HOW POLLS ARE MANIPULATED
Filed under: Media, blogging — admin @ 11:14 am
by Simon Owens
During the Republican National Convention, NOW, a PBS weekly TV news magazine, posted an unscientific poll on its website asking viewers to vote on whether they thought vice presidential nominee Sarah Palin was qualified for the position. Like most polls the show posts every week, it was taken down from the front page and replaced by a new one after gathering a few thousand votes.
But in the weeks after it was removed, someone unearthed the still-present URL for the poll and linked to it at the conservative website, Free Republic. The site has become famous for sending hordes of readers to crash unscientific online polls, so much so that the act of doing so has been termed “freeping.” In this particular instance, members of the Free Republic felt that the poll showed a sign of bias, and the poster linked to it to “provide them with a result they did not expect.”
“Send this email to every non-liberal you know,” the person wrote. “Let’s get some balance into this survey group. This is the easiest vote you will ever make. It takes literally two seconds.”
Predictably, the numbers on the poll in favor of Palin began to move up, but during the freep several liberal websites got wind of it. Typical of the blogosphere, the poll became a link-fest version of tug-of-war. Close to a hundred bloggers linked to it and liberals and conservatives began forwarding email chains to their friends asking them to vote (I actually received one of these emails less than an hour before I sat down to begin writing this article).
One of the bloggers who eventually linked to the poll was PZ Myers (pictured). An associate professor of biology at the University of Minnesota-Morris, Myers is arguably the most popular atheist and science blogger on the Net. His blog, Pharyngula, is published as part of the Science Blog network (owned by Seed Media Group) and averages more than 50,000 readers a day. In recent months, he and a small group of other atheist bloggers have begun a constant and often-successful campaign to crash online unscientific polls, usually to counterbalance or push back against what they see as either anti-science or overly-dogmatic beliefs.
After Myers finds a poll dealing with religion or science on a news website, he’ll provide a link to the site along with a pithy or mocking comment. “The Edmonton Sun asks, ‘Should God be left out of the University of Alberta’s convocation speech?’” he noted in one such post recently. “I should think so. They should also leave Odin, Zeus, and the Tooth Fairy out of it, unless it’s to make a joke. Surprisingly, though, 67% of the respondents disagree with me so far. Will that have changed when I wake up in the morning, I wonder…?”
Why Poll Crash?
I spoke to the science blogger, and Myers told me that when he links to a poll he can typically swing the results by 10,000 to 20,000 votes in a particular direction. Indeed, within an hour after he linked to the Sun’s poll, the results went from 67 percent of the respondents saying “no” to 91 percent “yes.” Though he has participated in poll crashes dating back to over a year ago, he has only begun conducting them on a semi-daily basis within the last month and a half.
“It’s a very popular thing with some people because they can flex a little itty bitty muscle, and a group going there and doing something shows we have some clout, a clout in expressing an opinion,” Myers said. “There have been a couple places where the polls are so poorly done and so easily manipulated, and people go nuts; they write a script and send in hundreds of thousands of votes. Which is kind of cheating, but the whole point is that these polls are silly and useless anyway.”
The bloggers’ motivation in linking to these polls, he said, was, in essence, to delegitimize them. Because these polls are unscientific and therefore largely biased toward the demographic of the website on which they’re posted, Myers argued that poll crashing makes it harder for people to use the polls simply to reaffirm their own biases.
“For instance, if I put a poll on my blog asking whether evolution is true, everyone would say ‘yes’ with just a few outliers,” he explained. “If you put it on something like [Christian conservative group] Focus on the Family, everyone there will say ‘no.’ So the point is to show that these are highly prejudicial polls, they’re sampling unscientifically, and they’re really kind of worthless. And you can’t use those results to say anything at all. I mean, what can you say about such a poll?”
But the inaccurate data isn’t the only problem that Myers has with these polls; he also detests the poor construction of many poll questions and the limited answer choices given. It’s not uncommon for him to link to a poll while issuing the caveat that — due to the perceived inanity of the question or answers — he doesn’t know which choice his readers should pick.
In speaking to Myers, I learned that his averseness to these polls sometimes carries over to even their scientific counterparts. He argued, as have others, that media coverage of elections is much too poll-obsessed and that covering a campaign in such a way perpetuates misconceptions about why voters should choose a particular candidate.
“If you look at the major networks’ coverage of the election, for instance, what you find is that they turn it into a horse race,” he said. “All they report is who’s ahead, who’s behind and by how much. It is distracting and detracts from the coverage of the actual issues. So that’s another reason to get in there and disrupt these polls: it’s because the polls really don’t matter. You shouldn’t vote on whether someone is ahead or not. What you should be voting for is whether they have policies that you agree with.”
Measuring Enthusiasm
I spoke to a few of the people responsible for publishing polls that Myers had crashed, and surprisingly there were no bitter feelings toward bloggers who deliberately try to skewer their results. In fact, both the people I interviewed said they welcomed such online participation. They argued that instances of poll crashes allowed them to gauge the level of enthusiasm for a particular issue.
Joel Schwartzbert, the director for new media for NOW, outright rejected the notion that the poll question on the website — whether Sarah Palin was qualified to be vice president — was somehow biased or leading. When the news magazine formulates each week’s poll question, he said, it bases it on a pressing issue that has become part of the national conversation. In this particular instance, there had been a sizable amount of discussion during the Republican National Convention over Palin’s qualifications for the position.
“As an example, during the Democratic convention, we asked people if they thought the party is unified,” he told me. “So we did not pull this issue out of a vacuum, it was the most relevant and talked-about issue. When the convention ended, that poll was retired. We don’t link to old polls, nor do we have an archive of old polls. So what people did was they found that poll sort of drifting in the vast outer space of the Internet, and looking at the source code found the URL, and that’s what became viral. It did not even begin to become viral until it was formerly retired on our website.”
To date, more than 50 million votes have been registered on the poll, both from constant freeping and from bots running rampant and falsely inflating the numbers. Eventually, NOW changed the poll to track a user’s cookie so they could only vote one time per computer.
Because of this one poll, Schwartzbert said, both NOW and PBS as a whole have experienced traffic numbers that far surpassed previous viewership records by wide margins. And in attracting all that traffic, they were able to drive readers to other NOW content linked at the bottom of the Palin poll. In this respect, the poll was able to engage the online community and expose a much larger audience to more reputable and scientific information.
I asked the new media director about the unscientific nature of such polling and whether it could be misleading in displaying public opinion.
“I don’t find any online polls to be accurate enough to be worthy of public broadcast,” Schwartzbert said. “We do not announce these poll results on air. If we were going to announce them on air you can be assured that it’d be a scientific poll that’d be very official. We don’t offer up these results to measure scientifically any demographics. The point of these polls and other polls is so that people can register their vote…And the poll engine has a way to generate enough excitement to look at our investigative reports, which are still very thoroughly vetted and meticulously fact checked and very scientific.”
Schwartzbert said that people like polls in the same way that they like memes and lists, and part of using new media is understanding that “these other devices are a way to get people to come to your table. But you want to rely on your bread and butter, and, in our case, the video investigations are the meat of what we do, and what best serves our mission. So the poll is a way for people to express themselves and bring people to our larger core mission, which is to reveal what’s going on in our democracy.”
http://www.socialistunity.com/?p=3056