Controlling duration dependence in conflict analyses: A replication and extension of "regime types and status quo evaluations"

Recent methodological work by Beck, Katz, and Tucker (1998) suggests that existing empirical analyses of binary time-series cross-section (BTSCS) data may be flawed. Analyses using such data fail to account for temporal dependence among observations of the dependent variable. In one such analysis, Lemke and Reed (1996) present a strong alternative to existing theoretical work on the democratic peace proposition. They are able to account for the lack of war between democracies within the theoretical framework provided by Power Transition/Power Parity theory. Since Lemke and Reed's analysis of BTSCS data may be troubled by the nonindependence problem, we replicate and extend their study using grouped duration methods to account for temporal dependence among observations. Vile find Beck, Katz and Tucker's method to be both useful and simple, and we show Lemke and Reed's findings to be robust in models controlling for the possible effects of duration dependence.