A Difficult Task

A few weeks ago, I posted my recent article from Law and Contemporary Problems. The article was part of a special issue, drawn from presentations at the September 2007 Duke Law School conference.

As often occurs at academic conferences, a few presenters offered contrarian viewpoints. The most extreme came from BC Law School Dean Michael Cassidy, who criticized portions of the disbarment ruling against Mike Nifong. Cassidy’s thesis: Nifong’s status as a candidate for election gave him a First Amendment right to utter many of the public denunciations of the lacrosse players that the DHC had ruled unethical.

Cassidy’s thesis certainly challenged conventional wisdom. But the dean’s proposed “candidate exception” to the Bar’s ethical canons has attracted scant, if any, support.

Another LCP article, penned by George Washington professors Robert Entman (a former Duke faculty member and parent of a recent Duke graduate) and Kimberly Gross, also challenges the majority view on an element in the case. As Entman summarized for me, the article contends “that, ironically, just as the coverage created misperceptions about the alleged crime, the coverage itself has been subject to distortion and selective memory. The media did not persecute the lacrosse players, at least not for very long, and there is no evidence in this case or more generally that ‘political correctness’ leads to more negative treatment of white than black defendants, or more credulous treatment of black than white accusers.” Entman added that his piece represents “the only systematic analysis of media coverage on this explosive case.”

Previous publications by Gross and Entman reflected prevailing academic assumptions on race-based issues. Gross, who teaches a course entitled “Race, Media, and Politics,” has published articles concluding that “there is a specific racial cast to [local TV] coverage – significant attention is given to black perpetrators while black victims are undercovered.” Entman’s 2000 co-authored book, The Black Image in the White Mind: Media and Race in America, posited “a subtle pattern of [media] images that, while making room for Blacks, implies a racial hierarchy with Whites on top and promotes a sense of difference and conflict.” (The book received an approving mention from Mark Anthony “thugniggaintellectual” Neal.)

The scholarly framework of Entman and Gross cannot easily explain an event in which—amidst at first considerable and eventually massive public evidence of actual innocence and prosecutorial misconduct—most coverage first presumed the guilt and then minimized evidence of innocence of wealthy, well-educated white defendants accused of a crime by a poor African-American woman.

---------

In their article, Entman and Gross analyze three newspapers (the N&O, the New York Times, and USA Today) and one television network (NBC), based on a search for words or phrases that implied guilt, innocence, or positive/negative judgments. (They do not code the cable news shows, on the grounds that such figures aren’t real journalists.) They conclude that while a media rush to judgment did occur, it was motivated not by political correctness or journalists’ presumptions that women who claim rape must be telling the truth, but instead the normal press bias in criminal justice cases, which favors the prosecution. In this respect, they contend, “most pretrial publicity is predictably prejudicial, and media behavior in the Duke lacrosse case was not unusual but typical.” Some people might consider such things as a fullpage Newsweek cover and massive coverage in the New York Times atypical in crime cases, but such editorial judgments don’t factor into the Entman/Gross schema.

Moreover, Gross and Entman cite a consistent coverage of the players’ possible, or even likely, innocence—an approach that grew stronger as the case progressed. Finally, their data reveals little difference between the local coverage of the N&O and the national coverage of the New York Times, other than expected gaps on most news stories between more nuanced local outlets and national publications that focus more on metanarratives. In any event, they deny that political correctness motivated either publication (or USA Today).

Some of the Gross/Entman conclusions confirm, rather than challenge, conclusions in both the book and blog. For instance, neither the blog nor the book ever contended that (apart from Samiha Khanna’s brief moment in the sun, early in the case) “political correctness” motivated the N&O’s coverage. Indeed, I strongly praised the N&O’s journalism, some of its editorials, and most of its op-eds. Yet a reader of the Entman/Gross article (which is framed, in part, as a critique of DIW and UPI) would have assumed just the opposite.

Indeed, consider this quote: “Over the entire period editorials were more likely to contain paragraphs emphasizing only innocence than they were to contain paragraphs emphasizing only guilt . . . This finding offers another indicator that neither a ‘politically correct’ nor a liberal editorial agenda was driving the coverage.”

That evidence sounds like a strong critique of UPI and DIW—until realizing that most of the editorials and the op-eds in the Gross/Entman study came from the N&O, which both the book and the blog had praised. Moreover, a central critique of the Times editorial page was not the content of the editorials it did run (zero, in all), but its decision not to weigh in on the case when all other major newspapers did, in late December 2006.

Or take footnote one of the Entman/Gross article cited both UPI and a DIW post as among the critiques “that trace media derelictions to ‘liberal bias’ or ‘political correctness.’” But neither the blog (in any of its 1300 posts) nor the book ever used the phrase “liberal bias” in this context—and the blog argued that political ideology provided no guide to predicting how actors in the case would behave. In response to an e-mail query, Entman conceded, “We have liberal bias and political correctness in quotes not meaning to attribute them as exact quotations of you or other authors. We meant them in the ‘so-called’ sense. It was probably unnecessary to have done this.”

