2.  PR Firm May Have Forged Political Polls

The American Association for Public Opinion Research scolded an Atlanta-based PR firm Wedesnday for concealing some of the most basic information about its political polls. But there's now a more damning accusation against the firm, Strategic Vision LLC (not to be confused with the more reputable Strategic Vision Inc.): statistical abnormalities in its polls "suggest, perhaps strongly, the possibility of fraud," Nate Silver reports at Fivethirtyeight.com. Silver applied a statistical technique widely considered to have indicated fraud in the recent Iranian elections, taking advantage of the fact "that human beings are really bad at randomization." Real randomization should occur in the final digits of polling numbers; that is, there should be roughly as many results ending in "2" (such as 42 percent of respondants favoring John McCain in a presidential poll) as results ending in "3," for instance. But Silver found that the Strategic Vision LLC polls were much more likely to end in some numbers (like "0," "5," and "7") than in others. "Over a sample of more than 5,000 data points, such an outcome occurring by chance alone would be an incredible fluke—millions to one against," writes Silver, who allows that "some intrinsic, mathematical reason that certain trailing digits are more likely to come up than others" may be an alternative, yet unproven explanation.

Read original story in Fivethirtyeight.com | Friday, Sept. 25, 2009