For most of the twentieth century, many people on the left and right shared an assumption: serious change required guns, and nonviolent protest was admirable but weak. In 2011 the political scientists Erica Chenoweth and Maria Stephan published Why Civil Resistance Works, a study that challenged that assumption with data. Their findings have shaped debates among activists, diplomats and scholars ever since, and they deserve both attention and scrutiny.

What they counted

Chenoweth and Stephan gathered information on 323 campaigns between 1900 and 2006. They included only campaigns seeking maximalist goals like regime change, ending a foreign occupation or achieving independence, not protests over wages or single laws. Each campaign was coded as mainly violent or mainly nonviolent, and as a success, partial success or failure. The headline result: nonviolent campaigns succeeded about 53 percent against 26 percent for violent ones. Nonviolent resistance was roughly twice as likely to achieve its full aims.

Why numbers matter

Their main explanation is a participation advantage. Joining an armed group requires physical strength, willingness to kill, secrecy and often leaving home. Joining a boycott, strike or march requires much less. Lower barriers let more kinds of people join: the old and young, women and men, workers, students, professionals and clergy. Larger and more diverse movements can use a wider range of tactics, from strikes that halt the economy to mass refusals to pay taxes or obey orders, which matters because rulers depend on the daily cooperation of millions.

Loyalty shifts and backfire

Diverse movements also produce loyalty shifts, meaning security forces refusing orders or defecting, along with officials, business leaders and others whose cooperation the regime needs. Civil resistance in this sense is not passive. It includes disruptive tactics such as sit-ins, blockades and general strikes, which impose real costs on rulers without killing anyone. Its force lies in the withdrawal of cooperation: soldiers who will not shoot, clerks who will not process orders, workers who will not work. A soldier ordered to fire on an armed rebel sees an enemy; a soldier ordered to fire on an unarmed crowd may see his neighbors, his cousin or his teacher. When rulers do use violence against peaceful protesters, it can produce backfire: repression of peaceful crowds stirring wider outrage, which brings new supporters and turns bystanders and foreign governments against the regime.

What comes after

The study also looked beyond the moment of victory. Countries where nonviolent campaigns succeeded were more likely to produce democracy afterwards, and less likely to slide into civil war, than countries where armed campaigns won. Part of the reason links back to the last level: armed victors already have an army and a habit of command, while nonviolent movements depend on persuasion, coalition building and broad participation, which are also the skills democracy needs.

The 3.5 percent observation

Chenoweth later highlighted a striking pattern in the data: no campaign that drew 3.5 percent of the population at its peak had failed. In a country of 50 million, that means about 1.75 million people actively taking part. The figure spread widely as the "3.5 percent rule." Chenoweth herself has cautioned that it is a historical regularity, not a law, and she has identified exceptions since, such as the large uprising in Bahrain in 2011, which was crushed with help from neighboring states.

A troubling trend

More recent data complicate the story. In the 2010s nonviolent campaigns became more common than ever, yet success rates fell even as campaigns grew more common. Researchers suggest several reasons. Governments have learned to answer with smarter repression, surveillance, co-opted media and paid counterprotesters. Movements organized quickly online can fill a square in days but may lack the organization, discipline and strategy that slower movements built, and they may rely on rallies more than on strikes and noncooperation. The record shows that nonviolent resistance often works, not that it works automatically.

Worked example: Serbia, 2000

By 2000 Slobodan Milošević had ruled Serbia for over a decade, through wars, sanctions and economic collapse. A student movement called Otpor, meaning "Resistance," had grown since 1998, using humor, street theater and a clenched fist symbol to make fear look ridiculous. Opposition parties, long divided, united behind a single candidate, Vojislav Koštunica, for the presidential election of 24 September. When the authorities claimed that no one had won outright and a runoff was needed, the opposition called a general strike. Coal miners at the Kolubara mines, which supplied much of Serbia's electricity, walked out, and police sent to break the strike backed off.

On 5 October huge crowds converged on Belgrade from towns across the country. The day was not entirely peaceful: protesters stormed the federal parliament, part of which was set on fire, and a bulldozer became a symbol of the day. But police and army units largely refused to fire, and Milošević conceded the next day. The case shows the participation advantage, the importance of unity, the power of strikes and the decisive role of loyalty shifts. It also shows that "nonviolent" often means "mainly unarmed," a point critics take up in the next level.