Friday, July 30, 2010

Malcolm Gladwell and the Popularization of Social Science

be the trouble not the balance
Tony Tost


Just before heading back from vacation, I read Brayden King's post on Malcolm Gladwell over at OrgTheory.net, and soon got into what some (including Gladwell himself, it seems) experienced as a heated (or at least "tiresome" and "silly") argument about the value of popular science books.

As one commenter suspected, I hadn't read much of Gladwell's stuff before. Like many others, however, I was aware of his work (I have a subscription to the New Yorker) and I had in fact read Steven Pinker's review in the New York Times, probably via Language Log's post on what Pinker called the Igon Value Problem. (It's the sort of scholarship "fail" that I like to use as an example in my writing instruction.) And since Gladwell had honoured us with his participation in the discussion*, I did the sensible thing and bought Blink in the airport on the way home. (A minor example of how your presence in the blogosphere can improve your book sales.)

Reading it confirmed my suspicions. So, when I returned to the conversation (after sleeping off the jet lag) I was able to apply my general criticisms of the popular science genre to Blink in particular. I'll get to that in a moment.

One thing that struck me in the course of the discussion was the mutual appreciation that was clearly expressed between Brayden (as a social scientist) and Gladwell (as a science journalist). It was perhaps most clearly expressed in Bob's somewhat over-the-top fawning over the "HUGE debt" (his emphasis) that sociology owes to Gladwell "for popularizing our research". The compliment was returned by Gladwell:

I’m grateful for the interest in my work. And I appreciate the thoughtful criticism, much of which I will take to heart. Just know that my mistakes, such as they are, tend to be made out of enthusiasm for what all of you do.

But what does it mean to say that social scientists "owe" Malcolm Gladwell a debt of gratitude for his promotion of what they do? I think the OrgTheory thread gives us a good indication.

During the discussion, Brayden pointed out that Gladwell's failings are not uncommon in academic writing. That may be true, but at least in academic writing they constitute a valid point of criticism. What Gladwell's supporters were arguing is that Gladwell's mistakes do not significantly diminish his contribution, probably because his aim is mainly to generate excitement. A falsehood, after all, is just as likely (and in the case of social science much more likely) to do this effectively.

Consider the example that I was ultimately accused of "harping" on about. Reading Blink, we meet the truly startling idea that there are psychologists who have a completely objective way of knowing with 90% accuracy whether a couple will get divorced based on nothing more than a 15-minute video of their interaction (page 21-2). When I read that, I was immediately skeptical, and, sure enough, the Wikipedia article on the inventor of the technique (the psychologist John Gottman) informs us of a chapter by Laurie Abraham, which has been excerpted in Slate.

It quite convincingly demolishes Gottman’s work by showing that he hasn't "proven" (Gladwell's word, on page 21 of Blink) his method at all. As it turns out, Gottman has only identified a pattern in his 57 videos that 90% of the divorced couples have in common. He identified that pattern after he knew they had gotten divorced. And he did not test the predictive power of this pattern on an independent sample after he had discovered it.

This was originally pointed out back in 2001, in a paper in the Journal of Marriage and Family, which it should be noted is the same journal that published Gottman's original results. Richard Heyman and Amy Smith Slep note that this lack of cross validation makes it "imprudent to inform the public and clinicians that researchers can accurately predict who will divorce." Four years later, however, Gladwell tells the public only about Gottman’s work.

As with Ekman's "lie detection" techniques, which Jeff Wise quotes a Yale psychologists describing as "hokum" in his Psychology Today post,*** Gladwell here relies on the claims of a particular scientist, while ignoring the claims of his critics. It is this marginalization of criticism that, to my mind, defines the genre and the tone of conversations that are spurred by it. Anything that might damp down the excitement, by questioning the simple truth of the claims being made, is frowned upon. I imagine this convention is strongly in force when Gladwell holds an invited talk (at TED for example). His job is to speak, to get people excited. The audience's job is to listen, and, yes, to get excited. Often there will be no room for serious questions; and if any come up, they will be booed down or the MC will deftly move on to another question. You know the drill.

