Ask a Democrat how Obama has done on unemployment, and you'll get a positive answer. Ask a Republican and you'll get a negative answer. Ask them both about George Bush instead and the results will be reversed. Is this because political partisans really and truly see the truth differently? A team of researchers decided to conduct a couple of tests to find out:
In both experiments, all subjects were asked factual questions, but some were given financial incentives to answer correctly. In both experiments, we find that the incentives reduce partisan divergence substantially—on average, by about 55% and 60% across all of the questions for which partisan gaps appear when subjects are not incentivized.
....In our second experiment, we therefore implement a treatment in which subjects were offered incentives both for correct responses and for admitting that they did not know the correct response. We find that partisan gaps are even smaller in this condition—about 80% smaller than for unincentivized responses. This finding suggests that partisan divergence is driven by both expressive behavior and by respondents’ knowledge that they do not actually know the correct answers. These results have important implications for our understanding of public opinion. Most importantly, they call into question the claim that partisan divergence in beliefs about factual questions is ground for concern about voters’ abilities to hold incumbents accountable for their performance. Partisans may disagree in surveys, but we should not take these differences at face value.
In other words, don't take polls like this too seriously. Even partisans mostly know the truth, but when they're asked questions with actual numeric answers they take the opportunity to trash politicians they don't like instead of answering correctly. After all, that's more fun, and there's no payoff for an accurate answer.
This might be true. But I think there's an alternate possibility: partisans are likely to answer a bit more accurately when they're forced to actually think about their answers. The cash reward is just a way of demonstrating that the pollster is serious about wanting accurate answers. But does this mean that partisans really do know the truth, and are therefore better than we think at holding incumbents accountable? I wouldn't make that leap. Campaigns, after all, are precisely the opposite of this test condition: an environment in which partisans are actively encouraged not to think about their answers. And that means they probably don't. During the heat of a campaign, their true beliefs are probably a lot closer to the inaccurate answers they gave when there was no incentive to think hard.
In any case, I wonder who cares? Partisans are the very people least likely to hold anyone accountable in the first place. By definition, they're the ones who just vote by party. A more interesting experiment would test for accurate responses among nonpartisans, the only group that might be likely to abandon cheerleading during a campaign and try to seek out the truth—though I think it's unlikely even in that case.
This is an interesting study (though I'd note that the questions they ask are really hard), but I'm not sure I'd take it too seriously just yet. I doubt that it tells us anything about actual voting behavior.