So, Entman and Gross established that the N&O’s coverage was basically fair; that most editorials published in the N&O leaned toward stressing innocence; and that a “liberal bias” doesn’t explain coverage. Since neither the book nor the blog challenged these arguments, on these matters, Gross and Entman are arguing against straw men.

A few other conclusions are a bit on the vague side: take, for example, Entman’s assertion that “the media did not persecute the lacrosse players, at least not for very long.” Leaving aside the fact that neither DIW nor UPI ever used the verb “persecute” to describe the media’s handling of the case, the clause contradicts itself: either the media “persecute[d]” the lacrosse players, or the media did not do so. The first half of Entman’s clause takes one position on the question (the media did not persecute the lacrosse players); the second half takes a completely different position (the media did persecute the lacrosse players—but “not for very long.”)

-----------

The Gross/Entman data model had two serious drawbacks, one avoidable, the other not. First, the avoidable item: the decision to bypass the Herald-Sun, which published more articles on the case than any paper except perhaps the N&O and whose editorial page read as if penned by Nifong himself. When asked about the thinking behind this decision, Entman explained, “We picked the N&O because it is a more influential paper in the Triangle and NC, though probably not in Durham proper. Also as you have suggested (and I as a resident of Durham off and on for 40 years have said), the Herald is a lousy paper and performed up to that standard in covering the lacrosse case. Yet it’s never been in those 40 years liberal editorially, indeed often endorsed a Jesse Helms view of the world, and therefore the Herald illustrates our basic argument that journalists’ worldview was only one among many factors shaping the coverage.”

The fact that the paper was conservative before Paxton Media bought the H-S and installed Bob Ashley as news and editorial page editorial editor is irrelevant to analyzing the H-S position on the lacrosse case. (The H-S editorial page tilted well to the left in 2006 and 2007--its positions often mirrored those of The Independent--and, of course, it endorsed Nifong protégé Tracey Cline in 2008.) Since Entman and Gross wanted to challenge those who portrayed the media as “politically correct,” it’s hard to understand why they analyzed the local outlet whose performance was widely praised as not motivated by political correctness—while ignoring the local paper whose articles and editorials were attacked as drenched in politically correct tenets.

Second, the article’s unavoidable problem: no statistical model can address editorial decisions that were key to any understanding of the media’s role in the case—such as Newsweek’s running a full-page cover with mugshots of Reade Seligmann and Collin Finnerty, under the title, “Sex, Lies, and Duke.” (The only “lies,” of course, came from Mike Nifong and Crystal Mangum; there was no “sex” at all.) Nor can the Entman/Gross model shed any light on what accounted for the editorial choices made by the New York Times, ranging from the critical decision to replace Joe Drape with a reporter more inclined to accept Nifong’s claims, to the August 2006 assignment of Duff Wilson to write an article that supposedly would reconsider the point of view offered in his own earlier articles, to the refusal to run corrections on the factual errors in that August 2006 article, to the more general decision to place such a high priority (more than 100 articles or columns in all) on a crime story in North Carolina.

Entman maintains, “There is no evidence in this case or more generally that ‘political correctness’ leads to more negative treatment of white than black defendants.” Yet editorial judgments shown in contemporaneous or near-contemporaneous cases suggest otherwise.

For instance, the lacrosse case wasn’t the only instance of high-profile college athletes being accused of sex crimes (USC, Cal-Berkeley, Oklahoma State) or an allegation of a racially motivated gang rape by college students (VUU/Richmond). Yet none of these cases—in which either Hispanic or black students were the accused, and in which a white woman was the accuser in at least two of the cases—attracted a media firestorm characterized by heavily moralistic coverage, or generated much coverage in the New York Times, or resulted in a guilt-presuming Newsweek cover, or received much attention on cable or network news.

Likewise, in 2006-7, the lacrosse case wasn’t the only high-profile, racially charged event in the criminal justice system. It coincided with events in Jena, Louisiana, in which African-American students were accused of attacking, without provocation, a white fellow student. In covering Jena, the national media, led by the same New York Times, stressed allegations of possible racial bias in the decision to prosecute, and presented the accused students in largely favorable terms. Only when it came out that the ringleader of the attack had previous convictions for assault (of other African-Americans), followed quickly by the ringleader’s guilty plea, did the media narrative somewhat change. The volume of coverage also dramatically declined.

None of these developments would have been predicted by the model that Entman and Gross present.

