Da Echo Chamber

Image via Wikipedia

We’ve discussed confirmation bias on here before: The tendency to search for and filter for information that confirms your preconceptions. It’s why, if you’re inclined to believe anyone Barack Obama appoints to the Supreme Court is going to be a Communist anti-white racist, you’ll tend to find more and more evidence for that proposition while you read the news. Or, alternately, why, if you think all Republicans are bigots, you’ll find more and more evidence for that proposition as you read the news.

And, of course, none of this is limited to politics. It’s why you’ll find more evidence that vaccines cause (or don’t cause) autism, why you’ll find more evidence that your husband or wife is the bestest (or a filthy pig), or why you’ll find more evidence that the government is (or isn’t) covering up UFOs. Etcetera.

So this has, of course, led to the question of whether people are getting better and better at filtering out discomfiting information on political and other matters as the Internet has made it easier and easier to create information cocoons, on the Left and on the Right.

A new study on the topic, “Looking the other way: selective exposure to attitude-consistent and counter-attitudinal political information,” set to appear in the journal Communications Research, suggests that, yes, people do have a pronounced tendency to read information they agree with on the Internet. Specifically, undergraduates in an experiment “spent 36 percent more time reading articles that agreed with their point of view than they did reading text that challenged their opinions.”

Here’s the experimental design, as recounted by the Ohio State press office:

The study involved 156 undergraduate students at an American university.  In the first of two sessions conducted for the study, the participants were asked their views concerning four hot-button topics: gun ownership, abortion, health care regulation and the minimum wage.  They were also asked about 13 other issues that were simply put in to cover the fact that the researchers were interested in these four issues.

Six weeks later, the students were invited to participate in another study, supposedly unrelated to the first.  In this case, they went to a computer lab, where they were asked to give their impressions of a new online magazine.  The online magazine had pro and con articles on the four topics that they were questioned about in the first session.  All the articles had headlines that clearly indicated what position they were advocating.

Participants were told they did not have time to read all the articles, so they should just choose which articles they found interesting, as they would normally with a magazine.  They were also told they didn’t have to read whole articles.  They were then given five minutes to read.

The key for this portion of the study was that the computers had a software program that unobtrusively recorded which articles they clicked on and how much time they spent with each article.

…

The results showed that participants clicked on an average of 1.9 articles that agreed with their views, and 1.4 articles that didn’t.  The participants had a 58 percent likelihood of picking an article that supported their viewpoint, versus 43 percent likelihood of choosing an article challenging their beliefs.

Participants were most likely to read only articles that were consistent with their views, the study showed.  Next most common was reading both views on an issue.  Very few people only clicked on articles that opposed their views.

It’s an interesting study, beset by the usual problems of this type of research — which I’ll get to in a second.

But, first, there’s one extra nugget, perhaps of interest to partisan combatants:

The study found that people with a stronger party affiliation, conservative political views, and greater interest in politics were the ones most likely to click on articles with opposing viewpoints.

“It appears that people with these characteristics are more confident in their views and so they’re more inclined to at least take a quick look at the counterarguments,” she said.  “Even if they click on opposing views, they’re not looking for insights that might change their mind.”

That’s certainly against the grain of the conventional wisdom, which holds that conservatives are more closed-minded. My purely anecdotal/intuitive guess is that conservatives, both in academia and in urban society, are much more likely to have their views challenged on a regular basis. They marinate in a culture that assumes they’re not just wrong, but actually evil. You don’t come to strong conservative views unless you’re willing, or perhaps even eager, to engage in constant intellectual combat. Again, just a theory, but it certainly matches up with what I’ve witnessed in years of covering politics (mostly the conservative movement).

Nonetheless, back to the problems with a study like this: The setting’s just too unnatural and the time frame’s just too small. The logistics of this would be tough, but the best way to do a study on Internet reading habits would be with, say, a Firefox plug in. Answer some questions when you download the plug in, consent to monitoring of your news reading habits for a week or more, track results by IP address as opposed to names. It’s not quite lab controlled, but people would have little reason to game the system. There’d be a selection-bias problem — perhaps people willing to participate would be more open-minded, in general — but I’m sure there are ways to design the experiment to overcome that.

The next question, of course, is what does it matter / what does it mean if people are segmenting themselves off by their views? Is there any serious or plausible reform to be proposed? Given that confirmation bias is an innate human trait, I don’t think you’re going to “solve” it anytime soon.

The best you can do is make people aware they’re doing it. So, here’s this post. You’re welcome, society.