Listening to people complain about bias in the media, you're reminded that there is more than one paranoid style in American politics. While the left has busied itself unpacking interlocking directorates and corporate ownership, the right has made a specialty of close reading, with an extraordinary attentiveness to the nuances of usage and address.
There's no better example of this than Bernard Goldberg's claim, in his bestseller Bias, that TV broadcasters "pointedly identify conservative politicians as conservatives" but rarely use the word "liberal" to describe liberals. As Goldberg explained the difference: "In the world of the Jennings and Brokaws and Rathers, conservatives are out of the mainstream and have to be identified. Liberals, on the other hand, are the mainstream and don't have to be identified."
To tell the truth, Goldberg's claim about the use of labels didn't sound that implausible to me -- not because I assumed the media were biased, but because the word liberal itself has become an embarrassment to so many people. Two decades of conservative derision have turned it into "the L-word," to the point where some Democrats won't own up to the label and others are careful to prefix it with "neo-," so as to distance themselves from those "unreconstructed" tax-and-spend stereotypes. And on the left, where suspicion of liberals has always run deep, most people have thrown the word over the side in favor of "progressive." But no one ever talks about "the C-word," and conservatives invariably wear that label proudly. So it wouldn't be surprising to find that the media, too, were more diffident about calling people liberals than about calling them conservatives.
Still, the psychology journals are full of studies that remind us just how deceptive our subjective estimations of statistical tendencies can be. And Goldberg is offering an empirical claim, even if he couldn't be troubled to back it up with any research. Granted, it isn't a simple matter to survey the language of TV newscasts, but the language of the press is readily available online. So I went to a Dialog Corporation database that has the contents of more than 20 major US dailies, including The New York Times, the Los Angeles Times, The Washington Post, The Boston Globe, The Miami Herald, Newsday, and the San Francisco Chronicle. I took the names of 10 well-known politicians, five liberals, and five conservatives. On the liberal side were Senators Barbara Boxer, Paul Wellstone, Tom Harkin, and Ted Kennedy, and Representative Barney Frank, all with lifetime Americans for Democratic Action (ADA) ratings greater than 90 percent. On the conservative side were Senators Trent Lott and Jesse Helms, Attorney General John Ashcroft, and Representatives Dick Armey and Tom DeLay, all with lifetime ADA averages less than 15 percent. Then I looked to see how often each of those names occurred within seven words of "liberal" or "conservative," whichever was appropriate, a test that picks out ascriptions of political views with better than 85 percent accuracy. [For more complete results click here.]
And indeed, there was a discrepancy in the frequency of labeling, but not in the way Goldberg -- or for that matter, I -- assumed. On the contrary, the average liberal legislator has a better than 30 percent greater likelihood of being given a political label than the average conservative does. The press describes Frank as a liberal two-and-a-half times as frequently as it describes Armey as a conservative. It labels Boxer almost twice as often as it labels Lott, and labels Wellstone more often than Helms. And the proportions of labeling of liberals and conservatives are virtually unchanged when you exclude opinions and letters to the editor. What's more, the discrepancy is almost as high even if you restrict the search to The New York Times, The Washington Post, and the Los Angeles Times, those pillars of the "liberal press."
The tendency isn't limited to legislators. For example, Goldberg writes that "it's not unusual to identify certain actors, like Tom Selleck or Bruce Willis, as conservatives. But Barbra Streisand or Rob Reiner . . . are just Barbra Streisand and Rob Reiner." But that turns out to be dead wrong, too: The press labels Streisand and Reiner more than four times as frequently as Selleck and Willis. (Nothing if not careful, I screened out examples that might include references to Reiner's portrayal of "Archie Bunker's liberal son-in-law.") Warren Beatty is labeled more often than Arnold Schwarzenegger, and Norman Lear is labeled more often than Charlton Heston.
It goes on. Goldberg claims that former Circuit Judge Robert Bork is always called a conservative whereas Laurence Tribe is identified merely as a Harvard law professor. But it turns out that Bork is labeled only a bit more frequently than Tribe is. And columnist Michael Kinsley gets a partisan label more often than either William Bennett or Jerry Falwell.
