OPINION

In recent years, Web polls have spread like a fungus across the Canadian online landscape, asking questions ranging from the banal to the blatantly biased.The polls,which suggest readers get involved, run on the same pages as news content. The poll results contain little explanation of what the numbers actually mean — which, according to academics and professional pollsters, is very little.

Some media outlets running the sites say the polls are a fun, interactive way for users to feel involved. It’s clear, however, that some people are taking their results very seriously.

So what does a Web-based poll tell us? Much more about the Web site’s users than about public opinion, say the experts.

Mike Colledge, senior vice-president with Ipsos- Reid in Ottawa, says a typical Web poll is about “as valid as any self-complete survey that you don’t control the sample on.”Put a ballot in any printed magazine — Media, perhaps — or paper and you have much the same thing.

Colledge says the Web,when properly harnessed, is proving itself a powerful tool for gathering public opinion.

“The gap between people who are online and people who aren’t online has closed considerably over the last four or five years,”he says, explaining that more and more people of all ages, income levels and sexes are going online. Concerns about surveys becoming skewed by the opinions of the young, the wealthy and the male are waning.

If participants are picked at random and given passwords, self-selection and vote rigging by small groups becomes less of a problem. Still, even rigorously conducted online surveys by professional pollsters remain less accurate than random-digit, telephone surveys, the pollster’s tool of choice.

And Colledge cautions that the point-and-click surveys done at most media sites do not even attempt to correct for the self-selection and multiple voting that true online polls try so hard to avoid. The simple fact that votes are tied to a particular Web site means they can never be truly reflective of public opinion, he says, since “we don’t know the general population went to that site.”At best they
give only the vaguest hint of what visitors to a particular site think. Frequently, however, the Web site can’t even do this.

According to the National Post Online, 64 percent of respondents last year felt “assimilation” should be “a priority for Canada’s policy on native peoples as a way of dealing with its First Nations.”

Meanwhile at globeandmail.com, a slim majority of participants in its Feb. 11, 2002 poll, some 51 percent agreed the United States was moving towards “a new crusade against Islam.”

Does this mean most Post online users really believe Canada should erase the identity of its native peoples by absorbing them into the general population? Does more than 50 percent of globeandmail.com’s audience really think George
W. Bush is itching to drive Muslims from the Holy Land? Nobody knows. The beauty of Web polls is that nothing is ever for certain.

Since respondents are self-selected,and since the vast majority of the polls allow people to vote as many times as they want, small,vocal minorities — even minorities of one — can easily take control.

Colledge tells the story of Chris Nandor, a programmer and fan of Boston Red Sox shortstop Nomar Garciaparra. In 1999, Nandor voted online for Garciaparra 39,000 times in an effort to ensure his American League hero was elected to play in the all-star game.

Nandor, then 25, did much of his voting with a computer program.

Hacking into the all-star ballot “definitely happens all the time,” he is reported to have said after he was caught. “I can’t give you a definite percentage, but there has never been a Web-based poll that people haven’t done this. It’s pretty much
standard procedure.”

Of course, one need not write a computer program, or even vote twice, to skew an online poll. Pointing a couple of thousand of your friends in the right direction is often enough.

Take,for example,the fate of the Modern Library readers’list of the top-100 novels of the 20th century. Created by the Random House division via an online poll conducted between July and October 1998, the list was meant as counterweight for a similar registry of the century’s top-100 novels compiled by Modern Library’s board of professional writers and critics.

But in leaving the readers’ list up to the whims of those surfing its Web site, the publisher got more than it bargained for. The two lists aren’t only reflective of differing points of view, but appeared to be from different planets.

A comparison of the top-10 novels of each list tells a startling tale.While the board’s first 10 choices — works by the likes of Joyce, Fitzgerald, Nabokov
and Huxley — are no big surprise, seven of the readers’ first 10 picks are by alleged cultists: objectivist philosopher Ayn Rand claimed four of the top spots, including the first and second, while scientology guru L.Ron Hubbard took three more. Neither Rand nor Hubbard is to be found anywhere on the board’s top-100.

Objectivists and scientologists probably don’t make up a majority ofModern Library’s readers. But they almost certainly make up a very loud minority.

As David Ebershoff, the division’s publishing director, carefully puts it, “the people who were drawn to go to the Modern Library Web site and compelled to vote have a certain enthusiasm about books and their favourite books that many people don’t, so that the voting population is skewed.”

