In response to my original post, John Samples of the Cato Institute expands on his comments on the post over at Cato’s blog. Because this is a long post, I’ll respond to several of John’s comments below the fold.
But first, let me highlight a paper cited by another commenter, which is authored by Valentino Larcinese, Riccardo Puglisi, and James M. Snyder, Jr. and is available here:
bq. We study the agenda-setting political behavior of a large sample of U.S. newspapers during the last decade, and the behavior of smaller samples for longer time periods. Our purpose is to examine the intensity of coverage of economic issues as a function of the underlying economic conditions and the political affiliation of the incumbent president, focusing on unemployment, inflation, the federal budget and the trade deficit. We investigate whether there is any significant correlation between the endorsement policy of newspapers, and the differential coverage of bad/good economic news as a function of the president’s political affiliation. We find evidence that newspapers with pro-Democratic endorsement pattern systematically give more coverage to high unemployment when the incumbent president is a Republican than when the president is Democratic, compared to newspapers with pro-Republican endorsement pattern. This result is not driven by the partisanship of readers. There is on the contrary no evidence of a partisan bias — or at least of a bias that is correlated with the endorsement policy — for stories on inflation, budget deficit or trade deficit.
Now, on to John Samples’ comments…
John writes:
bq. [The Gentzkow and Shapiro paper] confirms the original concern (or, at least, a legitimate concern) about liberal bias: responding to readers or viewers leads to a biased or distorted account of reality.
It is difficult to take the G&S findings and construct a specific description of how reality is biased. Essentially, G&S measure ideology by identifying phrases commonly used by either Democratic and Republican members of Congress (extracted from the Congressional Record in 2005) and then examining 433 newspapers (again from 2005) to determine how frequently each of these phrases appear. The measure aggregates across 1,000 different phrases. What is the size of the “bias”? How would this bias be manifest if you picked up a newspaper at random? By the use of the phrase “estate tax” instead of “death tax’? By the use of the phrase “American people” (a top Democratic phrase) instead of the phrase “stem cell” (a top Republican phrase)? By mentioning these examples of specific phrases, I am merely trying to show that it is not easy to decipher what the precise nature of any ideological bias is, given this kind of measure. I suspect that this is why G&S don’t make this finding the headline of the paper.
(Note: I am not critiquing G&S at all. I think their method is inventive and valuable. Using a text-based measure is much preferable to Groseclose and Milyo’s count of how often members mention thinktanks. John also asked if I might discuss the Groseclose and Milyo study. I am going to outsource a critique to Brendan Nyhan and Language Log; see also Groseclose and Milyo’s reply. Brendan’s comments are particularly trenchant, I think. I also prefer G&S to G&M because it covers many more media outlets and is based on a richer set of underlying data — e.g., a long list of phrases used in congressional debate, rather than merely the occasional reference to a thinktank’s research; and, in some cases, newspaper coverage from a longer time span.)
It is also quite important for the broader debate on media bias to emphasize again that the “ideal point” of newspapers is very close to that of their readers. Many of those who claim to see bias attribute it to nefarious influences — e.g., conservative corporate hegemons or leftist journalists. But if “bias” is merely a response to readers than the debate gets short-circuited. If we wanted to eliminate bias, it seems that we have to either ask newspapers to take a “position” that may lose them readers, or give every newspaper a readership in which the average reader is precisely in the middle of the ideological spectrum, however we would measure that.
The latter is an impossibility, unless we’re going to forcibly relocate people. The former gets at another of John’s comments:
bq. Why not relieve the media of market pressure as a way of dealing with bias? That prompts another question: Are NPR and the CPB free of political bias in their reporting and analysis?
Regarding NPR, Groseclose and Milyo would say no, since NPR is estimated a ways to the left of the American public, somewhere in between Ernest Hollings (D-SC) and Connie Morella (R-MD). And I’m not entirely sure that NPR and CPB are entirely removed from market pressures, in that they depend on attracting listeners and viewers (who will donate) and accept funding from private foundations and corporations, although they are certainly better insulated than purely for-profit entities.
