AP Poll: Setup for a Stolen Election?

As the Associated Press "tightens" the presidential race to a one-point margin, Michael Collins asks: Is the new AP "Poll" a Setup for another stolen election?

The Associated press came out with a remarkable poll Thursday. It claimed that the presidential race had tightened from the consistent leads Obama has shown over the past weeks to become one of those "too close to call" national elections. Their poll, conducted by German polling firm GfK, showed 44% for Obama and 43% for McCain. How could this be?

In the Real Clear Politics average of major national polls, Oct 24, Obama's average lead is 7.5%. The four national polls conducted on the same days as the AP poll show an average 9.2% Obama lead. The polls after the AP "tightening" effort shows an 8.7% Obama lead. Clearly the AP-GfK poll is an exception or outlier, as those are called. Yet the AP presents this as news, not opinion. What is AP Up To?

AP acknowledges that its poll is different from the other national polls but claims it's not "alone." It isn't. There is exactly one other poll that agrees with AP out of 16 presented on the Real Clear Politics summary for Oct. 24. AP ignores the tentative nature of the poll and presents the story as "news." It begins drawing conclusions from the poll at the start of the article with "rumors" from unidentified sources; "The poll, which found Obama at 44 percent and McCain at 43 percent, supports what some Republicans and Democrats privately have said in recent days: that the race narrowed after the third debate as GOP-leaning voters drifted home to their party and McCain's "Joe the plumber" analogy struck a chord." AP, Oct. 23, 2008 Reviewing the internal details and questions of the poll analysis by GfK, there are no questions about Joe the Plumber. Since this is the opening claim, mentioned again as central to the shift AP is selling us, it's important to look at how the debate was viewed by the public.

CNN and CBS both did national polls the evening of the debate. CNN reported that 58% of those watching the debate thought Obama won, with only 31% favoring McCain. CBS polled uncommitted voters who judged Obama the winner over McCain by 53% to 22%. This was the biggest margin for Obama for the series of three counters. What was going on with the AP sample of voters?

AP's Flawed Poll
The Associated Press is no stranger to polling. It is the lead sponsor for the national exit polls for Presidential and Congressional elections. Their recent effort in 2006 surveyed over 13,000 individuals and produced detailed demographic results on voters that actually contradict their current effort. The built in bias of the AP poll is obvious when one reviews the internals. 45% of the 800 person likely voter sample described themselves as "born again or evangelical" Christians. The remaining 55% said no (AP-GfK p. 24). AP should have known that it's 2006 exit poll showed that 36% of respondents were born again, a much lower percentage than included in their McCain friendly poll of yesterday.

Also, AP should have known the 2006 exit poll the group had a 60% preference for Republican candidates over Democrats. The election magic for McCain was complete before the presidential preferences were even sampled. The over representation of Christians and the born agains in the sample guaranteed a preference for McCain. But there was more. A look at the poll shows that 1,101 registered voters were surveyed. The final poll consisted of 800 "likely voters" taken from that group. The registered voters divided up with 48% for Obama and 38% for McCain. The reduction of Obama support and increase for McCain indicates that McCain's voters are highly motivated to turnout while Obama's are more likely to stay home. This contradicts the actual evidence available showing major gains in newly registered voters for Obama and huge advantages in primary turnout for Democrats compared Republicans. It's small crowds for McCain and record breaking gatherings for Obama around the country, including 35,000 in Leesburg, Virginia on Thursday Oct. 23.

In polling, the sample determines the outcome of a poll. If you bias the poll sample toward groups that are well known to support one party, then the poll will reflect that sample selection. In this case, AP, GfK did just that. They created a Republican friendly sample which displayed more motivation to vote because it started out Republican friendly. The AP-GfK "Giveaway" Within the details of the poll, there's a major clue that something is very wrong. The likely voter sample was asked questions that are directly related to attacks and negative campaigning. Voters are consistent in decrying what they perceive as unfair attacks.

READ MORE at Gather.com