Today, the Pew Research Center
released
a poll showing Mitt Romney rocketing ahead of Barack Obama to a four-point lead among likely voters. Needless to say, this is pretty remarkable. Is it true? Well ... maybe, maybe not. Just a few weeks ago Pew showed Obama ahead by eight points among likely voters, and I'm sure I wasn't the only one who found that unlikely. But referencing the recent "poll truther" insanity on the right, Slate tech writer Farhad Manjoo tweeted, "Watch for liberals to start questioning Pew's methodology/sampling/etc in 3, 2, 1...." Well, you can keep waiting. I have seen some liberals express the belief that these results may be inaccurate, particularly since they show the two candidates tied among women. I don't even think Ann Romney thinks her husband is going to be tied among women. But there aren't any liberals, as far as I can tell, questioning Pew's methodology or intentions.
And that's the difference. Whenever any of us see a poll with results we don't like, we have an impulse to explain it away. We look for evidence that there's something fishy going on, to conform to what we would like the truth to be. But faced with a bad poll, liberals don't say that the pollster is intentionally skewing the results to make their favored candidate look good. Yet that's precisely what conservatives have been saying as they saw polls showing an Obama lead. They saw not simply outliers, products of the inherent imprecision of survey research, or even methodological issues, but an actual conspiracy. They saw a similar conspiracy when the September unemployment figures came out and didn't conform to their beliefs about what ought to be happening.
Let's not forget that polls are indeed imprecise. That's the nature of sampling; when you take a bunch of polls, there will be variance in the results they get. That's why it's much more useful to use an average of the polls than any one particular one. When you average a bunch of polls you're effectively getting a much larger sample size, and thus a smaller margin of error, making it more likely that the true result is closer to what you're seeing. Even pollsters who do everything right will on occasion produce results that don't give an accurate picture of the world.
As far as I can tell, liberals may approach polls they don't like skeptically, but unlike many conservatives, they don't jump on board the crazy train at the first sign that the results aren't what they'd prefer. And don't forget, John McCain led Barack Obama for a time during 2008. The polls go up, and the polls go down.
