” (which can lead people to provide answers they might not otherwise provide) to “poor analysis of the final numbers” (should be self-explanatory) can throw off an entire survey.
    This is why, for best results, you don’t just rely on individual surveys by themselves if you can help it.

  • The Old Bill

    Often, political pollsters keep two sets of books: one for public consumption and another for their clients’ eyes only. The first is designed to do no harm to the client and to push his interests. The second is to show the client how he’s doing and his strengths and weaknesses. Reporters should be wary.

    Another factor is how manyo are unwilling to answer a poll and who they are. We just don’t know.

  • MP Smith

    And this has what to do with faith and religion?

    • geoconger

      Remember this blog is not about faith and religion. It is about reporting on faith and religion. The blog’s subject was the Marist polling session at the 2012 Religion Newswriters Association meeting and falls within these parameters.