Fox News Effect

How do people make decisions? Do they weigh information from all available sources to make rational choices? Or are they subject to persuasion–perhaps even manipulation–that causes them to be selective in what they consider when reaching decisions?

The outcome of the hotly contested 2000 presidential election between George Bush and Al Gore may have hinged on the answers to these questions, according to UC Berkeley economist Stefano DellaVigna. In an influential 2006 paper, "The Fox News Effect: Media Bias and Voting," DellaVigna and coauthor Ethan Kaplan found evidence that the introduction of the conservative Fox News channel into some local cable television markets boosted the Republican share of the 2000 presidential vote by a small, but significant amount–enough to put Bush in the White House. And the finding that some voters were apparently persuaded by a conservative news network may have broad implications not only for our understanding of media and politics, but also for economic theory.

Standard thinking in economics is that people make rational choices. They know their preferences and are able to filter out bias when evaluating information they get from others. That assumption undergirds the way the field of economics analyzes the behavior of investors, consumers, corporate managers, and interprets everything from home sales to the ups and downs of the stock market. In recent years though, some economists have challenged the notion that people typically make logical decisions. These scholars, known as behavioral economists, have used insights from psychology that factors such as social pressure or appeals to emotion can cause people to do things that don't make sense from a purely rational point of view.

The issue of rational decisionmaking versus persuasion prompted DellaVigna and Kaplan to look at Fox News and voting patterns. They exploited Fox's rise to conduct a natural experiment. Fox News was launched in 1996 and by 2000 was available on cable television in 20 percent of U.S. towns. The authors examined voting results for 9,256 towns, comparing data for places that were similar except for Fox's availability. This comparison is a break-through relative to the previous research on media bias, which largely compared people who decide to listen to Fox News to people who prefer to listen to other media. Unfortunately, in these earlier studies this comparison reflects the selection of who want to listen to Fox News, as opposed to an effect of Fox News itself.

In the DellaVigna and Kaplan study, instead, it is the naturally-occurring variability in when Fox News is added to the cable town which creates the variation. Imagine two similar towns, one which added Fox News to its cable system in 1999, while the second did so only in 2001. As of 2000, the towns differ mainly in the availability of Fox News, not in other selected ways. Does Fox News then make a difference? DellaVigna and Kaplan found the network's arrival added an estimated 0.4 to 0.7 percentage point to a town's Republican presidential vote. Fox pushed up the Republican Senate vote share 0.8 percentage point–even where the network didn't cover the Senate race heavily.

Depending on how Fox's audience was measured, the authors calculated that exposure to the network convinced between 3 and 8 percent of non-Republican viewers on the low side to 11 and 28 percent on the high side to back the GOP presidential candidate in 2000. Those results are evidence of a pronounced persuasion effect. "Our estimates imply that Fox News convinced a significant portion of its audience to vote Republican," DellaVigna and Kaplan concluded. The impact may have been decisive. The authors estimate that nationwide Fox added 200,000 presidential votes to the Republican column, enough to throw the all–important state of Florida to Bush.

Was the Fox effect evidence of irrational decisionmaking? The authors are cautious on this point. In 2000, Fox's Republican slant may not have been clear to voters, they suggest. Consequently, some viewers may have attributed Fox's fawning coverage of Bush to his high quality as a presidential candidate. But, as they learned to see Fox's biases, the network's ability to change their vote would disappear. If that was true, Fox's impact did not mean voters were irrational.

On the other hand, voters may not have discounted media bias strongly and may have been swayed to vote Republican. That might indicate that voters didn't take into account all available information, but instead were prey to irrational persuasion. "The two most plausible explanations, learning and persuasion, have very different long–run implications, DellaVigna and Kaplan concluded. "Rational learning predicts that Fox News' impact is temporary. Nonrational persuasion predicts that Fox News permanently altered voting patterns in the United States.”

DellaVigna and Kaplan in this paper provide evidence on the impact on voting of a polarizing media outlet in a close election. But the contribution of this paper goes beyond this particular case. The paper highlights a general methodology to study media bias: the use of natural experiment in media availability. Indeed, following this paper other authors used this methodology to analyze the impact of media bias in settings as diverse as public television in Russia and local newspapers in the early 20th century in the US. By and large, the finding that media bias is persuasive holds.