Wednesday, March 12, 2008

Rashomon With a Vengeance

I have been struck recently by how differently Clinton supporters and Obama supporters view news coverage of the campaign. Clinton supporters are absolutely sure that the news coverage is strongly anti-Clinton, and often in a sexist manner. Meanwhile, Obama supporters see just the opposite: For example, a front-page NY Times story treats as breaking news the fact that Obama's first book admitted occasional cocaine usage in high school and college, with the body of the story going on about how his contemporaries don't recall this being so. What a lame excuse to tout cocaine usage. Or take the fact that today's front page of the NY Times has no story on Obama's victory yesterday in Mississippi, buying into the Clinton narrative that states with a lot of black people (or a lot of highly educated people or whatever) don't "count." I don't bring this up to rebut the Clinton camp's arguments so much as to confirm what psychologists have long known: People tend to see bias against their own view. (The classic study was of fans of Dartmouth and Princeton football. Each thought that the referees were unfairly biased against their team.)

We might thus say that there is a pronounced Rashomon effect to the news coverage: Each side perceives bias against its favorite. But now I want to float a conspiracy theory that, if true, would suggest the Rashomon effect is even greater than at first appears. I noticed this morning that while the front page of the NY Times print edition did not cover the Obama Mississippi victory, the front page of the Times on the web prominently displayed it. (Link here to what appeared on p. 23 in the print edition.) This is NOT a matter of more space on the website, since the Mississippi story was among the top web stories for many hours. Nor is it a matter of the paper having to go to press before the story was done, because it's the same story in the print and web editions. Instead, let me suggest my conspiracy theory (for which I have NO other evidence.)

For some time now, more people have read the Times online than in print, although the two audiences overlap some and may read the Times with different levels of interest. In any event, the online readers undoubtedly skew a lot younger, while the paper readers skew older. Thus, online readers skew toward Obama, paper readers toward Clinton. And so what we may have here is a policy of slanting coverage in web and paper versions to appeal to readership. In other words, the coverage that Clinton and Obama readers see as slanting against their respective favorites is actually slanted (relative to an admittedly arbitrary baseline) in favor of their respective favorites. If web readers were to switch to paper and vice versa, claims of bias would increase.

As I said, this is a generalization from one data point, so take this theory---wild speculation might be a more accurate description---with a pillar of salt.

Posted by Mike Dorf