Vreeland, James Raymond. 2008. The Effect of Political Regime on Civil War: Unpacking Anocracy - The Web Appendix. Journal of Conflict Resolution 52 (3):401-425.
This Web Appendix presents several tests referenced in the main article. They serve to further establish that there is little evidence of a relationship between anocracy and civil war. Previous research suggests that anocracies – as defined by the middle of the Polity index of political regime – are more susceptible to civil war than either pure democracies or pure dictatorships. Yet, certain components of the Polity index are defined with explicit reference to civil war. When these components are removed from the Polity index, the original relationship disappears. I conclude that the overall Polity index should not be used in statistical analyses of civil war. Similar problems exist for Freedom House measures of political regime, which should also be avoided. Researchers looking to test the anocracy hypothesis may wish to employ the Vanhanen measure of political participation or the new SIP measure.
When it comes to civil war, controlling for linear and nonlinear effects of political regime may be unnecessary. But if control variables for political regime (democracy, dictatorship, or anocracy) are desired, researchers can use any of the following variables (together, separately, or combined in an index): X-POLITY, XCONST, XRCOMP, XROPEN, the Vanhanen measures, and SIP. The ACLP measure can also be used, although the dichotomous measure is not amenable to nonlinear hypotheses. For convenience, X-POLITY, XCONST, XRCOMP, and XROPEN are available here.