In the final weeks leading up to the June 8 Democratic Senate runoff in Arkansas, no data proved more pivotal in shaping conventional wisdom than a pair of Research 2000 polls showing challenger Bill Halter holding a lead.

And those surveys—which fueled the narrative that Sen.Blanche Lincoln was a goner—may have been bogus, according to the blog that commissioned them.

The prospect that polling data in a Senate contest of national consequence may have been faked has sent shockwaves across the campaign world, raising disturbing questions not only about the reliability of suddenly ubiquitous public polls, but about a new media environment where polling numbers are accepted without question even as they threaten to influence the outcome of campaigns.

The episode marks the second time in less than a year that a pollster’s results came under serious questioning—the Atlanta-based polling firm Strategic Vision, was also accused of falsifying data, and its failure to disclose information about its methodology led to a rebuke from the American Association for Public Opinion Research for violating its ethics rules.