Farah’s Dishonest Handling of Polling Data


Over the last couple years, the Worldnutdaily has contracted with Wenzel Strategies to do polling, as many conservative groups do. And now he’s promoting a new poll by Wenzel that purports to show what Muslim-Americans really believe — naturally, it’s what Farah thinks they believe. But he wants to establish the poll’s credibility by emphasizing that it’s “scientific.”

Before the presidential election, I got the idea to commission a scientific poll to find out what Muslim-Americans really think – and how it might affect their votes.

The results were eye-opening and alarming. They should be to every American who believes in the U.S. Constitution and Judeo-Christian morality.

And he calls Fritz Wenzel a “respected pollster.” Well let’s check the track record. Wenzel is a Republican polling company. Now that doesn’t necessarily mean his results are bad, but we can easily find out by comparing their polls to actual results. And they aren’t even close. For example, when all the other polling companies were showing Todd Akin being down by as much as ten points, Wenzel’s poll had him up by as much as five points, and showed him with a consistent lead.

Wenzel also was about the only company that showed Romney with a three point lead in Ohio just days before the election (Obama won by 2), and even had him pulling very close in Wisconsin (Obama won by 7). They also had Romney winning in Virginia just prior to the election (Obama won by 3). They were just as inaccurate in the Senate races, and all tilted toward the Republicans. They had Josh Mandel leading Sherrod Brown by five points and Tommy Thompson leading Tammy Baldwin by two points.

And Wenzel was one of the many conservatives arguing before the election that all the other pollsters were getting it wrong because they were oversampling Democrats. Here’s Fritz Wenzel in September explaining why his polls are accurate and all the others were wrong:

There’s a lot of noice out there right now regarding the polling of the race for President of the United States. Some polls show Democrat Barack Obama up by a few points, while others show it is a really close race. Almost no polls show Republican Mitt Romney with a meaningful lead. That does not necessarily reflect what is really going on.

The reasoon Obama seems to enjoy a lead over Romney is straight-forward, and by now, well-explained: it is because the survey sample includes more Democrats than Republicans.

Simple enough.

Yep, simple enough. And absolutely wrong. So once he was proven wrong on this, did he admit that his sample was wrong? Nope. He blamed it on stupid people who just didn’t realize that Obama was evil and Romney was better:

Pollsters and consultants across the country whose interpretations of the polling numbers were off the mark were expecting voters to be smarter than they apparently are. They expected voters to look at Romney’s weaknesses and strengths, and judge them against Obama’s weaknesses and strengths…

It was H.L. Mencken who once said that “nobody ever went broke underestimating the intelligence of the American public,” and this appears to have been the guiding principle to the Obama re-election campaign.

It worked.

It wasn’t his fault! It was all those dumb people who voted for Obama! He just has to put more dumb people into his polling sample! Alex Koppelman pointed out back in 2009, when Wenzel was putting out polls supporting the Worldnutdaily’s obsession with the birther issue, the kind of tactics they use in framing questions to reach a desired result:

Getting a result that favored WND’s position on the issue of whether Obama should release a long-form birth certificate (despite the fact that he’s already released a copy of his birth certificate and Hawaiian officials have said he was born there) involved a little sleight of hand.

The result trumpeted by WND is that 41.5 percent of respondents said “Obama should release all records, including long-form birth certificate,” essentially the “yes” answer WND was looking for. An additional 7.8 percent said they “are troubled by these questions,” which the site has been lumping in with the yes answer. The “no” answer, though, they split up into five different responses — “I am not concerned,” ”questions not valid,” “Obama has met requirements,” “Obama has answered all questions” and “requirements outdated — should be ignored.”

It’s a neat trick, and a fairly common one; by doing that, you can avoid providing a real reflection of the size of your opposition. In this case, the total of those five answers adds up to a majority of respondents, 50.6 percent.

But it isn’t just that Wenzel is a bad pollster who skews his polls to match what his paymaster wants, it’s also that Farah is being incredibly dishonest in his handling of that bad data. Here’s how he portrays the results of the poll on Muslims:

Nearly half of those polled, a scientific sampling of 600, said parodies of Muhammad should be criminally prosecuted with offenders given punishments as severe as the death penalty;

40 percent of Muslims believe they should not be judged by U.S. law and the Constitution, but by Shariah standards and Islamic judges;

Only 30 percent believe Christians have a right to evangelize Muslims;

20 percent say Muslim men should be allowed to have more than one wife;

58 percent said any criticism of their religion should not be allowed under the Constitution – even while nine of 10 said they support the First Amendment;

One in three said Israel has no right to exist or were uncertain about it;

72 percent said they would definitely be voting for Barack Obama, while another 8.5 percent were leaning in that direction;

Only 11 percent were supporting Mitt Romney at the time – just days before the November vote;

About 55 percent said the U.S. is generally on the right track.

