Various perspectives have been proposed to explain regime change under authoritarianism, yet the role of autocrats— ‘the elephant in the room’— remains largely unexamined in large- N studies. This study addresses this gap by employing the Randomization Inference method that reshuffles historical leadership sequences as counterfactuals. Analyzing data from 150 countries between 1946 and 2020, we find strong evidence that autocrats contribute substantially to variance in regime change and that their influence extends in both authoritarian and democratic directions. We also find that democratic leaders play a significant role in democratic backsliding but not in democratic deepening. Our findings contribute to the agency–structure debate on regime change by providing a systematic large- N validation of the agency perspective, thus complementing prior studies that have relied primarily on fragmented case studies.
Liu et al. (Thu,) studied this question.