The inability of a typical Web poll to capture a meaningful picture of the true opinion of a group of people using a Web site shouldn’t come as a surprise, according to Heather Pyman, research director of the Carleton University Survey Centre. The constituency of a Web-based poll is simply “not a representative sample of anything, and it cannot be if it’s self-selected,” she says.

The most bizarre example of online voteskewing concerns Time magazine’s poll for person of the century. Though the magazine eventually declared Albert Einstein to have embodied “the very best of this century as well as our highest
hopes for the next,” its online poll told a different story.

Until the magazine separated votes into five categories, it wasn’t Winston Churchill, Vladimir Lenin or even Adolf Hitler that respondents picked
as the person who had the greatest impact on the 20th century.No, thanks to a carefully orchestrated campaign, Mustafa Kemal Ataturk, autocratic founder of modern Turkey,was the one garnering the most votes — until a similar campaign for Winston Churchill helped push Ataturk out of first place in the magazine’s “Leaders and Revolutionaries” category. Churchill’s win was narrow indeed, however, and he took the top spot on the user’s poll by just 0.65 of a percentage point.

Maybe the frustrations of the major leagues, Modern Library and Time magazine have been enough to discourage American news sites from posting more Web polls. The surveys are certainly much less ubiquitous south of the border.And when they do appear, they seem more likely to come with warnings about how little they tell us.

CNN is one of the few large, mainstream American sites to prominently feature a Web poll. Results from its “Quick Vote” are always accompanied by warning that the survey “is not scientific and reflects the opinions of only those Internet users who have chosen to participate.The results cannot be assumed to represent the opinions of Internet users in general, nor the public as a whole.” This is something entirely absent from National Post Online, globeandmail.comand Canoe, the Sun newspaper chain’s main Web site.On these sites, results are simply posted without any explanation at all.

Norman Bradburn, the assistant director for social, behavioural and economic science at the National Science Foundation in Washington,D.C., has an acronym for surveys of the type used by media Web sites and radio phone-in shows (he
believes the two are equally inaccurate): SLOPs, or self-selected listener opinion polls.

Bradburn, who once headed the National Opinion Research Centre, says he fears that too often the general public doesn’t differentiate between SLOPs and legitimate polls.

“What happens is that advocacy groups can take them up and then purvey them as if they were real statements of fact about the world and use that to try to influence legislators or decision makers,”he says.

Asked about the unscientific nature of Web polls, Nancy DeHart, editorial director of the National Post Online when this article was written, says its “Post Vote”isn’t meant to be taken seriously. “It’s not scientific, nor do we claim it is,” says
DeHart about the survey.

“It’s a fun, interactive poll.”

“Our paper is always trying to stir up opinion,” she continues.“We just think an Internet poll is a great way to kind of tap into that.”

The backbone for Post Vote is provided by Vote.com, a Web site created by Bill Clinton’s former pollster, Dick Morris. Whereas many American news Web sites avoid Web polls,Morris has made them Vote.com’s raison d’être. Politics and policy questions are the site’s bread and butter,but the site also features movies, travel, even questions on gay issues.

Vote.com was started in 1999 and presently has about 1,810,000 registered users in the U.S. and 25,000 in Canada. It also has users in Japan, the
UK, Australia and France.While Morris readily admits the site’s polls may not be statistically representative of anything — and he prefers the term “direct voter referenda” to the p-word in recognition of this — he has no qualms about
sending the results to decision makers involved with the issues being considered.

“Our world needs a massive dose of direct democracy,” he says, because “within each of our nations, the special interests have developed a
stranglehold over our politics.”

If Web surveys really did free politics of special interest groups, there might be some redemption for them.But as the Post Online learned earlier this year, the polls can just as easily become a tool of special interests as a cure for them.

On March 2, 2002, the National Post ran a column condemning its own Web site’s Feb. 27 survey — in which 87 percent of respondents voted in favour of ratifying the Kyoto Protocol, in a poll the Post had declared it would send to
federal environment minister David Anderson.

“Before [Anderson] starts waving the results around as an endorsement, he should know — and we should all know — that the poll tells us nothing about the views of Canadians or Post readers on Kyoto but a lot about the tricky morality of Green activists,” wrote Post opinion writer Terence Corcoran, explaining that until the David Suzuki Foundation sent an e-mail to its supporters urging them to “let your voice be heard,” the survey had only been running two-to-one in favour of ratification.The column went on to attack the Suzuki Foundation,“the World’s
Greatest Global Scare Machine,” for rigging the Post Vote.

Would the Post have run a column attacking the results if they had been 87 percent for tossing the protocol in the dustbin?

Let’s vote on it .