Taking Democratic Differences Seriously: A Split-Population Approach. (with Mark Nieman) pdf version.
Many have confirmed that the relationship between democracy and conflict is at least partially endogenous. Nevertheless, studies continue to group democracy with other variables as predictors of conflict using additive models. This is inappropriate both theoretically and methodologically, and we demonstrate this with an example from the alliance literature. We use a split-population logistic model to show that democracies in alliances are no more reliable than other regime types, once the state-development process is also modeled. Further, we find that the alliances formed during times of conflict are particularly unreliable ``scraps of paper'', and the general reliability of alliances is concentrated in those coordination alliances existing in already peaceful environments.
Many have confirmed that the relationship between democracy and conflict is at least partially endogenous. Nevertheless, studies continue to group democracy with other variables as predictors of conflict using additive models. This is inappropriate both theoretically and methodologically, and we demonstrate this with an example from the alliance literature. We use a split-population logistic model to show that democracies in alliances are no more reliable than other regime types, once the state-development process is also modeled. Further, we find that the alliances formed during times of conflict are particularly unreliable ``scraps of paper'', and the general reliability of alliances is concentrated in those coordination alliances existing in already peaceful environments.