Just how accurate ARE the polls and what does it mean for the underdogs?

This is an archived article and the information in the article may be outdated. Please look at the time stamp on the story to see when it was last updated.

NEW YORK (PIX11) – “The most important poll is the one on election day,” a saying that hints at how unreliable of a science polling can be. Like it or not, when it comes to elections, polls seem to be all we have to go on for an early picture of where the candidates stand. Yet time and time again we see how inaccurate they can be.

Locally, the 2009 election serves as a perfect example. After Mayor Michael Bloomberg extended term limits himself it seemed over $100 million of his campaign war chest was more than enough to buy off or offset the New Yorkers who were upset about the restriction on politicians being overturned by a politician in office. Polls predicted Bloomberg would win well over 50% of the vote and demolish his opposition Bill Thompson.

All double digits, Quinnipiac said Bloomberg would win by 12%. Marist went even further saying he would defeat Thompson by 15%. Bloomberg’s campaign took advantage and sold a third term as an inevitability to the media and to potential endorsers. The Associated Press repeatedly sent test wires ‘Bloomberg Wins,’ never once even testing one for Thompson.

But on election day, Bloomberg barely got 50% of the vote, and the gap barely reached 5%. The polls were off on his margin of victory by some 60%. The surprise perhaps was that anyone was surprised. In 2005 polls said Bloomberg would win by as much as 38 points, he won by half that prediction: 19%.  In 2001, five of six pollsters said Mark Green would win over Bloomberg, yet we still rely heavily on polls. Even in the media.  Watch Arthur’s package for the candidates’ reactions to polls and their ability — or inability — to accurately predict.