Poll Dance

Stop comparing Sarah Palin’s approval ratings with Barack Obama’s.

Sarah Palin

According to the latest poll numbers, Sarah Palin is nearly as popular as Barack Obama. Or maybe it’s that Barack Obama is nearly as unpopular as Sarah Palin. At least that’s how some commentators see it: As the Los Angeles Times’ Andrew Malcolm noted Monday, “Sarah Palin’s poll numbers are strengthening. And President Obama’s are sliding.” In Tuesday’s Washington Post, former Bush strategist Matthew Dowd wrote that“Palin’s favorability numbers are a mirror image of those of Obama.”

The problem is, they’re comparing apples to oranges. Both columns refer to polls that show Palin’s favorability rating at around 43 percent—mere points away from Obama’s job-approval rating of 49 percent. But as Media Matters has pointed out, favorability and job approval aren’t the same thing. A politician’s favorability rating is a general sense of the public’s feeling about him. His job-approval rating is an evaluation of the work he’s doing.

When you compare favorability ratings—apples to apples—Obama still leads Palin by a distance. The latest Gallup poll puts Obama’s favorability 16 points ahead of Palin’s, ABC puts his lead at 18 points, and CNN says it’s 18 points higher. (Only Fox has the gap in single digits, with a seven-point spread.) It’s impossible to compare their job-approval numbers because, well, Palin doesn’t have a job.

You’d think the two measurements would be roughly the same—but they’re not. In general, politicians tend to have better favorability ratings than job-approval ratings. That has been the case with Obama since January, as it was with George W. Bush, who maintained high favorability (some might call it likeability) even when the public disapproved of what he was doing in office. There are exceptions: During the Monica Lewinsky scandal, Bill Clinton had a lower favorability than job-approval rating. Many people who despised him personally liked where he was steering the country.

Still, the two numbers get conflated all the time. The problem is the fuzzy ways pollsters word their questions. “Pollsters love vague questions because they’re easy to make comparable,” says Mark Blumenthal of Pollster.com. The more specific the questions’ wording, the bigger the disparity. For example, a recent Quinnipiac poll explicitly distinguished between liking Obama “as a person” and liking “his policies.” A full 74 percent of respondents said they liked him as a person. But of that group, a third said they didn’t like his policies. The gap owes partly to the realities of governing: If you ask people how Obama is doing at his job, they’re more likely to think of the 10 percent unemployment rate than if you ask about Obama in general. So when it comes to approval ratings, someone who doesn’t hold office often has a built-in advantage over someone who does.

The job-approval rating also predates the favorability rating. Since the 1940s, Gallup has asked Americans how presidents are doing at their job. It wasn’t until the 1970s and ‘80s that favorability ratings became popular. Favorability is especially useful for campaign pollsters, since it allows them to compare candidates who already hold office with those who don’t.

Nor is favorability/job approval the only common conflation. People also tend to compare different types of job-approval questions—even when the questions yield very different types of answers. For example, some pollsters ask simply, Do you approve or disapprove of how Person X is doing? Others ask respondents to rate a politician’s job as excellent, good, fair, or poor. The latter type of question tends to elicit a more negative response, because many people think of “fair” as a neutral, or even mildly positive, response.

Same goes for favorability polls. The Gallup poll goes for the simple binary: Do you have a favorable or unfavorable opinion of the president? Others, like the CBS/New York Times poll, reflect more nuance: “Is your opinion of Barack Obama favorable, not favorable, undecided, or haven’t you heard enough about Barack Obama yet to have an opinion?” The latter formulation produces lower favorability ratings. The reason, says Blumenthal, is a positivity bias. When respondents don’t really have an opinion but aren’t presented with a “don’t know” option, they tend to answer positively—a phenomenon that’s been dubbed satisficing. Hence the disparity between Obama’s approval rating when respondents are given the option to say they don’t know enough to decide (Quinnipiac and CBS News both put him at 50 percent) and when they aren’t (Daily Kos, ABC News, and Fox News put him at 55 percent, 61 percent, and 54 percent, respectively).

Things get even murkier when you get into issue-based polls. A NBC/Wall Street Journal poll caused a stir over the summer when it changed the formulation of a question about the public option. Initially, the poll asked, “How important do you feel it is to give people a choice of both a public plan administered by the federal government and a private plan for their health insurance?” When 76 percent of respondents said it was “extremely” or “quite” important, supporters of the public plan celebrated. Three-quarters of Americans want a public option!

But when the pollsters reworded the question to be more straightforward—”Would you favor or oppose creating a public health care plan administered by the federal government that would compete directly with private health insurance companies?”—only 46 percent of respondents said they supported the public option. Liberals cried foul. But, of course, they are different questions. “It was silly to conflate them as if they had same meaning,” says Blumenthal. “One was about the importance of having a choice. The other was asking them to react to a description of what the public option is.” Add on the fact that only a bare majority of Americans even knows that a public option has something to do with health care, and it’s easy to see why broad statements about public opinion on reform are so often misleading.

To make things even more confusing, Palin has her own ratings gap. Only 28 percent of Americans think she’s qualified to be president, according to a CNN poll. But people admire her personal qualities. A majority of Americans thinks she’s “honest and trustworthy” and “care[s] about the needs of people like you,” and 64 percent believe she’s “a good role model for women.” It’s this sort of disparity of views—hard to capture in a single number—that makes blanket dualities like favorable/unfavorable, popular/unpopular, and approve/disapprove, so limited.