Liberal Media Bias?

A new study by Tim Groseclose and Jeff Milyo purports to provide a definitive proof of a "liberal media" bias.

The gist of the study is that they measure the amount of times specific "Think Tanks" are cited by newspapers, and also measures how often those same think tanks are cited by congressman. A think tank that is cited by conservative members of congress (via ADA rankings) is labeled "conservative," one cited by liberal members is labeled "liberal." The study found that newspapers cited the liberal institutions at a far higher rate than the conservative ones, and thus concluded that there is a systematic media bias.

At first blush, its a compelling argument. But the flaws in the analysis are numerous. Joel Jacobs remarks on a few of them:

SOURCES CONSIDERED
To reach a fair conclusion about media bias, one would ideally consider all media. There are almost ten thousand newspapers in the country, mostly local. The study considers three papers, all national. There are thousands of magazines that have some political content. The study considers NONE. 22% of Americans get their news from radio; the study considers no radio sources. There must be hundreds (perhaps thousands) of television news sources, mostly local. The study considers three. Thousands of blogs and other Internet information sources; study considers one.

Certainly, studying all media sources would be difficult or impossible. Barring that, a responsible researcher would be obliged to study a number of media outlets that is significant enough, and representative enough, that one could safely generalize. The authors do not even attempt to argue that they have studied a representative sample. Given the absence of radio, magazines, and local media (which is where, I believe, most people get their news), it's hard to call this list representative. If the New York Times is liberal, but most smaller papers are conservative (which is plausible given that conservative chains like Murdoch and Hearst own a great many papers), then there's no overall "liberal bias."

TIME PERIOD

Though the study claims to cover quite a large time period, the data at the end reveal that the analysis of newspapers is limited to the GW Bush years. (And, as noted below, the data for the remaining media outlets are too limited to be useful.) The problem with concentrating on a period when there was a Republican President and a Republican Congress (or at least House) is that it's the government that, by and and large, is making news. Thus, a typical story might have a government official announcing or explaining a policy, and an "independent" source (perhaps a think tank) giving the other side. Most people would consider this sourcing (Administration says X, Think Tank argues Y) to be balanced. Under the study's methodology, though, this would be a "liberal" bias, because only think tanks are considered.


DATA USED

An overall conclusion about bias requires consideration of a number of components:

1. What stories media cover.

2. How prominently they cover them (e.g. putting anti-administration stories on the front page or at the "top of the news," and burying pro-administration stories on p.27 would be a liberal bias).

3. What facts are presented in the stories.

4. What sources the stories rely upon.

5. What positions the media's editorials espouse.

Even on its face, the study addresses only one very small piece of #4. The majority of sources that the media rely upon are not think tanks at all, much less the 200 "think tanks" (I'm sure the American Association of Retired Persons would be surprised to learn that it's a think tank, BTW)
. Considering only citations to these few organizations (and not other organizations, political leaders, academics, etc.) is like, well, a doctor saying "You're healthy because your hair looks ok."

The data are so limited as to be useless. While they don't show any overall bias, they do suggest that the media studied just don't cite think tanks all that often: in many cases not even every day. USA Today cited think tanks about 400 times in about 240 days. I don't read USA Today, but let's assume 30 political stories a day. That would mean a think tank is cited in only about 6% of stories. (The percentage would be lower if multiple think tanks are sometimes cited in a single story.) How can you generalize from that?

Similarly, CBS and ABC cited "think tanks" less than 100 times per year. Figure, what, seven political stories per night (I don't watch TV news either), and 260 days a year of nightly news (do they broadcast on weekends?) That's 1,820 stories a year. Again, think tanks are not even an issue for the vast majority of stories, so it seems an obscure variable to focus on.

OTHER EXPLANATIONS

Another problem with the study is that it ignores other, equally plausible explanations for the data. Consider the possibility that the Republicans in Congress are more right of center than the Dems are left of center. Assume further than many of the think tanks (or at least the ones whose names come up a lot) are relatively centrist. One would expect Dems to cite the think tanks more, and the study would then classify the think tanks as "liberal," which would throw the other results off. I'm not trying to prove this theory, just to point out that a good study must rule out explanations other than its preferred conclusion, and this study doesn't do that. It is at least possible that one party relies on relatively unbiased organizations more than the other party does (who knows, maybe it's the Republicans, but if true, it undercuts a key assumption of the study).

One must approach critically a conclusion that, the authors admit, they were expecting to reach before they even started the study. Media bias is a very difficult subject to study, and I just don't think this study comes up with anything that supports broad conclusions.

Mr. Jacob's second area of analysis in "Data Used," (on placement of stories) is especially interesting to me. I've seen numerous examples where stories that I think should be front page news are relegated to the back pages or omitted entirely. Often, this happens when a story isn't seen as fitting the "narrative" that has already been set up for the parties involved. For example, articles detailing the Bush Administration intense effort to undermine Homeland Security efforts in the post 9/11 era are virtually impossible to find because President Bush has effectively made his "tough on terror" persona the narrative that news stories had to fit within to be seen as mainstream. Jonathan Chait wrote a superb article in The New Republic back in 2003 detailing this very phenomena.

[E]ven when the mainstream media reports on Bush's efforts to limit homeland security spending, they still accept his basic assertion that homeland security is his top priority. Take, for instance, this Washington Post story from October 19, 2002:

[T]he White House appears to have put more emphasis on holding the line on overall spending levels than on winning the spending increases it has sought. The president's high-stakes demand for fiscal discipline in areas he has not emphasized has jeopardized his top priorities. In limbo are billions of proposed dollars to secure the nation's ports and skies, defend against bioterrorism. ... Instead of funding those proposals, lawmakers voted this week to keep federal agencies running at current spending levels until Nov. 22, leaving town with the non-military side of government practically operating as if Sept. 11 never happened. Yet White House spokesman Ari Fleischer sought yesterday to paint the impasse as a Bush victory.

Isn't it just a bit odd that the president would work tirelessly to scuttle his own "top priorities" and then revel in their failure?

Part of this is due to the way newspapers are set up. The New Republic wrote on October 31st 2003:

A Bush announcement is the province of the White House press corps, whose reportage is generally deemed the most newsworthy. Legislative arcana is the turf of less prestigious congressional reporters. So, when Bush announces a new, "compassionate" initiative, the story makes a big splash on the evening news. When it dies in Congress, the story ends up in the back pages of the newspaper.

The reverse is also true, once a narrative gets fixed in the public mind the media will often exaggerate it well beyond its logical bounds. The same (October) TNR article continues:

Once the news media has settled on a perception of a political figure, it becomes nearly impossible to dislodge. One reason is that the evidentiary standards for a piece of "news" drop if that news seems to fit a preconceived pattern. In 1988, for instance, reporters decided that GOP vice-presidential nominee Dan Quayle was stupid. Quayle certainly was not a brilliant man, but plenty of politicians with equally mediocre minds do not have their intellect savaged the way Quayle did. Part of the reason Quayle couldn't shake his reputation was that any tiny gaffe he committed became national news.

.

Finally, the last major problem with the Media is an effort for TOO MUCH non-partisanship, to the point where they don't dispute blatant falsehoods. TNR gives the example of the treatment of Bush's tax cut proposal on the campaign trail:

Democrats cited analyses--using models developed by nonpartisan economists at the Treasury Department--showing that Bush's tax cut would give more than 40 percent of its benefits to the wealthiest 1 percent of taxpayers. Bush responded with an analysis of his own showing that his tax cut would give a mere 22 percent to the top 1 percent. But Bush's analysis--or, more precisely, his economic flunkies' analysis--arrived at this number by excluding the elimination of the estate tax and the top income tax cuts, the very elements that most benefited the rich. In other words, Bush's "analysis" was a deliberate and obvious sham. But the press treated it as legitimate. As The New York Times reported, "the richest 1 percent of taxpayers would get between 22 percent and 45 percent of the tax benefits, depending on how the calculations are done."

At some point in time, one has to wonder when this myth of a liberal media will die. Are there some stories with a liberal slant? Sure. Are there some stories with conservative slants? Absolutely. If anything, the problem with the media is their willingness to go along with preset story lines even when the facts don't justify them. This ends up unfairly tagging all Republicans as heartless, rich white men and all democrats as soft hearted, pacifist pseudo-communists. If the media started reporting things as they are, rather than as the storyline says they should be, maybe this problem will start to go away.