---------

Even though Gross and Entman can’t analyze editorial decisions, their article, despite the superficial neutrality associated with a “statistical” analysis, employs some of its own editorial judgments—all of which seem to tilt toward one direction. For instance, in trying to explain why the media didn’t greet as conclusive proof of innocence the April 10, 2006 negative DNA tests, Gross and Entman bend over backwards to offer a neutral justification: “Because this DNA evidence did not mark an unraveling of the case for the prosecutor, journalistic routines and norms ensured that it did not mark an immediate unraveling of the case for journalists. Rather, reporters and editors apparently assumed that Nifong had other good evidence for proceeding with the indictments, and this colored their reporting.”

Yet in this specific case, the prosecutor’s office itself had, only seventeen days earlier, obtained an extraordinary court order—backed up by several Nifong public statements—that stated without caveat that the tests would provide conclusive evidence of either guilt or innocence. Most of the time, journalists show at least a smidgen of skepticism when government figures display an on-the-spot, 180-degree reversal of a previously held, public, position. Moreover, contrary to the expected performance in the Entman/Gross model (that journalists tend to be faithful to government sources they need), the paper that increasingly challenged Nifong was the local one (the N&O), not the national one (the Times), even though the Times, presumably, had no need to cultivate a permanent journalistic relationship with the Durham DA’s office.

Or take a passage, discussing the DNA evidence, describing a “second test on the accuser’s fingernails that had a match with the third player indicted.” Of course, no such “match” existed. The discredited Dr. Brian Meehan claimed that Dave Evans’ DNA, along with the DNA of two percent of all the males in the United States, couldn’t be excluded as a possible match, even though the State Lab had made no such judgment, using the same sample.

Or take a more subtle form of bias: Gross and Entman’s decision to reference by name, in the footnotes, the falsely accused Dave Evans, Collin Finnerty, and Reade Seligmann, even as they declined to identify Crystal Mangum. When asked why the duo pursued this strategy, Entman replied, “We didn't make a conscious decision not to name her. We probably should have.”

This sympathy for Mangum appears to have carried over into some of Gross and Entman’s coding decisions. Take, for instance, the following paragraph:

The damaging nature of certain keywords such as “single mother” and “stripper” must also be considered in any calculation of the balance between coverage of the defendants and coverage of the accuser. And this is a good example of the complexity of trying to decipher the polysemic nature of the coverage and its effects. It is unclear, for example, whether denunciations of the lacrosse players as hooligans caused any more damage to their reputations—no matter how subjectively painful these characterizations may have been to the players and their families—than being labeled as an unmarried mother and stripper damaged the reputation and credibility of the accuser.

There’s only one problem with this argument: Mangum was, in fact, an unmarried mother whose sole source of income came from stripping (or prostitution). The media, obviously, couldn’t falsely describe Mangum as married, nor could journalists falsely claim that she had another profession—say, computer programming. The accused lacrosse players, on the other hand, were not “hooligans,” unless the word is defined to apply to any college student who ever engaged in underage drinking or attended a tasteless spring break party—which would render the epithet all but meaningless.

Or take the following passage:

There is some support for the claim that early coverage contained information that reinforced impressions of Duke lacrosse players as spoiled, privileged, and loutish. There are references to prior misdemeanor charges, quotes from neighbors and professors, and discussions of the team’s collective reputation that can be characterized as undermining the individual defendants’ reputations . . . Although our analysis does not comprehensively evaluate this dimension of the coverage, some suggestive evidence comes from a search for the terms “record,” “criminal,” “violat[-e, -ing, -ion, -or, -ed],” or “assault” . . . Viewing this particular dimension as a frame contest, the score would be eight to forty against the defendants. On the other side, however, the news also offered many positive character witnesses testifying to the decency of these young men and relatively few testimonials to the character of the accuser. [emphasis added]

Again, the media can’t produce items wholly out of thin air: the fact that virtually no one who knew Mangum, including her fellow students at NCCU, had anything positive to say about her meant that journalists couldn’t offer positive testimonials to Mangum’s character. (From the start of the case, the NCCU campus paper, the Khanna period at the N&O, and national magazines like Vanity Fair tried to find people to say good things about Mangum.) The inability of most people in the media to find anyone willing to praise Mangum doesn’t explain data containing far more references to the players’ (minor) alcohol-related arrests than to Mangum’s arrest for robbery and trying to run over a police officer.

