The Hardest Problem in Social Science

April 26, 2010

Multiple Pages
The Hardest Problem in Social Science

British comedienne Catherine Tate did a very funny sketch with Daniel Craig, the latest James Bond actor. In the sketch she is a dimwitted, over-the-hill 36-year old who has hooked up with Craig through an internet dating service. The main joke is that Craig is besotted with her, and has moved in with her, while she is much cooler on him. (“I was hoping for someone slightly better looking, but …”) The secondary joke is that she is so clueless, she doesn’t know who Craig is. (“He says he’s an actor, but I’ve never heard of him … I think he works at Carphone warehouse …”).

After telling the “interviewer” how she and Craig first got in touch, the lexically-challenged Tate character adds: “The rest, as they say, is biology.” Realizing she has used the wrong school subject, she pauses to rethink, then corrects herself: “History.”

The joke there is of course that we were with her the first time. If an initial encounter turns into a love affair, the rest is biology, or at least an important part of it is.

So is a lot of other stuff. The Tate-Craig sketch came to mind when I was reading Wall Street Journal the other day: In 1900, David Hilbert produced a set of 23 problems that established an agenda for research in mathematics over the ensuing decades.

So he did. The event is dear to my heart: I got a book out of one of those problems. But where’s the Journal going with this?

Last Saturday, Harvard hosted a conference on “Hard Problems in Social Science,” sponsored by the Indira Foundation, that was explicitly inspired by Hilbert’s legacy. Twelve leading social scientists from a variety of fields and institutions were given 15 minutes each to present whatever hard problems they liked.

Seems like a neat idea. What’d they come up with?

<object width=“300” height=“243”; style=“float:left; MARGIN: 10px 10px 10px 10px”/><param name=“movie” value=“”>

<embed src=“” type=“application/x-shockwave-flash” allowscriptaccess=“always” allowfullscreen=“true” width=“300” height=“243”></embed></object><p>I have to confess that my attention wandered somewhat here. Twelve 15-minute presentations makes three hours. That’s a lot of lecturing to sit through when your college days are three or four decades behind you. Some of the topics are more accessible than others; some of the presenters are more watchable than others. What follows is therefore inexact and impressionistic. You can view the whole thing on your own time.

Nicholas Christakis of Harvard: How and why does the social become biological? A really good presentation on the culture ↔ biology feedback loop. The star attraction here, as always, is lactose tolerance: a change in culture (from hunting to cattle herding) causing actual genetic change. In the other direction, genes shape culture. How many friends do you have? How many does he have? And her? And him? And her? … the number varies over a population. Well, 46 percent of that variation is genetic.

Ann Swidler, Berkeley: How do societies create institutions? Lots of ancillary problems: Who has influence and why? How do societies cope with free riders and defectors? Nicely done, but not as much fun as the first.

Nassim Taleb, NYU: Yep, it’s the black swan guy, riding that same hobby horse (hobby swan?) Can we find a robust way to cope with events that have tiny probabilities but huge consequences? You need to be able to thrill to statements like: “One day in 40 years represents 80 percent of the kurtosis.” Good fun if you know the jargon.

Peter Bearman of Columbia, standing in for an indisposed Robert Sampson of Harvard: Something about the continuity of civil violence.

Nick Bostrom of Oxford: What is the biggest falsehood promulgated in social science today? I’m not sure that was actually his main problem. He shot off into some sub-problems, e.g. How can humanity increase its collective wisdom? (“There are 30 billion IQ points in the U.S,, but they don’t aggregate very well.”) Hard to follow; but then, Bostrom’s a philosopher. They’re supposed to be hard to follow.

Gary King of Harvard: something about post-treatment bias. Very jargony presentation on experimental design in social sciences & the methodological problems with experimental variables. Worthy, I’m sure, but zzzzzz for non-specialists.

Emily Oster of the University of Chicago: How do we get people to change their health behaviors? I dunno. Smack ‘em upside the head? More worthy-but-dull, skipped the second half.

Claudia Goldin of Harvard: Why do sex differences in economic outcomes exist? (She said “gender,” of course, not “sex.”) Not such a duh! as it sounds, and quite rigorously done. “More women than men want shorter, more flexible hours …” etc.

Susan Carey of Harvard: How do new concepts arise when there are no similar concepts for them to build on? I think that was her question. Dense, challenging canter through some ideas in cognitive science, with side excursions in philosophy of science. I wish Steve Pinker had been there to hold my hand.

James Fowler, U. Cal. San Diego: What causes clustering in social networks? Some good rigorous experimental sociology, up there with Nicholas Christakis. Good models, neat graphics. You can squeeze some predictive virtue out of four degrees of social relatedness: e.g. if your friend’s friend’s friend’s friend is a heavy drinker, I can make nontrivial probabilistic inferences about your boozing. Good stuff on the difficulty of untangling influence (we’re similar because we hang out together) from homophily (we hang out together because we’re similar).

Roland Fryer of Harvard: Why does the black-white skill gap persist? Gotta improve those schools! KIPP! HCZ! Charter schools! Good vigorous presentation, but same old questions left hanging in the air. E.g.: If these intensive interventions raise the school achievement of poor black kids, why wouldn’t they raise the achievement of poor white kids, too? Or for that matter, of middle-class and rich white kids? If you have a method of raising school achievement, that is a social good. On what grounds, moral or constitutional, can you withhold that good from children who are nonpoor and nonblack? And if you don’t, won’t you be left with the same gaps as before, just shifted up the scale? Then the really hard questions, like: How do you get unionized teachers to work KIPP hours?

Richard Zeckhauser of Harvard: the question was either What should we strive to maximize—income, happiness, or what? or else How do we aggregate information from different parties to make decisions, e.g. what’s the optimal election system? Sorry, I was nodding off by this point.

As a thumbnail (can you have a three-hour thumbnail?) sketch of the current state of the social sciences, this symposium was informative. Most informative of all, though, was to see how little biology was there, once Christakis was through. The Standard Social Science Model, a.k.a. the “blank slate,” is still going strong.

Take Prof. Goldin’s topic: Why do sex differences in economic outcomes exist? I liked her presentation, and I’m sure she’s got a lot of the answer covered. Might not some of it, though, be a matter of women being less aggressive, less ambitious than men, for merely biochemical—hormonal—reasons? And why are economic outcomes the focus? After all, they’re not life and death, like this “gender” difference.

Come on, social scientists. There are some good minds there, doing good quantitative work. When you’ve measured all the environmental variables, though, and crunched the numbers—ranked, sorted, histogrammed, and correlated them—and there’s still an explanatory gap, what’s left? The rest is biology.

Meanwhile, here’s my hard problem in social science: Can a pure meritocracy be stable?

Daily updates with TM’s latest


The opinions of our commenters do not necessarily represent the opinions of Taki's Magazine or its contributors.