Social science: the virtues of fact collecting

Donald Horowitz has written two big and deservedly famous books about ethnic conflict: Ethnic Groups in Conflict and The Deadly Ethnic Riot. Each of them collects a mass of facts from around the world, and organizes them loosely into themes, like intergroup comparison and party politics.

James Fearon wrote a tough critique of the latter book. He praised it as a source of facts, but criticized its hypotheses as unclearly stated and not properly tested. His arguments are typical of the modern political scientist’s cast of mind: we are supposed to develop clear, precise hypotheses and test them with statistical rigour. In short, to be scientists.

And yet. Some years back an article in Political Science and Politics made a big impression on me. It was from a US army officer serving in Iraq. History, he said, seemed more practically useful to him than political science.

I have a similar intuition. If I were asked to recommend a book on ethnic conflict for somebody practically involved in peacebuilding or peacekeeping, I would choose Horowitz, in preference to other work that is more careful about forming and testing hypotheses. (In preference, for example, to either Varshney’s or Wilkinson’s excellent work on India.)

The point is not that fact collection is an indispensable precursor to hypothesis formation. We can agree on that – Fearon and Varshney and Wilkinson have all put a lot of effort into collecting facts, and have created extremely useful resources for other researchers by doing so. The point is that, if my intuition is right, sometimes fact collection is a useful kind of knowledge in itself – useful for the practitioner, without further intervention from other scientists. How can that be? What kind of knowledge is transmitted by a big lump of semi-organized facts, that isn’t conveyed by those same facts pre-digested?

The answer may come from two facts about social science. First, it’s very hard. Biology is more complex than physics; biological laws are more probabilistic than physical ones and biology will never be reducible to four fundamental forces. Social sciences add another level of complexity: we deal with a very complex organism – the human – which forms even more varied and complex social patterns. Inference from one context to another is notoriously difficult in social science; two schools, towns or countries may work in very different ways.

The second fact about social science is that humans are good at it. Every human, even your worst undergraduate student, is a well-practiced, instinctive “folk” psychologist, organization theorist, game theorist and discourse analyst. (And rabbits are probably pretty good instinctive bunny sexologists, for the same reasons: they have to be.) We are good at spotting the patterns in unfamiliar situations.

What the undigested facts give the practitioner is context. When you are dealing with Sunnis and Shias in Baghdad, some unforeseen slice, taken through a bunch of anecdotes about ethnic conflict globally, may combine with your own local knowledge into a pattern that helps predict the future. The same anecdotes, prepackaged as a database and summarized in a hypothesis which fits them very well, may be irrelevant to your situation.

So, one thing differentiating us social scientists from the physical sciences is the kind of practical advice we can hope to give. We are not very expert, and our audience is quite expert. We therefore want to be humble and not to package our stories too neatly: practitioners will have their own stories to add to them.