Nothing ratchets up the perennial debate over media bias like a presidential election. But as Tim Groeling, a political scientist at the University of California, Los Angeles, observes, public discussions about media bias are often just “food fights,” with pundits and partisans throwing around anecdotes.

Groeling is hoping to advance scientific (and public) knowledge beyond this mush with research he used to demonstrate selection bias in television networks' decision to run or withhold the results of presidential approval polls. For an article appearing in Presidential Studies Quarterly this December, Groeling designed a method to deal with a problem that often besets research on the media: people can identify all the news that journalists saw fit to print, but it's more difficult to determine what they chose to ignore.

To counter the problem of the “unobserved population,” Groeling collected two different data sets: in-house presidential approval polling by ABC, CBS, NBC and FOX News and the networks' broadcasts of such polls on evening news shows from January 1997 to February 2008. Groeling found that, with varying degrees of statistical significance, CBS, NBC and ABC showed what Groeling calls a pro-Democrat bias. For instance, CBS was 35 percent less likely to report a five-point drop in approval for Bill Clinton than a similar rise in approval and was 33 percent more likely to report a five-point drop than a rise for George W. Bush. Meanwhile FOX News showed a statistically significant pro-Republican bias in the most controlled of the three models Groeling tested: its Special Report program was 67 percent less likely to report a rise in approval for Clinton than a decrease and 36 percent more likely to report the increase rather than the decrease for Bush.

Groeling's work is one of the few studies to quantify partisan bias in the media, a subject notoriously difficult for social scientists to research and discuss. These scientists work with theories such as the so-called hostile media effect to predict that ardent supporters of a cause will view media as slanted for the other side, and they have conducted hundreds of studies that have revealed imbalances in the ways journalists frame news on topics ranging from AIDS to the war in Iraq. But there is not a cohesive literature on media bias. Maxwell McCombs of the University of Texas at Austin, who pioneered agenda-setting theory, one of the leading paradigms on news media, says that a researcher would need a few years to make sense of existing data and develop an approach to study media bias. Like many scholars, McCombs sees “bias” as a loaded term, preferring to speak of journalists' “predilections.”

“Scholars hate the word 'bias' because they feel like they're entering the ideological fray,” says S. Robert Lichter, head of the Center for Media and Public Affairs (CMPA) at George Mason University, who prefers the term “tone.” Despite his efforts, Lichter himself got sucked into that fray. His content analysis of the transcripts of TV news broadcasts at the statement level is a respected and widely adopted methodology. This past summer, just as the view that journalists were going softer on Barack Obama than on John McCain was becoming widely accepted, CMPA issued a report showing that 72 percent of the statements in TV news reports about Obama in late spring and early summer were negative, whereas 57 percent of the statements about McCain were negative. When Fox News commentator Bill O'Reilly attacked Lichter's method during a radio interview, saying it would embolden liberal bias, Lichter responded, “You can take all my studies or none of my studies”—an allusion to past uses of his work to support conservative views.

In recent years disciplines not traditionally interested in media have turned their attention to them. In 2005 the Quarterly Journal of Economics invigorated the debate with a provocative study by Tim Groseclose, a political scientist at U.C.L.A., and Jeffrey Milyo, an economist at the University of Missouri–Columbia. Groseclose and Milyo created a scale and assigned 20 major news outlets and legislators in Washington, D.C., positions based on their citations of think tanks and policy groups labeled liberal or conservative. They also factored in the voting records of House and Senate members. Their measure determined that most of the major media were left of center of the average legislator—even the news pages of the Wall Street Journal were slightly left of the “average Democrat.” The exceptions were the Washington Times and Fox News's Special Report.

Most media scholars do not think the issue of bias can be settled by a formula, though. For example, Groeling observes that the context of news making, including professional definitions of newsworthiness, cannot be ignored when looking at the disproportionate front-page coverage of Obama.

“What more often occurs is this tendency for everybody to start seeing the story the same way,” says Elizabeth Skewes of the University of Colorado at Boulder, who analyzed journalists covering presidential races in her 2007 book Message Control: How News Is Made on the Presidential Campaign Trail. A former journalist herself, Skewes has a view similar to other scholars who have watched journalists work. She says the interplay of campaign logistics, journalistic norms and pressures from competitive editors “make it all but impossible” for different frames of issues and candidates to break into the evening news or the front pages. Journalists may have political biases, but that might not be why the news comes out the way it does.