Anyone who has come across “Why Civil Resistance Works” by Erica Chenoweth and Maria Stephan will be familiar with the idea that size matters for social movements. Their highly cited “3.5 percent rule” says that once movements actively involve at least 3.5 percent of the population they will inevitably succeed.
The idea that this is a cast iron rule has been contested — including by Chenoweth — on the basis that it was a description of the past rather than a prediction of the future. Others have shown that the rule has been broken in at least two cases. And although it was extracted primarily from a Global South context for countries resisting regimes, it has since, controversially, been applied to the strategy documents of prominent activist groups like Extinction Rebellion and been widely quoted in the media, including by the BBC, The Guardian and The Economist.
Far less contested, however, is another of the book’s major takeaways, which is the idea that nonviolence brings a higher success rate. Looking at civil resistance focused on regime change between 1900 and 2006, they found that nonviolent campaigns were more than twice as likely to succeed as violent ones: 53 percent of nonviolent campaigns led to political change, while the same was true for only 26 percent of violent campaigns.
As a nonprofit that helps inform advocates, decision-makers and philanthropists on the best ways to accelerate positive social progress, Social Change Lab was interested in seeing how the Chenoweth/Stephan findings hold up in today’s movement landscape. We wanted to see if the evidence from their historical data translated to the present, particularly as it relates to campaigns focused on more limited, area-specific goals rather than the high level goal of regime change. We also wanted to see what other factors might help bring about protest movement wins.
Our team has spent the last six months researching these questions. We did public opinion polling; interviewed academics, social movement experts and policy makers; and reviewed the literature. What we found — and published in our full report — not only underscores the recommendations of Chenoweth, Stephan and other movement strategists, but also builds upon them, offering insights into other key factors that determine the success of movements today.
What are the most important success factors?
Movements or social movement organizations optimize for different outcomes — whether it is changing public opinion, campaigning on a particular policy, prompting public discourse or something else. So a “success factor” is variable, and it can be hard to compare them.
Nevertheless, we wanted to try to give a sense of the relative importance of different ingredients of success — so we did this by combining and weighing evidence from multiple sources, beginning with data from existing experimental studies. We weighted our estimates based on the strength of evidence behind them. For instance, if most studies on a particular topic had similar results — and our experts also agreed — that would be strong evidence. Less agreement between studies, or a lack of studies, or disagreement amongst experts would be weak evidence.
What emerged from our findings were two tiers of success factors: one that showed a clear and distinct impact on a successful outcome and another whose impact was, though still important, less decisive.
The top three factors
1. Nonviolent tactics. Even though there is historical evidence for nonviolence being the best way to go, this tactical question is still widely discussed within social movements. Many activists are tempted to adopt more violent tactics because they think it’s a more expedient way of addressing the urgent problems we’re facing. There’s also debate about what other types of tactical actions might be most effective. Our own research has suggested that having a radical flank that uses more shocking tactics (like throwing soup at paintings) can actually increase support for more moderate groups focused on the same cause.
Our research here suggests that nonviolent tactics are more likely to lead to successful outcomes relative to violent outcomes. The experts we consulted were reasonably confident that violence is a less effective approach and the literature supported their view.
Omar Wasow, at Princeton University, published research in 2020, based on studying civil rights protests in the U.S. from the 1960s. He found that states where nonviolent protests occurred went on to see increased votes for Democrats (more-or-less in line with what protesters were aiming for). Violent protests, on the other hand, led to increased votes for Republicans.
Ruud Wouters, from the University of Amsterdam, used Charles Tilley’s Worthiness, Unity, Numbers and Commitment framework to conduct empirical research. In this framework, “worthiness,” which is a measure of “the absence of disruptiveness,” is a rough equivalent of nonviolence. Wouters’s 2019 study looked at support for asylum demonstrators in a sample of Belgian citizens and found big differences in support depending on whether protesters were seen to have “high” or “low” worthiness. He suggests that low worthiness alienates the public. A further Wouters study showed a similar effect — of the greater appeal of nonviolence — on elected representatives.
