The ideology of the owners doesn't correlate in any significant way with the political slant of their newspapers' coverage. When a single owner owns multiple papers, the authors find that each paper's language is tailored to its own market, rather than toeing a single, corporate line.
Diversity of the press has been a goal of American government since Thomas Jefferson - and in this era of newspaper consolidation, it is again an area of concern. Americans need a range of viewpoints on the news to maintain a healthy democracy, according to the theory, so newspapers should remain in the hands of people with different outlooks and backgrounds. Though seemingly simple, this theory has proved surprisingly difficult to test. At the heart of the matter is a lack of an objective way to quantify the political positions of different news outlets.
In What Drives Media Slant? Evidence From U.S. Daily Newspapers (NBER Working Paper No.12707), Matthew Gentzkow and Jesse Shapiro propose a solution to this measurement problem, developing a new technique for measuring the political slant of a news outlet. They go on to use their measure to argue that readers, not owners, play the most central role in determining a newspaper's slant. Media slant is a hard object to pin down, of course. To quantify it, the authors began by using data from the 2005 Congressional Record to identify the phrases that were most partisan, in the sense of being used much more often by members of one party than by members of the other. They came up with 1,000 such phrases, including "tax cut for the wealthiest" (used much more by Democrats than by Republicans) and "death tax" (used much more by Republicans).
They then counted the use of these partisan phrases in the news text of each of over 400 U.S. daily newspapers, and computed, for each newspaper, what type of congressperson (Democrat or Republican) uses language most similar to that of the newspaper. Not surprisingly, they found that the language used by liberal congressmen also found its way into papers such as The New York Times, Los Angeles Times, and Washington Post. The language of conservatives was more apt to show up in the Washington Times and The Wall Street Journal.
An illuminating example is the partisan divide over the tax on inherited wealth. In 2005, congressional Republicans, who generally oppose the tax, described it as a "death tax" 365 times, using the term "estate tax" only 46 times. Democrats did the reverse, saying "death tax" 35 times and "estate tax" 195 times. Similarly, the more liberal Washington Post used "estate tax" 10 times more often than it used death tax; the more conservative Washington Times used "estate tax" only twice as often.
Using the full range of phrases and newspapers, the authors find, contrary to conventional wisdom, that the ideology of the owners doesn't correlate in any significant way with the political slant of their newspapers' coverage. When a single owner owns multiple papers, the authors find that each paper's language is tailored to its own market, rather than toeing a single, corporate line. Their data also show no significant relationship between a newspaper's slant and the political contributions made by its corporate owner. What instead has a big impact on newspaper bias is readers. The study found that the political outlook of a paper's readers explained about 20 percent of the variation in slant that the authors uncovered. No other factor showed such a strong correlation.
The reason for this is that owners find it more profitable to reflect the views of their readers than to impose their own perspective, the authors conclude. And, most of the newspapers studied were close to the ideological "sweet spot" that would maximize their profits, the authors calculate. Even a small deviation from this ideal bias would cut circulation by some 3 percent, so newspapers hew closely to the ideological stance that makes them the most money. "Our work shows that consumers play a fundamental role in determining the ideological positioning of media outlets," the authors write.
-- Laurent Belsie