Thursday, November 09, 2006

NSF-Funded Study Finds Newspaper “Slant” Comes from Readers, Not from Owner

Once again, does ownership matter? A new, comprehensive, methodically unbiased study from two academic economists asks the question: “What Drives Media Slant?” Using robust statistical tools and a novel approach to measure “slant,” this National Science Foundation-funded study found that the largest single variable is that “Firms respond strongly to consumer preferences.” That is, to the extent there is a bias—or slant—in coverage-- newspaper editors tend to write for their audience. Bias comes from below, not above—from owners.

Whether it’s at the FCC hearings on ownership rules or at stacked conferences, such as National Conference for Media Reform, the rhetoric behind media ownership has been one of diversity and the implication—supported primarily by anecdotal stories—that large media companies stifle diversity. There is the subtext that the owners of the large media companies could -- if not actually do—promote their personal, political and/or cultural agendas.

This is despite the preponderance of evidence from sources that are not stakeholders or don’t have ideological axes to grind, that ownership matters, but not in the direction these advocates promote. As I have pointed out here before, describing a study undertaken at Harvard’s Kennedy School that finds the US has the most diverse and competitive media environment in the world, or the British-lead “Trust in Media” study that had a similar finding. The very well conceived content analysis study, “Does Ownership Matter in Local Television News?” from the Project for Excellence in Journalism reported very mixed, sometimes counter-intuitive results, such as that local TV stations with cross-ownership—in which the parent company also owns a newspaper in the same market—tended to produce higher quality newscasts.

The “What Drives Media Slant?” study is unique in several ways. Key is the method for determining slant: instead of researchers constructing a basket of content criteria and setting a handful of graduate students to laboriously try to categorize thousands of newspaper articles over a manageable period of time—often as few as two weeks-- Matthew Gentzkow and Jesse M. Shapiro draw their definition of slant directly from the words of ideologues—members of Congress. They started with a computer-driven analysis that examined the set of all two and three word phrases used by representatives in the 2005 Congressional Record, and identified those that were used much more frequently by one party than by the other.

For example, they discovered that when referring to estate tax legislation, Republicans tended to use the term “death tax,” while Democrats were partial to calling it the “estate tax.” Republicans were biased toward “tax relief,” Democrats liked to call it a “tax break.” And so it went: “personal account” for Social Security change and “war on terror” (strongly Republican) vs. “private account,” and “war in Iraq” (strongly Democratic).

The researchers were then able to take the 1000 most strongly identified terms and use computers to compare them to the full content of 400 newspapers accounting for 70% of daily circulation—far more than is possible using conventional content analysis. The key finding: “Using zip code-level data on newspaper circulation, we show that right-wing newspapers circulate relatively more in zip codes with high fractions of Republicans, even within a narrowly defined geographic market.” And by implication, these papers circulate less strongly in areas of a left leaning constituency.

The study further found that newspaper companies with publications in multiple markets followed the same principal: their papers published in “red” markets showed more of a Republican slant than those published in Democrat “blue” markets.

The political composition of the market did not account for all the explanation for slant, the study found, but it was the largest identifiable variable. The study was robust in that it controlled for many possibilities that could account for slant, but none undermined the central finding: bias is driven more from the readers, less from the editors or owners. In economist language, newspaper content is largely demand driven.

Advocates who claim to “know” that owners—often large media companies—set an editorial agenda that is their own invariably use anecdotal stories or a narrow, limited study to “prove” their point. Or they advance their position simply on faith. The empirical evidence keeps piling up that overall ownership does matter—but not in a single nor predictable direction.

Link to this entry
E-mail this entry

No comments: