Commentary

Panel Poseurs

Panel spammers, serial panelists, professional respondents: Whatever you call them, individuals who use online panels to earn spare cash are compromising online research results.

"Online research is barely 9 years old, and we're seeing a data-quality issue," says Ali Moiz, COO of Peanut Labs, a San Francisco-based provider of online samples. Moiz cites a 2005 comScore study indicating that one-quarter of 1 percent of panelists account for 30 percent of all surveys taken. Equally alarming: Online surveys have a 21 percent mean range of error, according to a 2007 study done by Knowledge Networks.

Moiz thinks Peanut Labs has a way to control serial panelists. In late June, the company plans to roll out OptimusID, a program that "fingerprints" individual computers without using cookies or downloads. Research firms can use those fingerprints to identify professional panelists, and then prevent them from taking surveys.

Fingerprinting might give a much-needed boot to panel spammers. But researchers say they've got bigger worries. "‘Are people who they say they are?' is a bigger issue," says John Ouren, executive vice president of panels and communities for MarketTools, a San Francisco-based market research firm.

Hard-core serial panelists commit fraud, pretending they're physicians, for instance, to cash in on the significant panel premiums offered to that select group. Garden-variety panelists misrepresent their demo group for the same reason.

"I may be an unemployed 45-year-old female, but if I represent myself as a 25-year-old Hispanic, I'll be eligible for more studies," Ouren explains. Fingerprinting, Ouren admits, is useful; indeed, MarketTools was a beta tester for OptimusID. But it won't catch impostors, he adds.

And, by the way, Ouren doesn't buy research that indicates one-quarter of one percent of panelists account for 30 percent of survey respondents. If it is that high, don't blame the panelists: Those researchers "aren't doing basic list hygiene that a direct marketer would do," he says.
Next story loading loading..