Major revolution datasets agree on most things (except one)

What is it about?

When a massive protest movement tries to topple a government, researchers often call it different things—some say "revolution," others prefer "maximalist campaign" or "popular uprising." This fragmented vocabulary has produced several major global databases, each built with its own rules, definitions, and timeframes. But are these databases actually tracking the same historical events, or are they measuring completely different phenomena? To answer this, the authors systematically compared six leading datasets covering hundreds of uprisings across the twentieth and twenty-first centuries. Their findings offer good news for students of political change: despite surface-level differences in labels and coding, most of these databases paint a remarkably consistent picture. They agree on where revolutions tend to break out, when they cluster historically, and what underlying conditions—such as economic development and population size—make them more likely. Crucially, they all confirm that peaceful, nonviolent uprisings are significantly more effective at producing lasting democratic improvements than armed insurrections. There is, however, one notable outlier: an older dataset that records over 2,000 "revolutions" in a single century—an implausibly high number. The authors show that this source mistakenly lumps together genuine mass uprisings with routine military coups and coup attempts, making it unreliable for studying true revolutionary dynamics. For scholars and policymakers, this study provides a practical roadmap: while different datasets serve different specific purposes, most can be trusted to deliver the same broad insights, and choosing the right one depends on whether you need detailed information about protest tactics, participant demographics, or long-term political outcomes. By clearing up confusion about how revolutions are counted, this work helps unify a fragmented field and offers clearer guidance for future research on political instability and democratic resilience.

Featured Image

Why is it important?

When researchers study mass protests that try to overthrow governments, they use many different names—"revolutions," "maximalist campaigns," or "popular uprisings." This fragmented language has produced several major global databases, each built with its own definitions, timeframes, and coding rules. But are they actually tracking the same historical events, or are they measuring completely different things? To answer this, the authors systematically compared six leading datasets covering hundreds of uprisings across the twentieth and twenty-first centuries. The good news is that, despite surface-level differences, most databases paint a remarkably consistent picture. They agree on where revolutions break out, when they cluster historically, and what conditions—like economic development and population size—make them more likely. Crucially, they all confirm that peaceful, nonviolent uprisings are significantly more effective than armed insurrections at producing lasting democratic improvements. There is, however, one notable outlier: an older dataset that records over 2,000 "revolutions" in a single century—an implausibly high number. The authors prove this source mistakenly lumps genuine mass uprisings together with routine military coups, making it unreliable for true revolutionary analysis. What makes this study unique and urgently timely is that it is the first to empirically test whether these datasets actually converge using hard statistical tools—overlap calculations, regression models, and matching analyses—rather than just comparing definitions on paper. Published in 2026, it arrives at a critical moment when the twenty-first century has seen a dramatic global shift from violent revolutions toward unarmed civic uprisings (from the Arab Spring to recent protests in Eurasia), creating an urgent need for researchers to know which tools accurately capture this new reality. The study's newly developed CSRA dataset, covering events up to 2024, fills a major gap left by older sources. The difference this work makes is practical and profound. It saves scholars years of wasted effort by proving that—for most broad research questions—choosing one reputable dataset over another won't change the core answer. At the same time, it provides a clear, practical roadmap: need to study protest tactics over time? Use NAVCO. Need detailed social composition of protesters? Use Beissinger's dataset. Need the most current events? Use CSRA. By cutting through methodological confusion, this study helps economists, political scientists, and sociologists speak a common empirical language, strengthens the credibility of nonviolent resistance findings, and ultimately gives policymakers more reliable evidence about what truly drives political change and democratic resilience worldwide.