But look at how he tries to establish the credibility of this study:

Now, maybe you question the authenticity or reliability of this survey. Maybe you think it was biased. Maybe you think the sampling of Muslims, 98 percent of whom have U.S. citizenship, wasn’t large enough.

Well, shortly after WND commissioned this survey by respected pollster Fritz Wenzel, another similar survey was commissioned by the Council on American-Islamic Relations, which bills itself as a Muslim civil-rights organization but which in fact is a Muslim Brotherhood front group.

The results were strikingly similar – at least on the questions asked in both surveys. This CAIR survey, conducted on a smaller sampling of 500, received extensive coverage by the media.

The CAIR survey found:

68 percent of Muslim-Americans planned to vote for vote for Obama with 7 percent going for Romney;
66 percent said they identify with the Democratic Party, while only 9 percent affiliated as Republicans.

In other words, the CAIR survey validates the Wenzel poll – even though it didn’t seek answers on some of the more troubling issues of the law and constitutional liberties.

Yes, he’s that dishonest. But like so many of Wenzel’s election-related polls, these results are in stark contrast to the results of surveys taken by non-partisan and non-ideological groups. A major poll by the Pew Research Center, for example, found that American Muslims are predominately moderate, mainstream and well assimilated into American culture, and that levels of religious and political extremism are quite low.

This is why you should not trust polls with a clear ideological purpose — and you should never trust the Worldnutdaily when they say almost anything.

Comments

  1. eric says

    Pollsters and consultants across the country whose interpretations of the polling numbers were off the mark were expecting voters to be smarter than they apparently are.

    Which is a complete nonsequitur and makes no sense whatsoever as an excuse. The relevant questions are: did you ask a representative group, did you ask questions that are good proxies for actual voting behavior, and did those polled tell you the truth. (If I had to guess, I’d say his polls went: no, maybe, yes)

    Ironically, I would expect that “stupid people” would have a more difficult time skewing a poll. He’s saying that the people he sampled were very very stupid….yet as the data analyst, he was not smart enough to figure out how they were lying/misrepresenting their vote. What does that say about him, hmmm?

  2. Hercules Grytpype-Thynne says

    The reasoon Obama seems to enjoy a lead over Romney is straight-forward, and by now, well-explained: it is because the survey sample includes more Democrats than Republicans.

    Simple enough.

    Yep, simple enough. And absolutely wrong.

    Actually he was absolutely right: the survey sample did include more Democrats than Republicans. But then again, so did most pollsters’ estimate of likely voter turnout, and so did the actual turnout on Election Day. Where Wenzel went wrong was in assuming that including more Democrats than Republicans in the survey sample was the wrong thing to do.

  3. slc1 says

    One thing’s for sure, Nate Silver did a hell of a lot better then Wenzel in predicting the state by state results (50 out of 51 with the 1, Florida, where he slightly favored Rmoney, actually being too close to call).

  4. baal says

    The smart / dumb comment has an almost reasonable explanation. Pollsters can’t rely on “who will you vote for” and “how likely are you to vote” questions entirely. Peoples subjective assessments on the clear questions don’t always track closely to behavior. If you add in a bunch of other questions that show how hard or soft your support is you get a better idea. If you ask for other ballot items as well, you get even more accurate – a moderately strong Obama supporter who is off the charts for marriage equality is very likely to make it to the polls despite hardships would be weighted more than a strong Obama supporter who doesn’t care about the equality vote, for example.

    The place where the (R) pollsters fall down is that they ask a bunch of right wing talking points and noise drowns out the signal when you’re gauging strength of support because the (R) are trained to take extreme positions consistently. It’s also entirely possible that the (R) pollsters are so inside the fake reality bubble that it’s not possible for them to think clearly enough to get the right type of questions and then get passed the problem of not wording the questions in a highly biased way.

  5. thisisaturingtest says

    It’s blatant dishonesty, too; he’s not even pretending not to cherrypick the survey results. “I’ll just ignore that the questions that are the heart of my argument aren’t even addressed in the CAIR survey, and claim that the similarity of incidence in one unsurprising response validates my survey.”
    This is the kind of thing that shouldn’t fool anyone with the critical thinking skills of a six-year-old child; but every time I say that to myself, I remember who Farah’s chosen audience is, and think, “oh, right.”

  6. Sastra says

    The Worldnutdaily should poll its readers:

    Do you believe that parodies of Jesus and insults directed at Christian faith should be criminally prosecuted?

    Does the U.S. law and the Constitution outrank Biblical standards and those who advocate the will of God?

