From the Archives: Another View of News Bias, as Selling Point

If bias is a product flaw, why doesn't it behave like auto repair rates and decline under competitive pressure?

Midjourney from prompt “Two cartoon characters holding newspapers. One is leaning sharply to the left. The other is leaning to the right.”

From 2000 to 2006, I wrote an economics column for The New York Times every four weeks. The column appeared in the business section and wasn’t supposed to be an opinion column. I decided from the outset to treat it like a science column, reporting on new research and interviewing economics. This article originally appeared on May 18, 2005 and it makes points many people still don’t understand or perhaps, confusing “is” with “ought,” simply refuse to accept.

To the consternation of many journalists, there is a widespread, and increasingly noisy, belief that the media are biased. The report in Newsweek, now retracted, about Koran desecration by interrogators at Guantanamo Bay has only intensified the criticism.¹

People in the news business are starting to feel a bit like auto executives in the 1970s under attack from Ralph Nader and others. Every day, journalists hear that they keep producing lemons.

But what exactly makes a story a lemon? Beyond getting the facts straight, good journalism is not as easy to define as a car’s accident or repair rate.

Some people say they want “just the facts,” and fault reporters for introducing too much analysis. Others complain that stories do just the opposite, treating all sides in a conflict as equally valid. The news-buying public seems to want contradictory things.

But one person's contradiction is another’s market niche. Those differences help answer an economic puzzle: if bias is a product flaw, why does it not behave like auto repair rates, declining under competitive pressure?

In a recent paper, “The Market for News,” two Harvard economists look at that question. “There’s plenty of competition” among news sources, Sendhil Mullainathan, one of the authors, said in an interview. But “the more competition there has been in the last 20 years, the more discussion there has been of bias.”

The reason, he and his colleague, Andrei Shleifer, argue, is that consumers care about more than accuracy. “We assume that readers prefer to hear or read news that are more consistent with their beliefs,” they write. Bias is not a bug but a feature.

In a competitive news market, they argue, producers can use bias to differentiate their products and stave off price competition. Bias increases consumer loyalty.

Reporters who firmly believe themselves to be disinterested observers may further this strategy if they share their audience's assumptions about how the world works and, hence, how to interpret particular facts.

Suppose, for instance, that the Bureau of Labor Statistics announces that the unemployment rate has risen to 6.3 percent from 6.1 percent.

In their article, the economists imagine two different takes on the story. One hypothetical version, with the headline “Recession Fears Grow,” notes that 200,000 people have lost their jobs in the past quarter. It quotes a gloomy John Kenneth Galbraith comparing the president to Herbert Hoover and is illustrated with a photograph of a long line of people waiting for unemployment benefits.

The other, called “Turnaround in Sight,” emphasizes the small magnitude of the increase -- just 0.2 percent. It quotes the stock analyst Abby Joseph Cohen as saying the “softness in the labor market bodes well for corporate profitability,” accompanied by a photo of a smiling Ms. Cohen.

“Each of these stories could easily have been written by a major U.S. newspaper,” write Professors Mullainathan and Shleifer. “Neither story says anything false, yet they give radically different impressions.”

But all the information is out there. Indeed, a wide-ranging reader would learn more from the two differently biased reports than from the raw unemployment figures.

That is one important difference between the relatively subtle choices that can bias real-world reporting and the simplifying assumptions that the two economists use to model bias mathematically.

In the economists’ model, news outlets announce that they intend to be biased in their coverage and specify the degree of that bias. The bias itself also appears mathematically as a simple deviation from the truth. For instance, if the unemployment rate is 6 percent, a news outlet biased one way might report a 5 percent rate and its opposite a 7 percent rate.

Professor Mullainathan acknowledged the model's limitations in the interview. “Sometimes when we write these models we face a fundamental trade-off of something that's simple and easy to deal with, versus something that's more realistic,” he said.

But, he argued, this research encourages economists to think differently about media competition, recognizing that consumers have different preferences.

The article makes some provocative predictions. It suggests that adding relatively moderate competitors may push rivals to take more extreme positions to hold onto their audiences.

Trying to correct Al Jazeera's bias, for example, by introducing pro-Western competition, as some analysts recommend, “might cause Al Jazeera and similar networks to further differentiate their product by advancing yet more extreme views,” write the economists. “The effect might be only to radicalize, rather than moderate, their audience.”

1. Here’s a contemporaneous NYT report explaining what this is about.