In Re: Obama v. Gallup

By
Democratic presidential candidate, U.S. President Barack Obama (R) pats Republican presidential candidate, former Massachusetts Gov. Mitt Romney on the back after the Presidential Debate at the University of Denver on October 3, 2012 in Denver, Colorado. The first of four debates for the 2012 Election, three Presidential and one Vice Presidential, is moderated by PBS's Jim Lehrer and focuses on domestic issues: the economy, health care, and the role of government.
Photo: Mandel Ngan/AFP/Getty Images

Gallup has a new poll showing Mitt Romney leading by 5 points among voters in swing states. The Obama campaign has released a memo criticizing the poll, prompting a wave of snickering journalists to compare Obama to the lunatics at unskewedpolls.com and other conservatives with oddball criticisms of polling.

It’s a silly comparison, for a couple reasons.

First, as I argued at the time, the GOP poll denialists may have been completely crazy, but they weren’t completely wrong. The charges of deliberate bias were crazed. But polling is a very tricky field, and as response rates to polls have plummeted, it has grown easier to imagine that respondents may be systematically titled toward one party or the other. What’s more, it’s also possible that high-profile events like a convention could charge up one party’s supporters and make them, at least temporarily, more likely to take the time to speak with a pollster.

Second, Obama’s campaign is making a completely different argument than the Republican poll denialists were. The poll denialist argument centered on the charge that polls were finding too many Democrats and not enough Republicans. Obama’s argument against Gallup centers on its likely voter screen. Gallup finds that, among registered voters, Obama leads by 4 points. But it asks questions designed to find out if the voter will really show up at the polls, and Gallup calculates that likely voters would support Romney by five. Thus, the likely voter screen shifts the race by nine points toward Romney. Obama’s campaign argues that this is too high and cites the previous election as evidence:

- 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.

This seems pretty persuasive. Historically, Republicans vote at a higher rate than Democrats. But the difference is usually a couple percentage points, not 9.

Some reporters have been making a second-order criticism: Obama may have a point about Gallup, but it looks bad to be criticizing the polls. That may be true, but it’s true in a purely self-fulfilling sense. If reporters decide to ridicule sensible criticism of polls, then sensible criticism of polls will harm a candidate. Anything a campaign does that reporters decide to ridicule is probably a bad idea by definition. But stating that tautology is kind of an odd approach to informing the public.