Media Bias is A Business Decision?

This won't be news to Rupert Murdoch. Here's Austan Goolsbee in the Times:

New research by two University of Chicago economists, Matthew Gentzkow and Jesse M. Shapiro, entitled "What Drives Media Slant? Evidence From U.S. Daily Newspapers" (www.nber.org/papers/w12707.pdf) compiles some compelling and altogether unusual data to answer the question.

Dr. Gentzkow and Dr. Shapiro started in the world of the political. They parsed the words of politicians -- all the words -- from the 2005 Congressional Record. They found the 1,000 most partisan phrases uttered in the year. They measured this by comparing how frequently a phrase was used by one side or the other.

In 2005, phrases like "death tax," "illegal aliens," "Terri Schiavo," and "nuclear power" came mostly from Republicans. Phrases like "minimum wage," "public broadcasting," "middle class" and "oil companies" came mostly from Democrats. Using those phrases, the two economists made a simple index of partisanship that comported nicely with standard measures like a politician's score on the Americans for Democratic Action ideological scale.

The study then analyzed 417 newspapers in the United States (accounting for some 70 percent of total newspaper circulation) as if they were politicians. The researchers measured, for example, all the times in articles about Social Security that a newspaper referred to "personal accounts" (Republican) or to "private accounts" (Democratic). Their measure of partisan slant came only from the news coverage. They did not include anything from the editorial page.

The index matched most popular perceptions of newspaper partisanship. Papers like The Washington Times or The Deseret Morning News of Salt Lake City used Republican phrases while papers like The San Francisco Chronicle and The Boston Globe used Democratic ones.

But more important, once the authors had this measure, they showed that the main driver of any slant was the newspaper's audience, not bias by the newspaper's owner.

The authors calculated the ideal partisan slant for each paper, if all it cared about was getting readers, and they found that it looked almost precisely like the one for the actual newspaper. As Dr. Shapiro put it in an interview, "The data suggest that newspapers are targeting their political slant to their customers' demand and choosing the amount of slant that will maximize their sales."

My girlfriend, who's a newspaper reporter for a liberal paper, insists that the study is flawed, since it assumes that any language which isn't used by Republicans is automatically "liberal". She points out that the estate tax is a more "fair and balanced" term than the "death tax," which is just a conservative framing device. While I agree with that point in general - Republicans have a knack for pithy rhetoric - it's worth noting that many of the phrases studied by the economists (tax break vs tax relief, oil companies vs. nuclear power, private vs. personal accounts, etc.) can be said to accurately represent some sort of bi-partisan reality. (For example, Bush's Social Security plan proposed accounts that were both private and personal.)

I'd argue that reporters favor a certain terminology because that terminology tends to generate the connotations that fit their larger political belief system. In other words, the slant of certain newspapers really represents an unconscious bias on the part of reporters, not a conscious business decision by editors or executives. (Of course, there are exceptions to this rule. For example, Murdoch has made himself rich by intentionally exploiting an underserved conservative audience.)