After Gallup released a poll showing Obama and Romney tied at 48% with battleground state women, the Obama campaign exposed a flaw in Gallup’s likely voter methodology.
The Obama campaign pointed out that Gallup’s likely voter survey was way off in 2010, “Only 2 years ago the distortions in Gallup’s likely voter screen were exposed, leaving Gallup’s survey 9 points off the mark just days before the election. Gallup’s likely voter model predicted a 15 point advantage for Republicans, 55-40, on October 31, 2010. The final result was a 6 point margin, 51-45. That year, Gallup’s registered voter survey was much closer to reality at 48-44.”
The Obama camp called the Gallup likely voter women battleground poll an outlier and said that in 14 polls in 8 battleground states since October 4, President Obama has led with women in every poll. In ten of the polls Obama held a double digit lead, and his average margin in all 14 polls is 10.2%.
Swing state polling chart:
The Obama campaign suggested that Gallup’s other polling was more accurate, but that there is a problem with the questions that Gallup uses to screen likely voters, “We believe the problem with Gallup’s outlying data is rooted in their 7 question likely voter screen, which distorts the composition of likely voters, leading to erratic and inaccurate results…Several of the likely voter questions create a bias against groups inclined to support Obama. For example, Gallup asks voters both whether they have voted in their precinct before and where people in their neighborhood go to vote. This creates a bias against registered voters who more likely to move from time to time, such as young voters, renters, minorities and urban dwellers, all of whom tend to lean toward the President.”
In general, the venerable polling firm has largely been out of step with the polling consensus in 2012. Unlike when the Romney campaign whines about polls, the Obama camp actually provided a methodological reason for their skepticism.
Since the election has tightened up, the Obama campaign is more sensitive to the perceptions caused by polling, but it is important to note that not all polls are the same. Too often, the media gives every poll equal weight when it shouldn’t. Some polls really are better and more accurate than others. Anyone who reads polling should do so not by focusing on one poll, but by looking at the overall picture presented by 5 or 10 polls.
Gallup’s recent track record over the last couple of years suggests that they are oversampling Republicans. Gallup has a ton of credibility because they are the founding father of political polling, but even legends make mistakes, and something smells fishy in Gallup’s latest women likely voter survey.