There have been three waves of scholarship on military coups d’état (or simply “coups”)—the unconstitutional replacement of chief executives by military officers—since the 1960s. The first used case studies to explore why the military overthrows governments. One of its central findings was that military uprisings were an integral part of political succession in many countries. A second wave produced the “aggregate studies” that were the first to deploy cross-national databases to identify the measurable features that distinguished more from less coup-prone political systems. These studies revealed, among other things, that coups proliferated in places with a history of instability. The third and current wave of scholarship takes advantage of the development of statistical software for limited dependent variables—then unavailable, now commonplace—to recast the quantitative research on coups.
Two core findings have survived disconfirmation since the start of the third wave. First, higher income countries have fewer coups, though the effects are small (and become even weaker when models only contain developing countries). Second, “political legacy effects” mean that the probability of a coup declines with time since the last military uprising. Much of the latest wave of research pinpoints factors—like coup proofing, less inequality, or the end of the Cold War—that reduce the probability of a coup. The development of ever more sophisticated statistical techniques to divine the causes of instability, nevertheless, relies on off-the-shelf data sets and coup catalogs whose validity—properly understood as accuracy—is questionable. Only a greater attention to accuracy and complementary methods promise to produce a comprehensive account of why the military topples governments in some, but not in other, places.