It would seem a fairly mischievously buried lede. The story you expect from the headline “Old Friends Say Drugs Played Bit Part in Obama’s Young Life” is the story of, well, drugs playing a bit part in Barack Obama’s young life. Turns out, as we don’t learn until the sixth paragraph of the article, the real news is this:

Mr. Obama’s account of his younger self and drugs [in his memoir Dreams From My Father] significantly differs from the recollections of others who do not recall his drug use.

Or, as the New York Times reporter who filed the story, Serge V. Kovaleski, puts it a little further on:

In more than three dozen interviews, friends, classmates and mentors from his high school and Occidental [College] recalled Mr. Obama as being grounded, motivated and poised, someone who did not appear to be grappling with any drug problems and seemed to dabble only with marijuana.

In other words, the story is more about the bit than it is about the drugs.

To be fair, it seems that someone at the NYT recognized the problem (albeit too late for the print edition) since Kovalevski’s story subsequently appeared under several different versions of the headline in the online edition of the paper, including:

+ Friends Say Drugs Played Only Bit Part for Obama [my emphasis]

+ Few From Obama’s Youth Remember His Drug Use

+ The young Obama: A time of alienation and discovery [in the International Herald Tribune]

If you’re an Obama supporter, as I am, and if you’ve noticed, as I have, the relative favor that Hillary Clinton seems to enjoy in the Times’s news pages, an example like this can’t help but send you springing to conclusions. Not even when it happens that those conclusions—most specifically that the wall between news and opinion is not as unyielding as the Times would have us believe—coincide with the conclusions that many conservative commentators have been making for many years. (And not only them: see former NYT Public Editor Daniel Okrent on the subject here.) The editorial page supports Clinton in the primaries, this thinking goes, and so consciously or not there will be a pro-Clinton bias in the paper’s news coverage.

But what if the bias has less to do with the people that produce the paper and more to do with the people that read it? What if editorial bias is a response to reader bias, rather than an attempt to create it?

That’s the conclusion of a new study by U. of Chicago professors Matthew Gentzkow and Jesse M. Shapiro. In their research for “What Drives Media Slant?” Gentzkow and Shapiro looked at news articles and opinion pieces in 433 U.S. newspapers and found that the political slant of any particular paper could be predicted by the political affiliation of its primary readership. What’s more, the political slant of a newspaper was calibrated to bring in maximal profit. As quoted in the January issue of the U. of C.’s Capital Ideas magazine, Shapiro said,

The slant newspapers choose is very close to what we would predict if all they were trying to do is maximize their circulation in their geographic market.

According to the study itself, the political preferences of a newspaper’s readership accounts for about 20% of the paper’s own bias. By contrast, when the study’s authors considered multiple newspapers that shared an owner, they found that the owner’s political bias had no discernible effect on the paper’s bias.

The most entertaining part of the study, as is often the case with these things, is the numbers. The authors found that if you chose a person at random from the geographic areas they studied and gave her a newspaper to run, it would cost her, on average, $2.34 per reader per year to favor her own bias instead of her readers’. (More exactly, to bring the paper’s bias one standard deviation closer to her own.)

The Gentzkow/Shapiro study dealt with obvious political differences—whether a local population contributed more to Democrats or Republicans, whether a newspaper used the phrase “estate tax” or “death tax”—but there’s no reason to think that their conclusions don’t also apply to the hairline fracture that separates Clinton and Obama. And thus the undeniable bias in the original headline of the Obama story might very well have less to do with dictates from above than it does with the dictates from below.

++++++++

Bonus paragraph: Another interesting finding from the study is that newspaper readers appear to skew left of the general population: “We fi…nd that the average pro…fit-maximizing point is also to the left of the average congressperson, and considerably closer to the average level of slant we observe. One possible explanation is that the consumers with the highest propensity to read— or whose readership is most sensitive to slant— tend to be to the left of the median voter.”