Measuring Media Bias

Media bias is mostly expressed through omissions of inconvenient facts.

Nat Brown
Tuesday, July 19, 2011

That the mainstream media have a liberal bias is hardly a belief limited to those on the right. Indeed, many liberals and even media personalities themselves will readily agree. But is it possible to scientifically measure media bias? Can one statistically prove its existence and determine precisely how strong it is? In his new book, Left Turn: How Liberal Media Bias Distorts the American Mind, UCLA political-science professor Tim Groseclose attempts to do precisely that.

Groseclose first explored this idea in an article he co-authored with University of Missouri professor Jeff Milyo in 2005. The article, which concluded that there was an overall liberal bias in the media (no surprise), initially attracted harsh criticism from many on the left, including Media Matters. In an interview last week with National Review Online, Groseclose recalled the initial reaction to those 2005 findings: “You get hate mail. I have to say, media bias generates the worst in people.” In spite of that, he decided to expand the study into a book, and the result makes for a read that’s both enlightening and entertaining.

Though Groseclose’s exact method of measuring media bias is somewhat complex, he does his best to explain it in layman’s terms. “It all has to do with comparing media reports to congressional speeches. The thought experiment is to ask what would happen if I gave you a set of articles from a news outlet like, say, the New York Times, and instead of telling you they were from the New York Times, I said they were speeches by a politician. If you were to read them, what would be your best guess of the ideology of that politician? That’s the whole idea behind my method.” More specifically, he derives his measurements by taking a list of the think-tank citations that a given media outlet makes, and putting it into a computer program that treats the citations as if they had come from political speeches and tabulates the results.

Groseclose gauges how liberal or conservative an individual is by using what he calls a political quotient (PQ). Simply put, a person’s PQ is a number on a scale of 0 to 100, with 0 being the most conservative and 100 being the most liberal. (For example, as Groseclose writes in his book, Michele Bachmann and Jim DeMint would have PQs of approximately 0, whereas Nancy Pelosi and Barney Frank would score ones of approximately 100.) Groseclose opted to use issues chosen by Americans for Democratic Action, a liberal interest group, to determine the score. He also uses a scale of 0 to 100 to measure how liberal or conservative a given news outlet is, in this case calling the number a slant quotient (SQ). Since the two scores are directly comparable, Groseclose can accurately describe the extent to which a given news program or publication leans to the left or right of a given politician.

In Left Turn, as in the original article, Groseclose measures 20 major news outlets in the period 1995 to 2004. Which came out on top as the most liberal? The answer might surprise you. Although the New York Times, CBS Evening News, and the Washington Post all scored on the left, the highest slant quotient went to the news pages of the Wall Street Journal, which scored 85.1 (this was before Rupert Murdoch’s News Corp. purchased the paper). “Dow Jones & Company gave a press release denouncing our study, but we stand by it,” he explains. “And since then we’ve had a number of people come up to us and say that that’s been the best-kept secret out there, that although the opinion pages are pretty conservative, the news pages are actually liberal.” In comparison, the New York Times scored 73.7, CBS Evening News also scored 73.7, and the Washington Post came in at 66.4. Which outlet was the most conservative? The Washington Times, which scored 35.4, measured farthest to the right (the next-lowest score was Fox’s Special Report with Brit Hume, which scored 39.7). These news-outlet scores aside, perhaps the most startling finding of Groseclose’s study is the statistic concerning the bias of the average Washington correspondent. “I think the most important fact to know about media bias is that Washington correspondents vote about 93–7 for the Democrat in the typical presidential election. That is not just high, that is statistically significantly higher than 80–20 or 70–30.”

Groseclose finds that media bias is rarely expressed through distortion of the facts, but rather through the omission of certain facts that would be inconvenient for the outlook of the person or group reporting. On the issue of the Bush tax cuts, for example, Groseclose finds a concrete example of when media bias shaped the debate. “All the liberals were saying that under this tax plan the very rich are going to get a highly disproportionate share of the benefits, and that was true — something like the top 1 percent would get 30 to 35 percent of the total tax cuts. But one fact a lot of media people didn’t mention is that the reason that was true is that the very rich pay most of the taxes. When Reagan did his tax cut, he just said, ‘We’re going to take whatever you pay in taxes, and we’re going to multiply it by 0.75, and everyone’s going to get 25 percent off what they usually pay.’ Bush didn’t do that. He said, ‘We’re actually going to have a bigger percentage cut for the poorest people.’ What that meant was that after taxes were cut, the rich would actually be paying a higher percentage of overall taxes than before. The Bush tax cut actually made the tax more progressive. So there are two facts that are equally true, but one was the fact that all the liberals were talking about, and the other was the fact all the conservatives were talking about.”

While the scores of each outlet are of interest, what’s really interesting (and frightening) about Groseclose’s book is his exploration of the effect this bias has had on the American voter. “I estimate that the average voter in the U.S. has a PQ of about 50, the views of Arlen Specter when he was a Republican,” he notes. He finds that media bias has a significant effect on this score, and concludes that “the slant of the media has moved the average voter about 20 or 25 PQ points to the left.”

So what would the average PQ of a voter look like if all media bias could be somehow magically removed? Groseclose’s answer: something close to Ben Stein. “To me, he’s about halfway between a Michele Bachmann and a Ben Nelson, the Democratic senator from Nebraska. If that’s true, Ben Stein has a PQ of about 25, and that’s what I estimate would be the average voter’s views if there were no media bias. It basically means that on controversial issues before Congress, Ben Stein would vote on the conservative side about 75 percent of the time, and 25 percent of the time on the liberal side.” That’s a long way from Arlen Specter, and Groseclose emphasizes the point. Without liberal media bias, “the average American voter would start thinking and voting about like the average Texan voter.”

If only.