Back to News

September 28, 2012

The Skewing of an Election

Are the recent presidential polls skewed? Could that impact turnout?

Well, that is what some in the GOP and others have been pondering, given revelations that the way polls are being weighted may favor one party more heavily. In fact, there are strong indications that the current industry standard media polling model is based on the 2008 turnout—a historic anomaly—instead of the more historically accurate and consistent 2004 election model.

And while pollsters use complicated models, the issue really comes down to one thing — sampling. The sample of a poll is the basis from which all other information is extrapolated. The bottom line is that when polls use 2008 voter turnout for modeling instead of the historic average, it will inevitably skew the poll to favor the winner of the 2008 race — Mr. Obama.

If there is a bias in these polls, it could have a major impact on turnout. The relentless drumbeat of the media in the past two weeks is that the polls now show a clear shift in momentum toward the incumbent. This conclusion is being cemented in place well before even the first Presidential debate.

If marginal voters – inconsistent participants who vote only when they believe the election is close — perceive one candidate as having a clear advantage so early, then the impact on turnout could be of real consequence.

Of course, if the polls are biased to favor one candidate over another the next question one must ask is whether or not mainstream media outlets and the polling firms they hire are attempting to affect the outcome of the election by skewing the polls.

Regardless of motivation, the fact that the polling models being used are based on the statistical and historic anomaly of the 2008 election is of real concern. The integrity of our electoral process is at stake if there is a real effort to use the power of polling and the media to affect the outcome of the election.