John Harvard's Journal
Putting the Science in Social Science
Can political scientists devise formulas to predict the outbreak of war? Although the notion seems far-fetched, James Alt wants you to say, "Sure!"
It has been just over three years since the Thomson professor of government became the first director of Harvard's Center for Basic Research in the Social Sciences (CBRSS--see www.cbrss.harvard.edu), a group of scholars focusing on ambitious and important--but distinctly unorthodox--work on the quantitative frontiers of their fields.
For instance, the center's International Conflict Initiative aims to use statistical methods to analyze both civil and international conflicts--and then use that analysis to predict future outbreaks of war. The Encyclopedic Enquiry project seeks to create an Internet research resource that will allow researchers and other curious folk to ask questions in natural language and retrieve answers from complex data sets without having to perform the statistical analyses themselves. A student could enter the question, "What is the correlation between age and blood pressure among citizens of Cambridge?" and get an answer instantly, computed from the data sets of a lengthy survey. In another initiative, the center has used an $800,000 federal grant to hire computer programmers to develop a universal and dependable standard for the citation of data sets: the approximate equivalent of the Modern Language Association format for citing publications and a seemingly impossible undertaking, since many of these data sets are on the Internet and can be changed or moved without notice.
By developing better methods of research and analysis, the center aims to promote advances in basic research methods, rather than applications. "Our motive is the same as anyone else's in the social sciences. We would like to improve the human condition. We would like to alleviate human suffering. We believe the way to do that is to develop and disseminate basic research tools," says Alt, who helps teach a departmental course on empirical methods. The International Conflict Initiative, for example, fits into one of the center's long-term, larger projects--the Program on Human Security, which seeks to develop a conception of human security that goes beyond the simple question of whether a country's borders are properly defended. The project, says Alt, was motivated by center affiliates' irritation when the magnitude of the Kosovo refugee crisis startled politicians and pundits. "It bothers us that people say they're surprised by things that we don't think should be a surprise," he says.
But this assumes that such a complex matter as the decision to go to war, with its dependence on economic conditions, politics, and the seemingly unpredictable variable of individual will, can be predicted statistically. It's an idea that at first seems to lead to the conclusion that human behavior can be predicted--which runs contrary to the idea of free will. This misunderstanding accounts for the disbelief that has greeted the International Conflict Initiative. Professor of government Gary King, director of the Harvard-MIT Data Center, with which CBRSS is affiliated, explains that free will is the very reason why such predictions can't attain a greater level of accuracy--but that doesn't mean the predictions aren't helpful.
King, as the head researcher for the conflict initiative, worked with a data set created during the Clinton administration in a project headed by then vice president Al Gore. Using factors like a country's form of government ("We've learned that democracies tend to fight each other a lot less often than countries that are not democracies," says King) and simple proximity ("Countries that are next to each other tend to fight each other a lot more often than countries that are not next to each other," he points out), the data aimed to predict the likelihood of state failure: that is, the loss of the head of state's power to enforce the rule of law in a given country.
The government team performed the original analysis, while King's team later replicated the government's findings independently, correcting mistakes and using more complicated statistical methods of analysis. The numbers they came up with indicated a given country's likelihood of having state failure within the next year--and some countries rated a 30 or 40 percent likelihood.
And what to do with these predictions? The answer is simple: use them to make sure they don't become reality and take action to prevent state failure where it's most likely. "The point of making these predictions is for them to be wrong," says King. The predictions can also be used to mobilize resources in advance of the public-health problems that result when state failure occurs. Most current public-health initiatives protect only from existing hazards and compensate for war's ravages only after the fact, ignoring risk assessment and prevention--two major modes of action. "They're not making use of the fact that political scientists have spent decades figuring out the causes of war," says King. The CBRSS project aims to change that.
This interdisciplinary approach is precisely what distinguishes CBRSS. "We are talking about crossing boundaries in a big way," says Alt. The idea is that at a decentralized place such as Harvard, a center not tied to any particular discipline is needed to create dialogue among the disciplines. "You could spend your whole life in one department and never have any interaction with anyone else, and that would be a real shame, given the diverse array of talents here," says King. Indeed, this interaction--between students and faculty, between students and other students, from inside Harvard and out--is what distinguishes CBRSS, say those who've benefited from its programs. "When I have questions, I can talk to the other grad students here," says Shigeo Harano, a sixth-year Ph.D. student in the political economy program who recently received a dissertation completion grant from CBRSS. "I know the people in psychology who do quantitative stuff, I know the people in sociology who do quantitative stuff, so there's also that peer connection that's really important."
Kevin Quinn, who spent two postdoctoral years working with King on a project to formulate better models for predicting mortality, now holds an assistant professorship at the University of Washington, dividing his time between the political science and statistics departments. He's part of an innovative program there that aims to foster research among social scientists and statisticians--not too different from what's going on at CBRSS. But the University of Washington program has more of a statistical focus, so CBRSS remains the only center in the world chartered on the idea of basic research, according to Alt. Besides the projects outlined above--for which it receives funding from Harvard and the government--the center funds in-residence fellowships for professors, graduate students, and undergraduates, and offers seminars, conferences, and workshops for Harvard students and affiliates.
Lodged in a quaint yellow house on Kirkland Street that dates from the eighteenth century, when the street was known as Professors' Row, the center has space for roughly 20 people at any given time, including the resident fellows and an administrative staff of six. Its acronym, which was King's idea, is pronounced "sea breeze."
Alt and King, among others, came up with the idea for the center, but Alt says that dean of the Faculty of Arts and Sciences Jeremy R. Knowles "was the person who made it happen." CBRSS, says Knowles, "has given us a much livelier presence at the more analytical end of political science. That means Harvard is a more attractive place for faculty and graduate students, and provides a livelier experience for undergraduates."
Of course, one of the risks of a center that tries new and even unlikely ideas is a certain degree of failure. But Alt and King welcome that as a sign of success. "How are we going to know unless we try?" asks Alt. "If you never fail, then you're not nearly close enough to the frontier."