The violence/nonviolence question has been widely studied by academics, and most studies reach similar conclusions, which is why we give this finding strong weight.
2. Larger numbers. Again, we found our evidence supported Chenoweth’s idea that size is really key, with bigger protests meaning a better chance of policy changes and other desired outcomes. Some interviewees suggested that although politicians invest a lot in learning about public opinion, they often don’t really understand the public — so big numbers at a protest give them a clear signal of public opinion. There might also be a virtuous circle here: As a protest gets larger, people think it’s more likely to succeed, so they feel more enthusiastic about joining it. So it gets larger, more people join and so on. Whether or not this is the explanation, we think the evidence for size being important is causal: A larger protest really will increase your likelihood of success.
In 2017, Ruud Wouters and Stefaan Walgrave looked at the attitudes of elected officials in response to protests. They found that officials were much more likely to take a position closer to the protesters when protest numbers were high. This change in their thinking also translated into behavior change and taking action, such as proposing a bill or asking a question.
Bouke Klein Teeselink and Georgios Melios also considered whether mass mobilizations bring about social change — this time by looking at the effect of the Black Lives Matter protests in 2020, following the death of George Floyd. Their research found that wherever large numbers participated in protests, the result was a greater increase in the Democrat share of the vote. In fact, for every 1 percent increase in the fraction of the population who protested, there was a raise in the Democratic vote share of 5.6 percent. Another study found that killings by police decreased by as much as 20 percent in municipalities where BLM protests occurred, and that police departments were more likely to adopt body cameras and community-policing initiatives.
In 2012, Stefaan Walgrave and Rens Vliegenthart looked at Belgian protests that had taken place between 1993 and 2000. Their analysis included more than 4 million people participating in almost 4,000 demonstrations. They too found a highly significant impact of protest size on legislation outcomes, suggesting this effect comes in part through bigger protests being associated with more media coverage.
3. Favorable sociopolitical context. There are other factors more outside the control of protesters — things like pre-existing public opinion, the response of the media, whether there are elites (like politicians or celebrities) who support the cause, as well as blind luck. This isn’t great news in terms of actionable evidence, as it can be hard to know what constitutes the right conditions and even harder to judge best timing. On top of those uncertainties, movements themselves have their own seasons and cycles, as Carlos Saavedra from the Ayni Institute has noted. There is little direct evidence on the effect of elite allies, but few would argue with a “best bet” of trying to win over influential people to your cause. Our experts agreed that winning a positive reception from elites was a really important factor — one even claiming that this factor alone explained 80 percent of the variance in outcomes.
Some researchers have tried to get a firmer empirical handle on the influence of elites, such as Marco Giugni and Florence Passy in Switzerland. Their 2007 research looked at the impact of elite allies and the effect they can have, over and above that of public opinion. They found that it was a combination of protest, supportive public opinion and the presence of political allies that led to policy wins. They also found that this combination of factors led to increased spending on environmental protection and reduced spending on nuclear energy (in line with protester demands).
Legislators adapt their policies and positions in response to public priorities — and the typical way to represent public priorities is through surveys. But protests offer another way to represent public views, and protests can also amplify public priorities. Luca Bernardi, Daniel Bischof and Ruud Wouters analyzed a data set covering nearly 40 years in four Western countries, looking at policy maker agendas, protests and public opinion. They concluded that it is very rare for protest alone to have an effect on legislators. Only when protests interact with the priorities of the public will legislators be moved to change their agendas.
Beyond the top three
So we know that numbers, nonviolence and a conducive climate are crucial ingredients for success. But there were other — albeit less well-evidenced — factors that also emerged as potentially important and worth the consideration of social movements.
Diversity. Striking school children are not something you see every day. According to our experts, the arresting images of children waving banners gave a particularly strong signal to politicians. Protests felt inherently less political given that it was children, rather than experienced activists, who were protesting. These were not the “usual suspects.”
The school strikes were a particularly strong example of diversity of protesters (diversity, in this case, from protester norms), but more generally we also found that greater diversity is likely to increase the chance of protest success. This might be because greater diversity appeals to more of the public — meaning there’s more chance they will support or join a movement. It also gives a clear signal to policy makers that the issue has broad public support.