It could be, of course, that the figures for people like Lott, DeLay, and Armey were skewed by the fact that they have leadership titles that might take the place of partisan labels in many stories. But the pattern is the same for other legislators. Substitute Richard Shelby or Strom Thurmond in the conservative group and the score for the group goes up a bit; substitute Jeff Sessions or Mitch McConnell and it goes down. Ditto the liberals -- inserting Ron Dellums would raise their score, inserting Nancy Pelosi would lower it.
The fact is, though, that however you cherry-pick the groups, there's no way to make the survey come out as Goldberg claims it should, where conservatives are systematically labeled more than liberals are. True, the proportions of labeling might be different in TV newscasts. But given the way the labels are used in the "liberal" press, the burden of proof is on Goldberg and other critics of liberal bias to do the studies that should have been done before making the claim. In the meantime, it's hard to believe the proportions would come out in a radically different way, not unless you're willing to assume that the language used on CBS and CNN is going to turn out to be much more liberal than what you find in The Washington Post and The New York Times.
Beyond showing that Goldberg is wrong about the use of labels, what does this all prove? If you took his premises at face value, you'd say that the patterns of labeling actually reveal a slight conservative bias in the media. But that misses the point, not just because it takes for granted the right's portrayal of the media as ideologically monolithic, but because it assumes that the press is really using these labels to slant its coverage in one direction or the other.
Why does the press use these labels, anyway? At a first pass, we might assume they're purely informative, a way of letting readers know where a particular politician or group is coming from. That does seem to be the way labels are used with interest groups and think tanks, which generally have a much higher rate of labeling than politicians do. This makes sense -- groups like these are almost always cited in the course of providing background on the "liberal" or "conservative" point of view on a story, rather than as its subjects. In fact the major finding here wasn't in the relative frequency of labeling of conservative and liberal groups, which was slight it was in the striking imbalance in the absolute numbers of citations for each side. The numbers confirm what Michael Dolny has shown in a series of annual studies for the group Fairness and Accuracy in Reporting (FAIR): Reporters' Rolodexes stop three times as often at the names of conservative think tanks as at the names of liberal think tanks.
But the informational needs of readers don't explain the labeling of politicians and other public figures. Readers scarcely need to be told where Helms or Kennedy stands on the issues, after all, but a label could helpful in situating McConnell or Pelosi.
It turns out, though, that Helms gets a label anywhere from eight to 10 times as often as Senators Tim Hutchinson or Sessions, and Kennedy gets one much more often than Pelosi or Paul Sarbanes does. Those discrepancies clearly have less to do with views than with viewers -- ADA or ACA ratings are less reliable as predictors of how frequently politicians are labeled than the number of times they've appeared on Hardball or Larry King Live.
In the end, then, the function of political labels isn't to inform or indoctrinate readers about the people and groups they're attached to. Rather, they're a way of reassuring us that the writer and publication are comfortably in the center, at a safe distance from the extremes on either side. And for these purposes, it's more expedient to use the labels with ideological archetypes like Helms or Wellstone than with a McConnell or a Sarbanes.
If there is a media bias in this, it consists, as Jonathan Chait has put it, of "ratifying the stereotypes that already exist." And if the media wind up labeling liberals somewhat more than conservatives, that's chiefly an indication of how phobic they've become about charges of bias from the right. In this sense, the disparities in labeling are chiefly a tribute to the success that critics like Goldberg have had in popularizing their picture of the media -- and by the by, to the extraordinary attention that the media have given to views like Goldberg's in the first place.
Shortly after I did a "Fresh Air" commentary about this study, I got an e-mail from a conservative who told me that the best indication of the liberal bias of the media lies in the fact that conservatives complain about bias much more often than liberals do. "If 90 percent of the bias complaints have been coming for years from the right," he wrote, "there must be something to their complaints." That's a misapprehension, of course, but an entirely understandable one. In newspaper articles published since 1992, the word "media" appears within seven words of "liberal bias" 469 times and within seven words of "conservative bias" just 17 times. If people are disposed to believe that the media have a liberal bias, it's because that's what the media have been telling them all along.