Quantitative social science is best when it provides rigorous evidence of counter-intuitive propositions, that is, when it can debunk commonly held perceptions about phenomena. A good example is Charles Murray’s careful analysis of social survey data in Coming Apart, showing that in fact low-income whites are abandoning religion and marriage much more than high-income whites—the reverse of what is typically perceived. But much of the time, social science is proving the trivial or the obvious, and offering little in the way of genuine advancement of our general understanding of how the world works.
My favorite recent example is a nine-year Finnish study of more than 55,000 people that found—are your sitting down for this?—that people who live close to bars drink more heavily! “Conversely,” the news story helpfully reports, “the farther away a participant was from the bar, the less likely they would be heavy drinkers.” And the news writer reminds us, just in case we forgot, “The Centers for Disease Control and Prevention warn that heavy drinking can lead to liver disease and cancer.” And, “The study appears in the journal Addiction.” Good to know that last part; I was beginning to suspect someone had slipped one from The Onion past me.
Meanwhile, in the “Important News You Can Use Department,” the good people at LiveScience.com reported yesterday that based on an analysis of 10,000 porn stars, the average adult film actress is a brunette with a B-cup named “Nikki.” There may be a problem with data quality, though, as LiveScience reports: “[M]any actresses are reluctant to help researchers, because they’re worried that the studies will be used against them by anti-pornography activists.” Even worse: the research was not peer reviewed! “Millward’s data is not published in a research journal or peer-reviewed by experts, so scientists like Griffith take it with a grain of salt.” Now, it would seem that finding the qualified peer group to review porn data shouldn’t be that hard, yet the story adds the dismaying news that “academic research on the porn star population is lacking.” This is just the sort of important research that is going to be curtailed if the sequester goes through.
But are these examples really more trivial than what often appears in the “mainstream” journals like the American Political Science Review? The latest issue actually has several articles that appear worth reading and on relevant topics of the moment, but I’m still guffawing over an article from the last issue of 2012 entitled “Tying Your Enemy’s Hands in Close Races: The Politics of Federal Transfers in Brazil.” Here’s the abstract:
This article uses a regression discontinuity design in close electoral races to disclose purely political reasons in the allocation of intergovernmental transfers in a federal state. We identify the effect of political alignment on federal transfers to municipal governments in Brazil, and find that—in preelection years—municipalities in which the mayor is affiliated with the coalition (and especially with the political party) of the Brazilian president receive approximately one-third larger discretionary transfers for infrastructures. This effect is primarily driven by the fact that the federal government penalizes municipalities run by mayors from the opposition coalition who won by a narrow margin, thereby tying their hands for the next election.
The mathematics in the article are very impressive, but at the end of the article the authors have merely proved what every citizen of Chicago knows intuitively: practical politics operates by rewarding your friends and hurting your enemies, with the allocation of public money being the best tool. Do we really need social science to prove this to our satisfaction? And people wonder why social science isn’t moving very fast to answer some of our biggest social problems. As I sometimes say to students about this, no one is ever going to solve the Israeli-Palestinian conflict with a regression equation.