Indeed, even as the duo faults the media for not using the case to explore the more general problems of the criminal justice system, Entman and Gross essentially give a pass to the faculty “activists” and civil rights groups that might have been expected to have advanced such an argument at the time. “The apparent story of privileged whites victimizing a black woman,” they write, “fit larger cultural narratives, making it almost inevitable that third parties would insert themselves into the struggle over media framing. The use of this high-visibility case as a platform for advancing more-general political agendas is also nothing unusual and, leaving aside the particulars of the case, is not even undesirable. After all, we are constantly being told that the United States enjoys a vigorous marketplace of ideas, so there is nothing wrong with spokespersons for various ideological points of view seizing on big news stories to promote their political interpretations and preferences. What was obviously problematic in this case was that these spokespersons tended to treat the defendants as guilty and used that conclusion as a basis for their larger political arguments about white racism and privilege. The weak factual basis of this particular case does not in itself logically preclude the possibility that their arguments to the more general points were valid. [emphasis added] A large body of evidence supports the view that white privilege and white racism do continue to oppress African Americans, particularly poorer African Americans. And there is a sound basis in this particular instance for suggesting that Duke University is a relative bastion of white privilege in a city where many working-class whites, blacks, and Latinos struggle to make ends meet and to receive adequate educations, job opportunities, health care, and other privileges that most Duke students (and professors) take for granted. In the context of increasing income- and wealth inequality nationally, and of continued black–white disparities across an array of indicators, the Duke lacrosse story, as it first seemed to be, was a reasonably compelling symbol.”

Perhaps I’m naïve, but it seems to me that when a preconceived worldview causes someone to erroneously rush to judgment and make highly public, moralized condemnations of innocent people, it’s time to ask what faults in that preconceived worldview caused such a misjudgment. Yet there’s no evidence that any of the Group of 88 (or similarly-inclined figures in the media) have engaged in any such critical self-reflection.

---------

At its heart, though, the fundamental problem with the Gross/Entman article comes in its pedagogical approach: in the lacrosse case—an instance in which, ultimately, there were not “two sides” to the story—statistical analysis can obscure as much as it illuminates. Take the example of Duff Wilson’s August 25, 2006 New York Times article. According to Entman, “The stories were coded at the paragraph level: 23 grafs coded as containing info suggesting guilt only, 26 coded not guilty only and 14 coded as both being present in the paragraph. This may appear more balanced than it would using a qualitative approach, but that's what the data say. Nonetheless, there's no disputing that such balance is problematic given that it appeared long past the point at which there should have been much credence to not guilty claims. However this piece was exceptional in this respect, not typical, as our data reveal.”

Recall that this nearly 6000-word, front-page, article:

• contained four outright errors of fact, each of which made the players look “more guilty”;

• purported to have examined the transcript of the photo ID lineup, but didn’t mention that Mangum had positively identified at least two people who weren’t in Durham the night of the party, or any of the other inconsistencies in the session, mentioning instead only that she correctly identified the lacrosse player who paid her, the only correct ID in the entire session;

• did not mention the fact that Nifong was facing a challenger in the Nov. election, and desperately needed an overwhelming black vote to prevail;

• did not mention that Jim Coleman (a figure with impeccable credentials on issues of race) had publicly branded Nifong unethical;

• did not mention how the police photos didn’t show bruising of Mangum, despite leading the article with Sgt. Mark Gottlieb’s “notes,” which purported to recall bruising at the photo session;

• and, perhaps most important, saw its central thesis—“While there are big weaknesses in Mr. Nifong’s case, there is also a body of evidence to support his decision to take the matter to a jury”—deemed false in a subsequent public announcement from the state’s Attorney General.

The fact that, in the Entman/Gross coding system, this article is classified as stressing more the players’ likely innocence suggests that the coding system doesn’t help us much in analyzing the media’s response to the case.

This coding system, alas, leads Gross and Entman to offer the following conclusion: “Interestingly, The New York Times, which received substantial criticism for its purportedly pro-Nifong and anti-accused coverage, actually more disproportionately favored ‘not guilty’ inferences than the other sources . . . Certainly these figures do not portray a media homogeneously attacking the lacrosse players.”

Even Times executive editor Bill Keller and Times sports editor Tom Jolly aren’t making such claims; both have apologized for the paper’s downplaying the evidence that supported innocence. In this respect, Gross and Entman have positioned themselves as more royalist than the King, minimizing errors in coverage that the paper’s editors no longer defend.

----------

In the interests of full disclosure, I’m a historian of Congress, a topic dominated by political scientists, most of whom employ the kind of quantitative analysis used in this article by Entman and Gross. In the course of writing three books on Congress and foreign policy (two published by Harvard University Press, one by Cambridge University Press), I have found that quantitative poli-sci literature on Congress obscures as much as it illuminates. So in general terms, I’d be unfavorably inclined toward this article’s research approach.

That said, I’m disinclined, for the reasons presented above, to adopt the thesis presented by Entman and Gross. To conclude with a question directly related to the article’s thesis: virtually everyone closely involved in the case—from Nifong on one side to the defense attorneys on the other—believed that the Times and the N&O offered dramatically differing portraits of what occurred in Durham. What does it say about the merits of the Entman/Gross coding system that their model assigns comparable scores to the two papers?