Fact-checking articles, newscasts and speeches has become an online cottage industry. But what about public opinion polls? Splashy majority opinion numbers create news and high-traffic headlines, but how do we know when, or if, the poll results are accurate?

PollCheck, a new and unique project by iMediaEthics, conducts its own polls to fact-check the accuracy of headlined polls and their widely quoted media poll numbers.

Increasingly, various polling organizations claim percentages of public opinion that conflict with each other, or present results that seem to defy common sense. Through the strict use of approved polling and scientific methods, PollCheck will examine conflicting or odd polling conclusions to produce a more realistic picture of what the public thinks.

iMediaEthics is a not-for-profit, nonpartisan media ethics news website that has been holding the media accountable since its launch (then as StinkyJournalism.org) in 2004. By examining concrete, measurable errors of fact and ethical breaches regularly encountered in the press, iMediaEthics has established itself among the top 20 most-trafficked media watchdog websites, according to Alexa.

PollCheck will be monitored by David W. Moore, Ph.D., a veteran Gallup pollster who writes the popular bi-monthly “PollSkeptic Report” for iMediaEthics. “Polls can be an important part of the democratic process,” he writes, “but it’s necessary to determine when they are giving us a realistic picture of the public, and when they are not.”

PollCheck fact-checks polls by examining the impact of question wording, with specific attention to whether polls measure the percentage of people who have no meaningful opinion on the issue.

“This is an important element in understanding what the public thinks that typically the major pollsters gloss over,” says Moore.

Moore will compare original question wording with a more objective wording he says will be “as neutral as possible, and will take into account both non-opinion and opinion intensity. Typically, when polling results are suspect, it’s because the polling organizations have fed respondents biased information, pressured respondents to produce ‘opinions’ they don’t have, and/or ignored the intensity with which respondents hold their views.”

PollCheck will address these issues by conducting its own iMediaEthics polls. Typically, the question wording of the media polls will be tested against a more neutral wording, or against a wording that explicitly allows respondents to admit they have no opinion. This “poll checking” process will allow Moore to demonstrate how a more realistic picture of public opinion can be obtained.

In iMediaEthics’ first PollCheck, Moore examines national public opinion about the U.S. government’s loans to Chrysler and GM to prevent the car companies from going bankrupt.

To support the PollCheck feature and Moore’s analysis of media polling, iMediaEthics has created a new section of its website devoted to media polling. In iMediaEthics’ Polling Center, all PollChecks, PollSkeptic columns, Moore’s PollTalk blog, the annual Dubious Polling Awards and polling resources are organized for readers’ easy access.

iMediaEthics is published by Art Science Research Laboratory (ASRL), a not-for-profit directed by Rhonda Roland Shearer, an adjunct lecturer at the School of Journalism and Mass Communication at the University of Iowa. Shearer co-founded ASRL with her late husband, Harvard professor and scientist Stephen Jay Gould.

“iMediaEthics’ polls and the PollCheck project are not just about exposing flawed polling methods and their results, but educating the public what to look for in polls so they can judge for themselves,” Shearer says. “Ultimately we want to hold the media accountable for producing scientific polls that report, rather than manufacture, news about what the public really thinks.”