    Do Muslims have a right to evangelize Christians?

    Are criticisms of Christianity allowed under the Constitution?

    And be sure to ask these questions right after one of those persecution articles. So it will be fair.

  7. says

    “Pollsters and consultants across the country whose interpretations of the polling numbers were off the mark were expecting voters to be smarter than they apparently are.”

    Translation: They were letting their political biases influence their interpretation of the polling data. It’s not their fault!

  8. says

    @2: “Where Wenzel went wrong was in assuming that including more Democrats than Republicans in the survey sample was the wrong thing to do”.

    Most surveys don’t “include” more Ds than Rs, or anything like that, they just do a random sample. If you get more Ds than Rs, that means there are more Ds than Rs out in the public.

    Of course, this is predicated on having a truly random sample. If you know that you can’t get such a sample (for example, you can’t call cell phones), then you might weight the poll according to party ID, if you have some independent means for knowing what that is. But it’s much more common to weight for age, race, or other demographic qualities for which you can use reliable census data.

    Either way, in the end, Wenzel and the rest went wrong by assuming that their side had more support than it really did, and when the polls showed otherwise, rather than accept what the evidence was saying, they chose to interpret the evidence to fit their preexisting conclusion. This seems to happen a lot with these folks.

  9. raven says

    New Pew Research Center Survey Finds Moderate Attitudes Among Muslim Americans
    No Signs of Growth in Alienation or Support for Extremism

    Washington, D.C. — As the 10th anniversary of the 9/11 attacks approaches, a comprehensive public opinion survey by the Pew Research Center finds no indication of increased alienation or anger among Muslim Americans in response to concerns about home-grown Islamic terrorists, controversies about the building of mosques and other pressures on this high-profile minority group in recent years. Nor does the new polling provide any evidence of rising support for Islamic extremism among Muslim Americans.

    On the contrary, as found in the Pew Research Center’s 2007 survey, Muslims in the United States continue to reject extremism by much larger margins than most other Muslim publics around the world, and many express concern about the possible rise of Islamic extremism.

    Very few Muslim Americans – just 1% – say that suicide bombing and other forms of violence against civilian targets are often justified to defend Islam from its enemies; an additional 7% say suicide bombings are sometimes justified in these circumstances.

    Fully 81% say that suicide bombing and other forms of violence against civilians are never justified. Comparably small percentages of Muslim Americans express favorable views of al Qaeda, and the current poll finds more holding very unfavorable views of al Qaeda now than in 2007.

    According to real polls, most American Moslems are moderates.

    That is not surprising. Many of them are here because their own societies are boring, going nowhere, or have flat out failed i.e. Somalia and Afghanistan.

  10. gopiballava says

    “40 percent of Muslims believe they should not be judged by U.S. law and the Constitution, but by Shariah standards and Islamic judges”

    I would say that Shariah and Islamic judges should be the ones to judge how well a Muslim adhered to the rules during Ramadan, for example. If somebody wants their wife to be judged and penalized according to Islamic rules for being out in public alone, now that’s a bigger problem. But this question doesn’t go there. I’d expect that most Christians expect their adherence to Christianity to be judged based on the Bible rather than the constitution and, again, that’s perfectly reasonable.

  11. raven says

    Pewforum 2006:

    For example, six-in-ten white evangelical Protestants say that the Bible should be the guiding principle in making laws when it conflicts with the will of the people, a view rejected by an equally large majority of Americans, including most Catholics and white mainline Protestants.

    If you look at the views of the American Moslem extremism equivalent, the fundie xians, they aren’t much different from extremist Moslems.

    60% of American evangelicals are in favor of biblical law. Biblical law is just as bad as Sharia law and probably a lot worse.

    There are ca. 20 death penalty offenses for mostly trivial reasons, disobedient children, nonvirgin brides, sabbath breakers, false prophets, atheists etc. It is estimated that under biblical law, 99% of the US population would end up dead under a pile of rocks.

    The best one can say is that at least 99% of all fundie xians would also end up dead under pile of rocks. Maybe we should legalize biblical law but make it voluntary. The fundies could have a ball stoning each other to death and we would be rid of a serious social problem.

  12. says

    “Pollsters and consultants across the country whose interpretations of the polling numbers were off the mark were expecting voters to be smarter than they apparently are.”

    This idiotic statement is not only insulting to voters, it demonstrates how stupid the residents of the bubble have been. They looked at the data and told themselves, “that can’t be right. A majority of Americans have to see Obama as an illegitimate Muslim radical Marxist foreigner. Don’t they watch Fox?” And the just fudged the numbers.

