In the first few chapters this book contained all the typical signs of political ”science”, with emphasis on the quotation marks: misuse of the words ”explain” and ”causal” when the correct expression would be ”correlates with”, use of GDP as a putatively exact measure of something it wasn’t intended to measure, and of course, use of the word ”empirical” for studies where the authors read or browsed through collections of newspapers and books and drew their conclusions.

However, meaningless correlation coefficients between ”coded” variables are fortunately not a central part of this book, even though it contains a few tablefuls of them. The authors actually have a fairly interesting theory about democratic and dictatorial developments in Latin America in the twentieth century. The theory puts a lot of weight on the regime preferences of politically important actors and on international influences. Especially the authors' qualitative discussion of Argentinean politics in chapter 5 seems to support their argument. The authors also provide a nice critique of alternative explanations for democratic / dictatorial regime changes, such as modernization theory, class theory, and theories of mass political culture.

I sympathize with the author’s explanatory dilemma. They have a good historical argument on Latin American democratization and they seek to defend it through statistical evidence. But statistics turns into complete pseudo-science when you try to assign numerical values to abstractions, such as regime preferences of various Latin American political actors who were on the scene many decades ago. Due to the great unreliability of statistical "coding", the general parts of the argument are really not interesting. However, the same argument is made much better through citation and synthesis of written sources in the country-specific case studies of Argentina and El Salvador, and these sections were certainly worth reading.