For most of President Obama’s tenure in office, Gallup’s Presidential approval polls have shown the President performing at lower levels than most other pollsters. Their polls were often used by Republicans to support the narrative that the President and his policies were unpopular.
Throughout the summer of 2012 Obama’s approval numbers lingered in the mid-40s as he was locked in a dead heat with Mitt Romney. Despite overestimating Obama’s support by four points in 2008, Gallup came under fire from the left for being biased. When Gallup defied the polling consensus, some polling observers such as Mark Blumenthal complained that Gallup was missing potential Democratic voters because they were not sampling non-telephone households. In other words, Blumenthal thought Gallup’s polls were skewed towards Republicans.
The complaints about Gallup from the left reached a crescendo in August when the Department of Justice joined a former Obama staffer in filing a lawsuit against Gallup for overestimating costs on a government contract. Conservatives complained that this amounted to working over the refs. Coaches rough up the refs for one reason – it works. And sadly it appears that it worked on Gallup as well.
Gallup’s new polling procedures take three steps to further secure the Democratic base.
1. Gallup attempts to account for more minority voters by weighting their samples more heavily toward urban respondents. The logic being that urban areas are more densely populated and current geographic sampling leads to a rural bias. In addition, Mark Blumenthal noted happily that Gallup now appears to be polling non-telephone households, which should boost minority poll participation.
2. Gallup added emphasis to include early voters in their likely voter sample. Although the history of early voting is still limited, the early voting trends thus far have leaned heavily Democratic.
3. Gallup is significantly increasing their cell phone samples. In the past Gallup favored landlines by a 60-40 margin. They will now split calls between landlines and cellphones evenly. Cellphone respondents lean noticeably Democratic.
I have no fault with Gallup’s reasoning on any of these changes. The range of respondents should always match actual demographics as closely as possible. No voter is more likely to vote than someone who has already cast a ballot –and it is a fair assumption to say that half the country uses a cell phone as their primary phone.
When taken individually Gallup’s new guidelines seem to make sense. The problem comes from Gallup locking in Democratic demographics and creating a floor of support for the Democratic candidate. Consider the following:
· The minority vote is weighted to at least 25% (actually exceeding 30% in some recent Gallup samples) and votes Democrats at about an 80% clip.
· Early voters are given priority treatment in likely voter samples and vote Democrat by a roughly 60-40 margin.
· 50% of respondents are called on a cell phone and lean Democratic.
With these demographics locked in it will require some hard work for a Democratic candidate to actually have a bad poll.
Unfortunately, pollsters make no such consideration for Republican demographics. Pollsters have been under-polling evangelical Christians by as much as 50% without their consciences troubling them.
Gallup is now taking the approach, along with many other pollsters, that some samples are more random than other random samples. Apparently, the randomness only comes in after the Democratic base is safely counted.
What affect has this had on Gallup’s polls? In this past couple of weeks despite a national shift towards Mr. Romney and a disastrous debate performance, President Obama is reaching highs in approval last seen in 2010. According to Gallup, the President’s approval has been running in the low 50-percentile range. They are the only pollster to currently have the President’s approval above 50%. Within just a couple of months, Gallup went from being the harshest on the President to the most favorable.
The change in polling tactics, which coincided with the Department of Justice lawsuit certainly raises questions. The fact that these “tweaks” appear to have significantly shifted Gallup’s polling outlook in favor of the President appears even worse. If Gallup overestimates the Democratic support once again they will certainly damage the brand they have worked so hard to build.