It’s difficult to envision a foolproof strategy to insulate the media from market pressures. To do so, we might have fewer newspapers whose corporate parents are publicly traded and thus accountable to shareholders who may care more about dividends than investigate reporting. The Tribune Company appears to be going private, with Sam Zell as its new owner. Of course, then critics wonder if this merely substitutes market bias for the owner’s personal biases (a familiar concern about Rupert Murdoch’s acquisition of the Wall Street Journal). Some suggest a non-profit model, like the Christian Science Monitor. I have no ready answer here.
To me, the biggest consequence of mitigating market pressure has nothing to do with ideological bias. Instead, newspapers worried less out about the bottom line will likely produce more and better journalism because they will not be forced to cut staff or foreign bureaus — changes that are well-documented (e.g., here). That is what consumers and analysts would notice, more than any shift in ideological bias.
bq. Is there a market for unbiased reporting? You would think so, but perhaps not. Maybe it doesn’t matter. We may just dump media messages, bias and all, into the marketplace of ideas and trust that something like an unbiased political result will come out the other end.
Two responses. First, I think there is a market for good journalism, whether or not any one story is on balance favorable to a liberal point of view or conservative point of view, or to a Democratic leader or Republican leader. Good journalism may not be the preference of the majority of Americans, many of whom are not dedicated consumers of news. But some Americans appreciate journalism that goes beyond “he said, she said” to investigate and evaluate the claims of political actors. There are enough such Americans to make good journalism profitable. Second, I do not really know what an unbiased political result looks like. What I do know is that media coverage is but one factor in any political process (elections, policymaking, etc.). So, as John says, there is a sense in which media coverage is “dumped” into a bigger process, where political elites have contending ideas and where citizens (who may or may not be influenced by media coverage) may or may not have strong opinions. All of these things can matter, and thus it is difficult to tease out empirically the relationship between bias in media coverage and bias in some political result, presuming we could agree on the nature of the bias in each case.
bq. The post also prompted the following thought: I have worked on the campaign finance issue for many years now and I have never talked to a reporter from major media who doubted any part, much less the whole, of the reform case. Political scientists have not found that campaign contributions have much effect on members of Congress (see the earlier link). But that has not affected the prior beliefs of reporters . One raw assertion of corruption by Fred Wertheimer outweighs a hundred careful studies of the influence of money on politics. That might suggest that how monolithic liberalism is in the media depends on the issue. But still, do reporters favor reform because they are liberal or because they get to write “Look, corruption!” a couple times a week? Or do they favor reform because it tends to suppress accounts of reality and messages that compete with the product offered by their employers? In other words, do they support regulations that confer directly nonproductive deadweight rents on their employers.
(Note: Some of John’s work on campaign finance reform is in his new book, which I have not read but plan to.)
I vote for “Look, corruption!” In Deciding What’s News — a book that deserves addition to Lee’s list — Herbert Gans argues that media coverage reflects certain underlying values, one of which is “altruistic democracy.” In short, “news implies that politics should follow a course based on the public interest and public service.” When politicians don’t act in the public’s interest — because of corruption, incompetence, or whatever — it is news. This value can motivate journalism that is either liberal or conservative. It is true that Democrats were more supportive of campaign finance reform, which perhaps makes it a “liberal” issue. But altruistic democracy also underlies coverage that some might label conservative, like John Stossel’s “Fleecing of America” segments. Political scandals can support a liberal view (“The government screwed up. It should do its job better.”) or a conservative view (“The government screwed up. It shouldn’t be doing this job.”) A value like altruistic democracy also fits with reporters’ belief that part of their job is to serve as a watchdog and (at least sometimes) identify ways in which political leaders fall short. That to me is the simplest explanation for why journalists cover campaign finance in the way that they do.
More broadly, the values that Gans identifies, combined with the biases discussed by Dan Bartlett in the original post, suggest that the entire debate over ideological bias is misplaced. What if we spent more time questioning horse race coverage of elections — which is as prevalent as ever — than we do wondering if media coverage favors Democrats or Republicans? Commentators spend a lot of time and energy looking for every sentence or paragraph that might suggest ideological bias, when there are bigger biases at work, biases that are well-known and have larger consequences for what media coverage contributes to the public’s knowledge of politics and to broader political debates.