Gallup Explains How It Screwed Up 2012 Presidential Polls So Badly

Gallup offered up reasons this week for why its polls on the 2012 election were so disastrously wrong, showing Mitt Romney with a small but comfortable lead when all others put Barack Obama ahead.

The polling firm saw a hit to its reputation when its predictions of the 2012 presidential election ended up so wrong, and offered a bit of an apology on Tuesday. The organization’s daily tracking poll showed Romney consistently in the lead, including a 1-point edge in the final poll. Obama would win by 4 percentage points.

On Tuesday, Gallup offered up a 17-page report detailing four reasons why the polls were so off. The main issues were misidentification of likely voters, underrepresentation of regions, problems with weighting for voters of different ethnicities, and the use of a non-standard sampling method.

“None of these factors are large, in and of themselves,” said Gallup Editor-in-Chief Frank Newport, “in the sense that they are not moving the numbers 10 points, 5 points.”

But each moved the polls slightly toward Romeny, Gallup noted, enough of a factor that the pollster says “they made a significant difference in our overall estimate of who was going to win the presidential election last fall.”

In announcing the review a few weeks ago, Newport said Gallup had been doing presidential polling since 1936, and when polls are off like in 2012 “we take it very seriously.”

Gallup’s errors were compounded by the rise of polling analyst Nate Silver, whose blog FiveThirtyEight aggregated polls to create a more full picture of the electorate.

Utilizing data from hundreds of polls, Silver used a mathematical formula that showed the opposite of Gallup’s poll — President Obama holding a small but stead lead over Mitt Romney.

Newport has hinted that after the mistakes of 2012, Gallup could include polling informed by other sources in the future, including “non-probability data, administrative data, social media data.”