Network News

X My Profile
View More Activity

Re: Measuring epistemic closure


Henry Farrell says it has been done:

Take levels of political awareness (i.e. respondents’ ability to answer questions about which party has more members in the House etc) as a proxy for exposure to political information (both biased and unbiased). Divide the population according to the most appropriate metric for capturing the putative cocooning effect you are interested in (i.e. liberals v. conservatives; Democrats v. Republicans or whatever). Then test to see whether increased exposure to political information makes respondents more or less likely to give the right answer to politically salient questions where you know the right answers.

Best of all, Larry Bartels has done this already. In Unequal Democracy, Bartels examines how better informed and worse informed liberals and conservatives respond to a question asking whether economic inequality (as measured by income differences) had increased or decreased over time. The differences (see the graph [top]) between liberals and conservatives are striking. The better informed that liberals are about politics in general, the more likely they are to answer (correctly) that income inequalities have increased over time. The better informed that conservatives are about politics (in general), the less likely they are to give the correct answer. In other words, greater exposure to political information makes conservatives less likely to be right. This strongly suggests that conservatives face epistemic closure, at least on this issue. The more conservatives ‘know,’ the more likely they are to be wrong.

That's just one question, of course, and it's a question that liberals have an incentive to get right and conservatives have an incentive to get wrong. Which is to say, it's perfect, but we need more of them, some of which offer the opposite incentives. Seems like a good job for the Pew Foundation.

Original post on "epistemic closure" here.)

By Ezra Klein  |  April 26, 2010; 12:58 PM ET
Save & Share:  Send E-mail   Facebook   Twitter   Digg   Yahoo Buzz   StumbleUpon   Technorati   Google Buzz   Previous: Filibustering the legislative process
Next: The big winner from immigration reform


The question about income inequality is a question that liberals care deeply about (and thus will be more informed about) than conservatives. Not only that, but conservatives focus more on where that question leads, and are wary to acknowledge it because they believe that the cure is worse than the disease. To conservatives, income inequality is less of a problem than government attempts to rectify it.

I would imagine that if you asked a question that conservatives care more about (for example market distortions caused by government intervention in the market) I suspect you would get an opposite result. Again, liberals are more willing to tolerate unintended side effects of government intervention if they believe in the merits of whatever the intervention is attempting to achieve.

Posted by: sold2u | April 26, 2010 1:50 PM | Report abuse

Standing alone, this is a terrible, terrible example.

You need a control: a question that liberals "should" be motivated to get wrong and conservatives "should" be motivated to get right. All you've demonstrated in this instance is that people are more knowledgeable about a subject when the answer pleases them. You've said nothing about liberals vs conservatives.

"Which is to say, it's perfect, but we need more of them, some of which offer the opposite incentives."
Nope, sorry. The lack of a question with "opposite incentives" does simply leave the question incomplete, it completely distorts the point you're making. Those graphs don't lead to the conclusions you claim. Bad data is worse than no data; you're using the illusion of objectivity by numbers to hide gigantic, subjective leaps of logic. I'm a big fan, Mr. Klein, but this is abuse of numbers, plain and simple.

Posted by: CarlosXL | April 26, 2010 2:02 PM | Report abuse


I don't know if having a control makes the information any more meaningful. Even if people are very clear about what they are talking about--income equality vs. overall poverty levels or how well off todays American poor are in comparison to medieval serfs--I still think it's out of context. There are related questions:

Does income inequality always increase when the overall economy does well?

Are there cases where the poor get appreciably richer but the wealthy do not?

What is the difference in income inequality by age group and immigration status?

What attempts have been made to address income inequality in the past, and how have they turned out in regards to reducing income inequality and/or helping the lower classes?

Unless "Epistemic Closure" is just about finding out that the people we don't agree with are ill-informed, foolish, stupid, or led around by the nose by Machiavellian svengalis. In which case, we already knew that!

Posted by: Kevin_Willis | April 26, 2010 2:21 PM | Report abuse

"it's a question that liberals have an incentive to get right and conservatives have an incentive to get wrong"

I'm not sure I buy this particular incentive argument. Why exactly would I, as a liberal, have an incentive to be aware of income equality trajectories? Might it be that I have an incentive to be a liberal since I'm aware of income inequality trajectories? A chicken-and-egg question maybe?

Posted by: slag | April 26, 2010 3:28 PM | Report abuse

"Then test to see whether increased exposure to political information makes respondents more or less likely to give the right answer to politically salient questions where you know the right answers."

It depends on what the right answer is. Conservatives may have read and agree with the paper I've linked to below, which suggests that studies showing increasing income inequality are flawed.

