Research on civil war has become a growth industry in political science in the last five years. One of the most active areas has used panel data looking at 160-or-so countries' experiences over time since 1945 or 1960. Our paper looks at a broader range of possible determinants of civil war onset than most previous studies, and poses an explanation for the observed patterns.

A number of different statistical techniques are used in the paper, including some nonparametrics that are not yet common in political science. But the main interest I think is in the presentation of a lot of evidence against the view that ethnic diversity, or any particular ethnic or religious demography, has a direct positive impact on the odds that a country will have a civil war.

The remarkable spread of persistent civil war in the poorest countries of the world has had little to do with ethnic demographics, religious hatreds, economic inequality, dependence on primary commodity exports, or even absence of political rights. Instead, it mainly reflects the success of a set of military techniques—rural guerrilla warfare—in poor countries whose state administrative and police capabilities are weak and underdeveloped. Peacekeeping operations to civil war-torn countries with low state capabilities will fail unless "peacekeeping" becomes successful state building.

Political violence is an enduring research topic of mine, and, for my colleague David Laitin, ethnic identity and conflict. We are currently working on a book project on the subject of civil war since 1945. This article summarizes some of our main empirical results and our theoretical interpretation of them.