Published on November 1, 2014
1 *Field-work was funded by the Social Science Research Council's International Predissertation Fellowship Program with funds provided by the Ford Foundation and the Duke-UNC Program in Latin American Studies with funds provided by the Tinker Foundation. Introduction This paper re-analyzes democratization in Honduras in the 1950-1963 period. This is part of a larger project which seeks to explicitely explore the relationship between militarization and democracy in Latin America during the Cold War. The data are derived in large measure from archival research and a reading of the Honduran newspapers of the time-period. Two principal findings resulted. Perhaps the most interesting and surprising findi ng is that the militarization issue engendered a long-running and serious national debate in Honduras. Hondurans were deeply divided over whether they should build a strong military institution or follow the Costa Rican model. Many editorials in the 1950s eerily predicted the results of the creation of a strong military caste in the country. The second key point of the paper is that democracy in Honduras was not predestined to fail due to sociological or econonomic conditions. Indeed, many factors in the country were favorable for democracy. I conclude that the sudden and dramatic emergence of a strong and autonomous military institution was a principal factor in the breakdown of democracy in 1963. Before I present the case, it is prudent to first introduce the logic of counterfactuals. The use of counterfactuals and the criticism of that strategy have been around for generations (see Tetlock and Belkin 1996a for a full discussion). Yet, many social scientists have concluded--and I think correctly--that counterfactuals are an inevitable element of any comparative research (Przeworski in Kohli 1995, 20). Even statistical research that purports to show that the dependent variable varies with changes in the causal variable is often interpreted by social scientists with counterfactual reasoning: If case A would have reduced causal variable X by 50%, then case A would have achieved a score of Z in the dependent variable. Tetlock and Belkin (1996b, 3) argue that we "can avoid counterfactuals only if we eschew all causal inference and limit ourselves to strictly noncausal narratives of what actually happened." The Honduran case is implicitly based on a counterfactual: Without an autonomous military in 1963, a golpe de estado would not have occurred and democracy would have endured at least another day. The issue in causal comparative research is not whether one uses counterfactuals, but whether it is explicit or implicit in the argument. Tetlock and Belkin (1996b, 18) and Dawes (1996) provide advice on how to produce better counterfactual inferences. Dawes in particular insists that causally convincing counterfactuals must be derived or justified by statistical analyses; "I argue that good counterfactual inferences--again, productive, reasonable, helpful ones--should be point predictions arising from statistical expectations" (1996, 301). That is, stand-alone counterfactuals should not produce much confidence.