In what may have been a response to complaints from liberals and the Obama campaign, Gallup changed some of its sampling methods. Even with that change, supposedly in President Barack Obama's favor, the latest tracking poll shows Mitt Romney with a seven percentage point lead, the largest lead for Romney of any recent national poll.
Mark Blumenthal, senior polling editor for The Huffington Post and editor of Pollster.com, has been one of those criticizing Gallup's methods. In a June column for The Huffington Post, for instance, he argued that the Gallup polls were skewed in favor of Mitt Romney. The reason, he said, is that Gallup did not properly weight their sample in a way that would appropriately represent the proportion of nonwhites in the population.
On Oct. 10, Gallup announced changes to its survey methods that appear to be a response to some of those criticisms. Among the changes, Gallup will now include more cell phones in its surveys, from 40 to 50 percent of the sample, will add a weight for population density to, apparently, give greater weight to urban voters, and will make slight changes to the question wording used to determine the most likely voters.
"In all cases we are continually tweaking, modifying, and improving our methodology," wrote Frank Newport, editor in chief at Gallup.
Though Gallup does not report the racial makeup of its samples, Alan Abramowitz, professor of political science at Emory University, calculated (using extrapolations of the racial breakdowns of presidential approval) that the changes increased the proportion of nonwhites in Gallup's weighted sample.
"The increase in the percentage of nonwhites in Gallup's latest weekly tracking poll sample is a welcome development that should produce more accurate estimates of not only presidential approval but also support for the presidential candidates among registered and likely voters," Abramowitz concluded.
Some conservatives, though, complained that Gallup may have bowed to pressure from the Obama administration. The Justice Department recently filed suit against Gallup for overestimating costs on a government contract unrelated to Gallup's presidential election polls. Brandon Gaylord, editor of HorseRacePolitics.com, raised this concern in a Oct. 13 editorial for Townhall.com.
While noting that Gallup's changes may be a legitimate way to improve its methodology, he wrote, "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."
Even with Gallup's changes, though, its polling continues to show Romney leading by the greatest margin of similar recent polls.
The most recent Gallup tracking poll, conducted from Oct. 15 to Oct. 21, shows Romney with a six percentage point advantage, 51 to 45 percent. The next closest of the recent polls, Monmouth, has Romney leading by only three percentage points, 48 to 45 percent, and a recent IBD/TIPP tracking poll has Obama leading by six percentage points, 48 to 42 percent.
In a Sept. interview with The Christian Post, Scott Keeter, director of survey research at Pew Research Center, pointed out that any single poll could be off by a few percentage points. Given the sheer number of polls taken this close to an election, some of them will likely be even outside the margin of error, even when using sound methods, due to the uncertainty involved in sampling.
Keeter also pointed out that the variation in polling results can partly be explained by the different methods used to determine who is a likely voter.
"This process of building a likely voter scale in the polls in something that's very individual for different pollsters. It's also the case that most pollsters don't disclose in great detail how they do it. That could be having an impact on the differences among polls, but it could also be having some impact on what your bottom line is," Keeter said.
There are two additional points worth noting, though, about the Gallup poll. Gallup's results are not based upon a single day's poll, but a rolling average of daily polls. Also, Gallup has the largest sample size, 2,700 likely voters, of any of the recent polls. All else being equal, pollsters believe that a large sample size is more likely to provide an accurate result than a small sample size.