On a personal note, I take Gladwell's polite "bowing out" (when I suggested he was "making a mess of the popular imagination", immediately after I had brought up the Gottman stuff, which he then did not find it necessary to address) as part of this drill. It always irks me (yes, it happens now and then) because it is essentially an oblique way of saying "Who let this asshole into the discussion?" Gladwell is entitled to his emotional responses to what is said about his work, but I think we can expect major public figures (as we should expect all academics, whether major or minor) to get beyond the (in this case very mild) rhetorical flourishes of their critics and address the issues, rather than use the occasion to feel insulted and leave the room.

In any case, it is because I agree with Gladwell (and Brayden) that "you can’t really expect" more of "a work of popular non-fiction" that I disapprove of the genre as a whole. (In truth, I would like to demand and/or expect more of popular science writers, but I have learned not to. It is in the nature of the thing that a popular writer first defines the audience and then tells it what it wants to hear. I.e., what it will find exciting.) Popular writing is not intended as an occasion for criticism.

Indeed, not even a blog post explicitly about the "backlash" was a suitable occasion for criticism. As a number of commenters tried to tell me, the appropriate thing to do in this area of discourse is to, first, acknowledge the great service Gladwell is doing to social science, second, admit that, like anyone else (being human) he makes mistakes, third, (if you must) point some of those mistakes out but only if you also, fourth, frame that criticism in the context of a specific things he "gets right".

That is, although it can be easily shown that Gladwell misleads his readers on the "blinking" abilities of at least two psychologists (I could write an exciting book about ESP, UFOs, and cancer cures to that standard in a flash), thus claiming that science has "proven" that snap judgments are often very valid (which is surely a dangerous idea if false)**, we are supposed to be 'fair and balanced' in our criticism of his work, always saying something nice about him before we offer some polite points of criticism that could, as one commenter put it, "tell Malcolm Gladwell how to be Malcolm Gladwell only better". My suggestion in this vein, I guess, is that he might try his hand at writing pulp fiction, and leave what he calls "middle-brow non-fiction" to people who care about how minds and societies really work.

Brayden suggests we blame not Gladwell but his readers. If only they would read the social science literature for themselves, then people like Gladwell wouldn't have to bamboozle them into thinking about these issues with "misleading anecdotes" and "premature conclusions" (Brayden's words, which I have simplified in my own mind as "anecdotes and prejudices", i.e., the stuff of folk sociology or everyday moralizing). I prefer to blame neither the reader nor the writer but, rather, the genre, i.e., the conspiracy of writers and readers that sets a particular critical standard, that defines a set of expectations. In this case, it is a standard that pretty much suspends disbelief altogether.

(Update: While finding the links for this post I also found this critique of Blink in the CHE, which is worth reading.)

[Read more on this topic here.]
_________
*Teppo and Ezra raised suspicions about whether the commenter was the real Malcolm Gladwell. He certainly claimed to be, and he didn't seem to be doing it either as a joke or to embarrass "himself". OrgTheory requires that commenters leave an email address (which is not published, of course). So I assumed early on that the good bloggers at OrgTheory were satisfied that this was not an imposter.

**Consider the hope he expresses in his introduction. What if Blink is right aboutpersuades us of the importance of "thinking without thinking", he asks? "I think that would change the way wars are fought, the kinds of products we see on the shelves, the kinds of movies that get made, the way police officers are trained, the way couples are counseled, the way job interviews are conducted, and so on." That is: it would change the way power is exercised in society.

[***(added August 1, 2010) Wise's most unequivocal statement is probably this one: "it has been well established by peer-reviewed studies that, Ekman's claims notwithstanding, no person can reliably tell whether or not another human being is lying simply by looking at them." There's also a good article on Nature's website that details the criticism of Ekman, and especially the SPOT system (which is a good example of the danger implicit in these ideas).]

No comments: