Jonah Lehrer, a neuroscientists blogging at The Frontal Cortex from withing ScienceBlogs, comments on new research on partisan bias in perceiving reality:
Yesterday, we looked at some new research that found that when conservatives were exposed to evidence demonstrating the falsity of a partisan belief – such as a report demonstrating that Iraq didn’t have WMD, or that lowering taxes doesn’t increase government revenue – they became more convinced than ever that those beliefs were actually true. The scientists call this “the backfire effect”.
The researchers argue that conservatives are particularly vulnerable to this cognitive flaw, as their beliefs tend to be more rigid and immutable. But I’m not so sure. As a liberal partisan hack, I’m very aware of how my political biases distort my processing of information. I fixate on news that jives with my beliefs and tend to ignore those inconvenient facts that contradict my inner talking points.
We discuss this in our book, in chapter eight:
Polarization as a perceptual screen on reality appears ubiquitous. We can see an example of this in a recent survey on 9/11 conspiracy theories. One prominent conspiracy theory about the attacks centers around the claim that President Bush knew about the attacks in advance. This conspiracy comes in the form that Bush either planned the attacks himself or was too incompetent to do anything about them despite his knowledge of their imminence.
Respected commentators of all political stripes have categorically rejected this theory. But what about the public? Republicans in the survey rejected the idea that President Bush knew about the attacks in advance by a 7-1 margin. On the other hand, Democrats were close to evenly split on the question.
The country is divided geographically as well as by party. For example, Edward Glaeser and Bryce Ward note that in 2004, “twenty-three percent of respondents in Oregon, Washington and California thought that Saddam Hussein was personally involved in the September 11, 2001, attacks. Forty-seven percent of respondents in Texas, Oklahoma and Arkansas had that view.” Glaeser and Ward also report that “56 percent of Mississippi residents think that AIDS is God’s punishment for immoral sexual behavior. Only 16 percent of Rhode Island residents share that view.”
Looks like partisan filters are a generic feature of being human (one might describe them as especially strong priors). But do these filters have policy consequences? Perhaps Condorcet’s jury theorem can save us (if errors are independent), perhaps not (if they are not). Or perhaps the moderates that help decide elections are far less inclined to such mistakes than committed partisan. I’m not sure; I’d like to see some evidence on this score.
Boris,
This reminds me of the literature on the discounting of source bias. For example, if you hear good things about a candidate, you’re likely to believe it, or at least be influenced in that direction. That’s the principle of much political advertising. Even if people know it’s advertising, they don’t discount the source, or don’t discount it enough.