  13. Hercules Grytpype-Thynne says

    @9:

    The pollster then weights the various demographic categories to make sure there are enough men, enough retirees, enough Democrats, enough Catholics, enough college graduates, and so on.

    From Electoral-Vote.com

  14. Michael Heath says

    Sastra writes:

    The Worldnutdaily should poll its readers:
    […]
    Does the U.S. law and the Constitution outrank Biblical standards and those who advocate the will of God?

    Where the most popular answer would be:
    The Bible and the Constitution are in perfect harmony and always have been.

    Except in the confederacy where the answer would be:
    The Bible and the Constitution are in perfect harmony and always have been; except for the 13th – 15th Amendments – all three are ill-gotten gains during the war of Northern Aggression.

    A distant third would be the above answer with objections towards the 16th (income tax), 17th (Senators elected by popular vote), 19th (women’s suffrage), 21st (repeal of 18th – prohibition of liquor), and 24th (prohibition on poll taxes).

  15. John Hinkle says

    @Sastra:

    Do you believe that parodies of Jesus and insults directed at Christian faith should be criminally prosecuted?

    Does the U.S. law and the Constitution outrank Biblical standards and those who advocate the will of God?

    Do Muslims have a right to evangelize Christians?

    Are criticisms of Christianity allowed under the Constitution?

    To make your questions more like a Wenzel poll, I’ve modified them as follows:

    Do you believe that parodies of Jesus and insults directed at Christian faith should be criminally prosecuted, or should only Muslims have that privilege as they do now, you poor persecuted Christian you?

    Does the U.S. law and the Constitution outrank Biblical standards and those who advocate the will of God, or do only Muslims get that privilege with their holy “koran” , you poor persecuted Christian you?

    Do Muslims have a right to evangelize Christians, or do you feel that Muslimness is a political ideology and not a religion at all so they should just shut their mouths and go back to Mexico?

    Are criticisms of Christianity allowed under the Constitution (unlike criticisms of Muslimity, which are disallowed under the Constitution so WTF is up with that?)?

  16. says

    @14:

    Unlike race, gender or age, all demographic traits for which pollsters weight their samples, party identification is considered an attitude that pollsters say they should be measuring. When party identification numbers change, it’s an indication of deeper political change that a poll can spot.

    “If a pollster weights by party ID, they are substituting their own judgment as to what the electorate is going to look like. It’s not scientific,” said Doug Schwartz, the director of the Quinnipiac University Polling Institute, which doesn’t weight its surveys by party identification.

    http://www.nationaljournal.com/2012-presidential-campaign/republicans-to-pollsters-too-many-democrats-in-your-surveys-20120925

  17. ricko says

    “Wenzel also was about the only company that showed Romney with a three point lead in Ohio just days before the election (Obama won by 2), and even had him pulling very close in Wisconsin (Obama won by 7).”

    Yeah, Obama won by more than 7% but a little less than 8% in Wisconsin.

    Perfect polling by Wenzel.

  18. Ichthyic says

    to me it’s simple.

    Silverman (amongst others – he didn’t really invent the models) uses models that are proven to work.

    any polling organization that chooses other models should be able to similarly show a proven track record.

    your track record is abysmal?

    then, you fail, and people should laugh at you, if not at least stop funding you.

    simple.

    if anything else occurs, there is corruption afoot.

  19. Ichthyic says

    There are ca. 20 death penalty offenses for mostly trivial reasons, disobedient children, nonvirgin brides, sabbath breakers, false prophets, atheists etc. It is estimated that under biblical law, 99% of the US population would end up dead under a pile of rocks.

    That’s it!

    Evangelical Christianity is a Muslim Plot to destroy America!

  20. Ichthyic says

    …Obama is a secret Evangelical Christian, out to employ his Muslim Master plan to destroy America!

    I can haz TV show?

  21. Childermass says

    slc1@4 : “One thing’s for sure, Nate Silver did a hell of a lot better then Wenzel in predicting the state by state results (50 out of 51 with the 1, Florida, where he slightly favored Rmoney, actually being too close to call).”

    Actually, Mr. Silver switched to an extremely slight Obama win for Florida a day or two before election day. So he really did call all 50 states + DC.

    FiveThirtyEight still has his last projection up, so you can verify it. For my prediction the day before the election, I took Silver’s prediction and reversed him on Florida to say Gov. Romney would win it.

    (Of course, Silver was more than a little bit lucky as his own model suggested that what he called would probably not be right every time. After all Florida was very close to a coin flip in his model and several states had non-trivial chances of going either way in his model.)

  22. slc1 says

    Re Cildermass @ #24

    Even at 50.3% to 49.7% in favor of Romney, which was the last Silver numbers I recall, that’s in the noise level and essentially says that the race is almost even steven.

Leave a Reply