How Bias Works
Over at the Volokh Conspiracy, Tim Groseclose is spending the week guest-blogging his new book, Left Turn: How Liberal Media Bias Distorts the American Mind. In one of the posts he recounts his experience with a distorted LA Times article which made it seem as if the underrepresentation of blacks at UCLA after the passage of Prop 209 was much more dramatic than it actually was. Groseclose attributes this to liberal media bias; some of his critics in the comment attribute it to the time pressures of reporting. As it happens, I think they're both right.
While it's undoubtedly true that some reporters consciously repress facts which threaten their ideological priors, I don't think that's really the issue in most cases. What bias does--in science, in media, in any situation where information is gathered--is affect what questions you ask.
Of course, I don't know that the author of that article thinks affirmative action is swell, and Proposition 209 was a bad idea, one gets the strong feeling from the article that this is indeed the case. What I'm fairly sure of is that she wasn't a supporter of Proposition 209. Because if she had been, she would have looked harder for evidence that things might not be all that dire. She would have discovered, for example, that the number of black transfer students had actually risen, so that the total number of blacks enrolled at UCLA was about the same (up by 2 students, in fact).
I don't know that this is the "right" way to tell the story, or that these are the most important omissions. But Groseclose is almost certainly right that her ideological priors greatly influenced what facts she found, and the stories she told. And at least as presented by Grosclose, that that story did, in fact, objectively overstate the extent of the problem.
I think that she did not report those facts because she didn't know they existed. Unless you know a whole hell of a lot about the California public college system, there was no particular reason to wonder whether almost half of UCLA undergraduates are transfers rather than incoming freshman, or whether the NCAA had changed its rules about academic standards for athletes in a way that reduced the number of African American athletes could admit. The thing is, someone who supported Proposition 209 would have gone looking to see if there was a deeper story here, rather than looking at the obvious superficial metrics.
Before I was a journalist, I used to wonder why journalists were suppressing obvious important facts; after I became a journalist, I realized that it's often incredibly hard to know that there's a fact you're missing. I seem to recall a scathing editorial about mortgage financing in maybe 2005, in which the outraged writer pointed out that banks were booking the asset value of the loans as if they were going to realize 100% of the payments . . . even though some of those loans would go bad! Fraud! Malfeasance!
. . . er, accounting. Anyone who knows anything about accounting knows that what the writer was saying is absolutely true . . . and that if you look on the other side of the balance statement, you find that they have booked a corresponding allowance for bad debts. You can argue that in 2005, those allowances weren't big enough--hell, I think at this point, it's not an argument, but a fact. Nonetheless, the opinion column was ludicrously wrong.
And yet, understandably wrong. I'm sure that the writer had been fed the story by a consumer activist (I can probably name the group). He looked at the financial statements, and the numbers checked out. He'd never heard of such a thing as an allowance for bad debts, so how was he to know it was there?
Bias matters not because liberals deliberately slant their stories, but because they are much more likely to interrogate the facts that contradict their ideological beliefs, than the ones that support them. When they come across an uncomfortable fact, they'll go out of their way to figure out why it isn't really true. When they come across a fact that confirms what they believe, they'll be more likely to accept it at face value.
I'm not claiming that liberals do this more than conservatives (I think that being human, they're equally prone to this phenomenon)--only that in the media, liberal bias is mostly what matters, because the media is overwhelmingly somewhere to the left of the American center. Even if you have a conservative reporter prone to insufficient interrogation of convenient facts, those same facts are going to set off alarm bells with his editors, who are quite likely to question the whole story.
This is why I think liberal media bias is worrisome--not because it's a vast conspiracy, but because it creates a giant store of pseudofacts that "everyone knows", like all those ludicrous statistics about abortion and domestic violence that used to appear everywhere. And new pseudofacts are being created, or resurrected, all the time. Longtime readers of the blog have endured my jeremiads against Republicans who say that cutting tax rates raises revenue. A few weeks ago I was flabbergasted to find out that a very smart left-of-center blogger I admire seemed to think it was common knowledge that there's really no evidence that tax cuts provide stimulus . . . a belief that I then find out seems to be common among some of the left commentariat. This is not at all what any mainstream economist I'm aware of believes (the debate is over the relative size of secondary and tertiary effects of various kinds of stimulus, not whether tax cuts are stimulative at all.) But somehow this belief had become common . . . presumably because in the milieu where it was transmitted, no one was disposed to question something that fit so neatly with what they already believed. Add in a few hundred repeitions from people you like, and voila, "Everyone knows . . . "
This is not a problem that can be overcome by us all resolving to be better, which--along with denial--is the usual response to conservative complaints about the media's skew. Journalistic ethics, a committment to truth, and so forth, are supposed to compensate for the fact that yes indeed, 95% of my peers vote Democratic in almost every national election.
But intention is no substitute for that fiery, almost angry "that can't be right" reaction which drives us to dredge for more information--you'll quickly get a pretty sophisticated rebuttal of Laffer Curve nonsense from a run of the mill liberal blogger, and I'm fairly sure that it wouldn't have taken an ordinary conservative blogger more than half an hour to learn enough Keynesian economics to be able to explain why the same model that shows that deficit spending stimulates the economy, also shows stimulative effects from deficit-financed tax cuts.
But we don't investigate things that everyone knows--reporters do not start off each new story by checking if gravity is still in operation. The more things that everyone knows, the more unnoticed holes there will be in stories. And there is no one without blind spots--the best you can achieve is getting together a bunch of conscientious people who all have different blind spots.
For all the laments over partisan media, there are actually ways in which it is a check on this sort of blindness. Of course there's no guarantee that at any moment, the partisans raising questions will be right--and often they aren't. So of course they're reviled by people who are sick of contesting, over and over, questions that, um, everyone knows the answer to.
But the process itself is healthy--yea, even though it frequently produces its own misinformation. A while back, a friend of mine asked what he called an "epistemological question" about the journalism of the financial crisis. A lot of things, he pointed out, are reported only once, and then they become the basis of other reporters' stories. How, he asked, do we prevent wrong "facts" from entering into the data stream and corrupting our understanding of what happened?
I had two answers, neither of them perfectly satisfactory. The first is that journalists are competitive, and overturning, say, Andrew Ross Sorkin's account of an important meeting is a pretty major journalistic coup. But the second is that this is actually what ideological journalism is good for: when facts appear that contradict peoples' world views, partisans frequently go out and exhaustively investigate those facts to see if they can be in any way mitigated or disproven.
Both of these things have ugly side effects, of course. But in the end, I think that both of them provide the only check we have against the tyranny of the things we all know that ain't so.