[Posted by Karl]
I really don’t mean to pick on Jonathan Chait here — he just happens to be representative of many left of center in this instance. Last Friday, the story was all about Obama’s improved standing in the polls, with Gallup as an outlier (to be dismissed like Rasmussen, as if that makes sense). Yesterday, he was groping around to explain away bad numbers from the Washington Post/ABC News and New York Times/CBS News polls (both historically have pro-Obama house effects), while highlighting that Obama’s approval number was up at Gallup. The… flexibility is amusing, but the question of what happened is worthwhile. There are a number of possible explanations.
Statistical noise: Jonathan Bernstein notes this as a factor — and it almost certainly is. However, Bernstein adds that it’s better to look at the poll averages — and they may tell the same story. The variation in Obama’s overall approval rating has not varied much outside margin of error recently, but it seemingly has declined slightly over the past three weeks. The same seems to be true of Obama’s approval on the economy. And the gap between Obama and Romney has been shrinking over the same period.
Sampling: As Ed Morrissey noted, this was likely an issue for the WaPo/ABC poll. After not releasing the party ID for the samples in January and February, this poll disclosed them. The February sample was D+11, while the March sample was D+4, which accounts for much of the results. However, it is unlikely this was a problem in most polls taken over the past three weeks.
Gas prices: This was the explanation given in the WaPo news coverage of its poll. However, as WaPo Wonkbook blogger Brad Plumer noted, the correlation of gas prices and presidential approval is statistically suspect and perhaps indirect (I have noted the same). If you want to dig further, it appears that rising gas prices could cost Obama a few points in approval and a point or two in vote share.
The “War on Women”: Both Allahpundit and Mickey Kaus at least allude to the idea that the media effort to reframe the HHS mandate as a conservative war on contraception has failed and the issue is hurting Obama. I tend to doubt this, but the issue has received increasing media coverage over the past three weeks.
Good economic news: Chait floats a theory that I find implausible in one way, but illuminating in another. Chait, noting that Obama’s campaign to convince voters that “America is back” is a dud, hypothesizes last weekend’s polling reflected that Americans “do not think that it’s morning in America and respond badly to any suggestion that they ought to feel cheerful.” I tend to doubt last weekend’s polling reflected backlash to happytalk about Friday’s jobs report, because the decline — if the start of a trend — dates back to mid-February. However, Chait’s piece caused me to look back at the news coverage from mid-February… and it turns out there was a lot of happytalk about the economy (and how it was boosting Obama) then, too.
The question not being addressed is whether Americans’ polled perceptions of the economy are more accurate than those of the establishment media. I don’t want to overdo this — economic confidence is on the rise. But job creation may be stagnating. Real disposable income has fallen recently. Underemployment seems to be rising. A disproportionate share of the jobs that have been added over the past two years have been what Democrats once dismissed as low-wage “McJobs.” The official job numbers look good, but everyone from Gallup to Wall Street economists are puzzling over the disconnect between the unemployment and growth statistics. Are the strong jobs numbers the real story, or the weak growth stats? The conventional wisdom is that unemployment is a lagging indicator, which suggests we should already have seen better growth statistics. But if the conventional wisdom was always right, the economy would already seem much better than it is.