The question of what makes countries become democracies remains a key issue in academia as well as in the global development community. The billions of dollars invested every year in democracy-related support are mostly directed to strengthen factors endogenous to democracy itself. However, existing scholarship provides poor guidance on the efficacy of these factors in facilitating democratization. We advance an argument about one institution that is both malleable and often the object of interventions by both domestic and international actors: electoral management bodies (EMBs). We suggest a series of theoretical reasons why we should expect an increase in the autonomy and capacity of EMBs to have independent effects on democratization. We test the expectations applying a novel causal inference technique, Generalized Nonlinear Difference-in-Difference-in-Differences (NOCNOC) to a population of over 250 democratization episodes from the past 120 years.