from women, who, as we will see, tend to be more dovish in their voting than men. Most advocates of the Democratic Peace are willing to write off the centuries before the 20th, together with new and unstable democracies, and insist that since then no two mature, stable democracies have fought each other in a war.
Critics of the Democratic Peace theory then point out that if one draws the circle of “democracy” small enough, not that many countries are left in it, so by the laws of probability it’s not surprising that we find few wars with a democracy on each side. Other than the great powers, two countries tend to fight only if they share a border, so most of the theoretical matchups are ruled out by geography anyway. We don’t need to bring in democracy to explain why New Zealand and Uruguay have never gone to war, or Belgium and Taiwan. If one restricts the database even further by sloughing off early pieces of the time line (restricting it, as some do, to the period after World War II), then a more cynical theory accounts for the Long Peace: since the start of the Cold War, allies of the world’s dominant power, the United States, haven’t fought each other. Other manifestations of the Long Peace—such as the fact that the great powers never fought each other—were never explained by the Democratic Peace in the first place and, according to the critics, probably came from mutual deterrence, nuclear or conventional.221
A final headache for the Democratic Peace theory, at least as it applies to overall war-proneness, is that democracies often don’t behave as nicely as Kant said they should. The idea that democracies externalize their law-governed assignment of power and peaceful resolution of conflicts doesn’t sit comfortably with the many wars that Britain, France, the Netherlands, and Belgium fought to acquire and defend their colonial empires—at least thirty-three between 1838 and 1920, and a few more extending into the 1950s and even 1960s (such as France in Algeria). Equally disconcerting for Democratic Peaceniks are the American interventions during the Cold War, when the CIA helped overthrow more-or-less democratic governments in Iran (1953), Guatemala (1954), and Chile (1973) which had tilted too far leftward for its liking. The advocates reply that European imperialism, though it did not vanish instantaneously, was plummeting abroad just as democracy was rising at home, and that the American interventions were covert operations hidden from the public rather than wars conducted in full view and thus were exceptions proving the rule .222
When a debate devolves into sliding definitions, cherry-picked examples, and ad hoc excuses, it’s time to call in the statistics of deadly quarrels. Two political scientists, Bruce Russett and John Oneal, have breathed new life into the Democratic Peace theory by firming up the definitions, controlling the confounding variables, and testing a quantitative version of the theory: not that democracies never go to war (in which case every putative counterexample becomes a matter of life or death) but that they go to war less often than nondemocracies, all else being equal.223
Russett and Oneal untangled the knot with a statistical technique that separates the effects of confounded variables: multiple logistic regression. Say you discover that heavy smokers have more heart attacks, and you want to confirm that the greater risk was caused by the smoking rather than by the lack of exercise that tends to go with smoking. First you try to account for as much of the heart attack data as you can using the nuisance factor, exercise rates. After looking at a large sample of men’s health records, you might determine that on average, every additional hour of exercise per week cuts a man’s chance of having a heart attack by a certain amount. Still, the correlation is not perfect—some couch potatoes have healthy hearts; some athletes collapse in the gym. The difference between the heart attack rate one would predict, given a certain rate of exercise, and the actual heart attack rate one measures is called a residual. The entire set of residuals gives you some numbers to play with in ascertaining the effects of the variable you’re really interested in, smoking.
Now you capitalize on a second source of wiggle room. On average, heavy smokers exercise less, but some of them exercise a lot, while some nonsmokers hardly exercise at all. This provides a second set of residuals: the discrepancies between the men’s actual rate of smoking and the rate one would predict based