We think it’s interesting that — while most of the social movement experts we interviewed didn’t talk about diversity — all the policy makers thought that it was important. The three U.K. civil servants thought diversity was the second most important protest factor after size. They also felt that unexpected protesters — or people who don’t often protest, like school children — give a much stronger public opinion signal than so-called typical protesters.
A nonviolent radical flank. The “radical flank effect” refers to the influence that a radical faction of a movement can have on support for more moderate factions. It can be positive or negative. Overall, we believe a violent radical flank is likely to have negative overall consequences, while the effects of a nonviolent radical flank are more likely to be positive.
Evidence from a recent 2022 experimental study from Brent Simpson, Robb Willer and Matthew Feinberg supports the idea that radical flanks can have positive effects. They found that having a radical flank that uses radical tactics leads to a better impression of a more moderate flank, whereas a flank with a radical ideology (but not radical tactics) has no impact on the more moderate flank.
Other research by Eric Shuman and colleagues at the University of Groningen in the Netherlands found that actions that break social norms — like being disruptive or radical — may be the most effective way to persuade those who are resistant to change. They tested this idea of “constructive disruption” in a variety of experimental settings and found that it was more persuasive than either violent or typical (non disruptive) nonviolent actions. Evidence that radical flanks can also carry a cost comes from work by Elizabeth Tompkins. She found that a radical flank increases state repression, which in turn decreases mobilization — though she also points out that these effects are not “necessarily detrimental” to the overall success of a campaign.
Trigger events. These highly visible, often shocking actions that vividly reveal an existing problem to a wider public can have a significant impact. The arrest of Rosa Parks in 1955 and the murder of George Floyd in 2020 were very potent trigger events, both of which led to dramatic widespread protests. This suggests that it is important for protest groups to build the groundwork so that if an opportunity comes, they can grasp it and use the chance to build momentum.
There is very little direct research on trigger events, partly because their unpredictability makes them hard to study, but many of the experts we spoke with mentioned their importance. If they are right, social movement organizations would be wise to plan and organize for the need to respond speedily and convincingly, mobilizing in large volume at short notice.
Numbers really count… but they’re not the whole story
In their 2016 book, “This Is an Uprising,” authors Mark Engler and Paul Engler wrote persuasively of the impact mass movements can have when pursuing strategic nonviolence. Our previous research has also found that social movement organizations that use protest as a main tactic can significantly impact public opinion, voting behavior, public discourse and to a lesser degree, direct policy outcomes.
If protest is an important tool of influence, it is important to think about how best to go about it. Some of our key findings are not a total surprise: numbers matter and nonviolence is the best strategy. These findings support recommendations of many social movement thinkers and help to build a clear set of guidance as to some of the key decisions social movement strategists should take to make their campaigns effective. Our evidence also points to some less well known factors that are worth considering. Being able to act on trigger events, adding to the diversity of your protester base and expanding your movement to incorporate a nonviolent radical flank might all be valuable strategic additions.
One important note, however: Our research is not an exhaustive guide to what protest movements should do to be successful. Instead it should be seen as a summary of the current available evidence. Some factors are easier to measure than others. For example, it is a lot easier to get an idea of protest size than it is to assess an organization’s internal culture. So, there could be a bias towards some factors in the research.
Additionally, there are some pieces of evidence that are hard to act upon. The importance of timing, external factors and luck certainly leave some open questions. And while social movements may ebb and flow in cycles, it’s not clear what grassroots organizations should be doing in their fallow periods. Should they be focusing on internal organizational improvements or concentrating on building the sort of supporter base able to mobilize at short notice? These are areas where we think we could explore further.
As protests grow and spread around the world, becoming an ever more popular tactic for achieving social change, we need to understand them better. We hope that our research has added value in addressing some unanswered questions about the best approaches for protest movements in their efforts to improve the world.
If you liked this article, please donate $5 to keep NationofChange online through November.