Does a movement win once 3.5% of a country joins it?

Not automatically. The finding is that, in one well-known dataset of campaigns against governments, nearly every campaign that reached 3.5% of the population succeeded, and every one of them was nonviolent. Political scientist Erica Chenoweth and Maria Stephan studied the major violent and nonviolent campaigns to change a regime between 1900 and 2006. Comparing 323 of them, they found that only 26% of violent revolts succeeded, against 53% of nonviolent campaigns. Chenoweth then found that nearly every campaign with active participation from at least 3.5% of the population had succeeded, and that every campaign that reached that level was nonviolent. It became known as the 3.5% rule.

Why would nonviolence do better? Chenoweth's explanation is about numbers: nonviolent campaigns attract larger numbers, partly because they have fewer requirements for physical ability or weapons. In the data they averaged four times as many participants, and 20 of the 25 largest movements studied were nonviolent. The 3.5% figure came from Chenoweth revisiting an older proposal by Mark Lichbach, who had suggested that 5% could topple a government and that no opposition movement could surpass that because of the free-rider problem.

Chenoweth warns that it is a rule of thumb, not a hard law. It is descriptive, not prescriptive, and omits factors such as momentum, organisation and strategic leadership. Critics add that the dataset was too narrow, that it left out failed nonviolent uprisings and successful violent revolutions, and that Brunei and Bahrain passed the threshold without regime change. Chenoweth answers that those were small monarchies backed by powerful allies.