Conservatives who are more exposed to information (and presuamably prefer to read articles which confirm their own biases) are more likely to have seen articles like this.

By the way, if the CATO article is broadly true, that's probably good for liberals - it means that income inequality isn't actually as big a problem as they might fear.

Posted by: justin84 | April 26, 2010 3:55 PM | Report abuse

It's an uncomfortable reality that most people, even those "well informed" simply believe what is comfortable, what reinforces their existing views, and ignore or discount (make up new statistics, etc.) contradicting evidence.

And there's more to this filtering of information (bias) than only comfort or consistency of belief.

Posted by: HalHorvath | April 26, 2010 4:19 PM | Report abuse

It seems to me that Bartels is stacking the deck a bit, though, by picking that particular question. Income inequality as a concept is important to the left’s philosophy, but not to the right’s. The left would say that’s due to callousness on the right, while the right would say that it’s due to a commitment to different values, but I think both would agree that it’s true as a descriptive statement. There’s thus incentive for the left to learn this figure, while the right would consider it irrelevant.

It’s as if Bartels polled Jews and Christians over the cost of a Bar Mitzvah party, and from there concluded that Christians were ignorant about the economy.

Posted by: tomtildrum | April 26, 2010 4:33 PM | Report abuse

"Income inequality as a concept is important to the left’s philosophy, but not to the right’s."

I'm honestly trying to think of a question that could be characterized as important to the right, and it's really hard. Marginal tax rate trends? I wonder how they'd do on that one.

Other than that, all I've got are "Number of affairs Bill Clinton is alleged to have had" and "Number of times Obama has bowed to foreign heads of state". In all fairness, I'm pretty sure conservatives would nail both of those.

Posted by: slag | April 26, 2010 5:13 PM | Report abuse

timtildrum, good point, but there is an exception -- the idea that Republican economic ideas cause general economic growth which is then claimed to lift all boats. "Trickle down" is another example. In other words, there were specific claims repeated in political talking exactly about whether the middle and lower income people would gain economically over time in response to economic policies.

So if after the policies were in place, the result for the middle class was a deterioration in real total income, then you have a direct contradiction of previous talking points.

And then, it's time for the information bias to eliminate the uncomfortable facts from consideration.

Posted by: HalHorvath | April 26, 2010 5:13 PM | Report abuse

Yipes. This is a great measure of liberal vs. conservative knowledge about income inequality, and a TERRIBLE measure of liberal vs. conservative epistemic closure. I mean, I'm perfectly willing, as a liberal, to believe that the conservative community is more epistemically closed, but nothing about this result suggests that the liberal community isn't simply a very epistemically closed community that happens to believe something that is, in this case, correct.

I'm also a little unsure whether any test that examines particular static opinions at particular points in time really says anything useful about epistemic closure. I think Julian Sanchez' concept is a little fuzzy, but to the extent that I understand it, it seems to me you could test it better by

(a) actually just surveying breadth of sourcing for political information (do they watch multiple channels, etc.) and level of trust/distrust for multiple sources (do they have minimum levels of trust for multiple sources, or do they only trust a small number of sources), and/or

(b) isolating groups from each side that believe something factually inaccurate and seeing how well they responded to provision of factually accurate information.

Posted by: nickel79 | April 26, 2010 5:28 PM | Report abuse

In addition to the valid criticisms described above, I'm not so sure a simple "fact check" test would tell us much. And it could actually mislead us about who is "most informed."

I'm sure 9/11 conspiracy theorists know many more details, many of which probably are true, about events surrounding 9/11 than I do. It's their interpretation of those facts that goes off the rails.

It doesn't surprise me that -- as you mentioned in your post earlier today -- Rush listeners are more informed than Fox News watchers. Rush identifies the enemy, tells his listeners what to watch out for, gives them some details into why a given liberal is destroying the world. Knowing these "facts" does not tell me that a listener is better informed.

Posted by: dpurp | April 26, 2010 9:23 PM | Report abuse

It's hilarious watching liberals wrench their arm sockets patting themselves on the back about how open-minded they are.

I'm still in shock over the high percentage of liberals, including Mr. Klein, who resolutely ignored all information that the CBO numbers on the healthcare bill were based on a pack of lies. The latest report by the HHS Medicare actuaries is yet more proof that liberals deliberately ignored unfavorable information or, at the very least, made sure that information was not publicly available before the final vote.

Posted by: ElmerStoup | April 27, 2010 8:33 PM | Report abuse

The comments to this entry are closed.

RSS Feed
Subscribe to The Post

© 2010 The Washington Post Company