By Richard Morin
By Richard Morin
Cyberspace is a confusing place, made ever so much more so by the proliferation of online pseudo-polls. The consistent failure of online surveys to accurately measure public attitudes is well documented; the reasons they're so unreliable are well understood. Still, the media continue to report these ghastly little horrors as if they were real barometers of public attitudes.
Consider the recent sad performance of leading news organizations in embracing the dubious results of two online surveys conducted by ABC.com. In each case, Web site visitors were asked to respond to a question or questions prominently posted on the site. Then the clicks were counted, and the results seized upon by reporters in the throes of a long, hot, newsless summer.
In the first instance, ABC.com asked people on its Web site to indicate their choice in a hypothetical Democratic primary matchup among Vice President Gore, former New Jersey senator Bill Bradley and actor Warren Beatty. Thousands of people responded, voting early and apparently often for their favorites.
When the results of this little beauty contest were tallied, webmasters reported that Beatty was the choice of about 25 percent of those who responded, with Gore at 45 percent and Bradley claiming 29 percent of the hypothetical vote. Beatty's strong showing was so astonishing that New York Times columnist Maureen Dowd reported it straight in her nationally syndicated column, citing it as a sure sign that we've entered "nutty season" in American politics.
Well, maybe. The uncharitable might cite her use of an Internet survey as even stronger confirmatory evidence that political reporting is now far over the top. As she should have known, Internet polls are based on samples of people who don't look very much like the population as a whole. Online surveys also remain hugely susceptible to electronic ballot stuffing; few sites restrict access, so you can answer the posted question as many times as you want.
That's why no other scientific poll, before or since, has Beatty getting anywhere near 25 percent of the vote, including surveys conducted by Dowd's own news organization. Even in California, a recent survey showed Beatty with a solid 1 percent among Golden State Democrats, merely the latest real poll that showed the actor straining to break into double digits.
The bogus Beatty poll was quickly followed by an ABC.com online survey on Internet addiction that managed to generate major buzz around the world, suggesting that 5.7 percent of all Internet users were hooked on the Net. It was a finding startling enough to merit air time on "Good Morning America" as well as make the front page of the Boston Herald, where the reporter breathlessly claimed that "the computer mouse may as well be a syringe full of heroin for an estimated 11 million worldwide . . . ." Oh, my.
The ABC.com poll, which was not conducted by the ABC News polling unit, was "hopelessly flawed," says James Beniger of the University of Southern California, a past president of the American Association for Public Opinion Research.
In addition to the usual problems, Beniger notes that the Internet addiction "survey accompanied on that Web site an ABC story on Internet addiction, which thereby most likely influenced the responses of all those who completed the survey after reading that story on the very same topic." Also, "the survey appeared on only a single Web site, that of ABC News, which means that it is completely biased toward the . . . kinds of people who visit that site."
Michelle Bergman, manager of communications at ABC.com, says the Web site doesn't claim that its online offerings are "scientific polls." She seemed surprised that Dowd had picked up the Beatty result, and was unaware the "Good Morning America" had partially based a segment on its Internet addiction survey.
She says the site attempts to offer online visitors an "engaging online experience while maintaining our integrity. We certainly realize they are not scientific. We certainly do not actively promote those surveys. We make it clear to our users that they are not scientific."
At least they do now. Last Tuesday, the ABC webmasters, in response to criticism inside and outside the company, added this disclaimer to the online poll's question of the day: "Not a scientific poll; for entertainment only." "We keep evolving and perfecting our site," Bergman says.
© Copyright 1999 The Washington Post Company