Z-Tel-0001-Stock

In 2001, University of Chicago law professor Cass Sunstein predicted that the Internet would increase the political polarization of America. If conservatives congregated at conservative websites, and liberals at liberal sites, then it would reduce the "unplanned, unanticipated encounters" that are "central to democracy itself." With sites like Daily Kos, Drudge Report, and the Huffington Post becoming almost mainstream, many observers have echoed his observation. A new paper by economists Matthew Gentzkow and Jesse Shapiro suggests that Sunstein's prediction has not come true.

The authors begin by pointing out that since the Internet reduces the cost of reaching a news source, it can increase the diversity of one's reading by increasing its quantity. They go on to analyze data from comScore, a company that tracks Internet traffic, to calculate the number of users on 119 news and politics websites. They combined the usage figures with data reflecting the political outlook of 12,000 panelists who participated in an electronic survey and classified themselves as conservatives, moderates, or liberals.

The authors then investigated how the Internet usage of conservatives and liberals differed. For each of the sites in their analysis, they calculated the average percentage of conservative users, and then looked to see if conservative and liberal users tended to concentrate their viewing on sites that shared their views.

For the most part, their data confirmed what we already know about websites. For example, 88 percent of the visitors to foxnews.com are conservative, whereas 40 percent of the visitors to nytimes.com are. Fully 98 percent of visitors to glennbeck.com are conservative, while only 6 percent of those visiting thinkprogress.org are.

But the results also suggest that Internet users are voracious readers who flit from site to site with abandon. In the end, that means that the typical conservative visits many sites that are more liberal than he is, and the typical liberal visits many sites that are more conservative than he is.

The authors capture these movements with something they call the "isolation index," which is plotted in the nearby chart. If conservatives visit only sites with all conservative visitors, and liberals visit only sites with all liberal visitors, then the index is 100. If there is no political pattern to site visits, then the index is 0.

The chart shows the isolation index for different types of media. In the case of the Internet, the average conservative viewer visits websites with a viewership that is 60.6 percent conservative. The average liberal visits websites with a viewership that is 53.1 percent conservative. The difference between those two, 7.5 percent, is the isolation index for the Internet. That figure is distinctly higher than the index for magazines, broadcast news (the evening newscasts of ABC, CBS, NBC, PBS, or the BBC), cable television (CNN, Fox News,MSNBC, CNBC, or Bloomberg), and local newspapers. But it is lower than for national newspapers (the New York Times, USA Today, and the Wall Street Journal), and for random face-to-face interactions within one's zip code.

The authors also explore viewing habits, and they find that individuals who frequent more extreme sites tend to offset that usage with visits to less extreme ones; visitors to extreme conservative websites are more likely to visit nytimes.com than the average user.

The Web may be increasing the attention paid to screamers, but it has also made it costless for people to try out other points of view as well. Politics may be increasingly polarized, but don't blame the Internet.

Kevin A. Hassett is a senior fellow and the director of economic policy studies at AEI.

Photo credit: iStockphoto/ahlobystov