Michael Albertus and Victor Menaldo raise three important criticisms of Dictators and Democrats. All are well taken, but they also reflect contested terrain about how best to study regime change and rare political events more generally. The jury is far from out on these core theoretical and methodological issues.
The first criticism concerns our focus on Third Wave transitions, a criticism that we explicitly address in the book (pp. 7–10). Studies of democratization over a longue durée are based on the assumption that omitting the early European transitions introduces bias. Although Albertus and Menaldo are not as insistent on the need for a longer time frame, they similarly suggest that we forgo “a rich series of time-series variation within countries.”
Yet studies that “go long” face their own challenges. As the time frame becomes longer, the problem of panel heterogeneity also increases, including with respect to the outcome of democracy itself. Earlier transitions, for example, were often marked by incremental and partial franchise extensions; contemporary ones typically move toward universal enfranchisement rather quickly. Do we believe that these very different democratizations are driven by similar processes, in social, economic, and international settings that also vary widely? We are both skeptical of the heroic efforts to control for these sources of heterogeneity, and less concerned about the loss of generality. Getting our heads around transitions during the postwar period or Third Wave poses plenty of challenges as it is and has the advantage of practical relevance.
A second criticism concerns our gamble on “a typology of democratic transitions in which different types of transitions have their own players, logic, and end points.” Albertus and Menaldo make reference to some finer-grained distinctions we draw between types of elite-led transitions, but our basic conceptual distinction is in fact quite similar to theirs, and is motivated by a particular theoretical target: class-conflict models in which inequality and mass mobilization are expected to play a role. The basic players in these transitions—elites and masses—are in fact the same, although with varying organizational and institutional capabilities. And it is important to emphasize that the end points are not different either. In coding both elite- and mass-driven transitions, we rely entirely on two extant data sets’ coding of when transitions occur and assume equifinality.
We also find, as do Albertus and Menaldo, that different transition paths can affect the quality of democracy, although these differences decay over time. This conclusion is similar to their claim that democracies emerging from elite-biased transitions are less redistributive than their popularly-based counterparts, but contain the potential for more progressive reforms over time.
Finally, Albertus and Menaldo express a concern with selection on the dependent variable, namely, the well-known criticism that it “creates problems for causal inference, in that it potentially introduces selection bias.” We would note that our book takes a multimethod approach, with standard panel designs complementing our large-n qualitative analysis. But we are also convinced that our qualitative focus on successful transitions—the dependent variable—is in fact appropriate, as most other contemporary analyses of qualitative designs have also concluded (for example Gary Goertz, Multimethod Research, Causal Mechanisms and Case Studies: An Integrated Approach, 2017). Frequentist approaches favored by Albertus and Menaldo can identify the average treatment effect of a chosen causal variable, but do less well in identifying the cases that conform—and do not conform—with the specified causal mechanisms. Our mixed-method design allows us to ask not only whether the causal factor in question has a generalized effect, but also whether it operates as stipulated in the cases the model purportedly predicts.
