What the Data Reveals About Effective Protests

A deep dive into the tactics and details that make a difference

As protests and other types of resistance have continued in Minneapolis and spread to dozens of cities across the US over the past few weeks, one question we might ask is: What makes a protest more likely to succeed?

Answering this question requires three steps: defining “protests,” defining “success,” and designing research in such a way that we tease out causes (not merely correlations) that might lead to more or less successful protests. That said, we must manage expectations. Just as knowing that eating more vegetables is likely protective against cancer doesn’t imply that eating carrots every day will guarantee good health, knowing that protests can help activists move toward their goals doesn’t imply that any given protest will guarantee the outcome we want.

Context also matters: the health benefits of our daily carrots will quickly become less relevant if we eat them while our house is on fire. Similarly, the effectiveness of protesting will vary not just by what’s going on in the protests but by what’s going on around them.

Finally, protests differ from carrots because it’s not always obvious what counts as one. Also unlike recognizing carrots, declaring something a “protest” rather than, say, a riot, an insurrection, or terrorism is itself highly politicized, and highly consequential for our ability both to learn about what works and to exercise our First (and other!) Amendment rights.

What counts as a “protest”?

The first thing to know about protests is that there is no single definition of them. For example, in the last few years, three different papers have been published in the Annual Review of Political Science, a prominent journal, that approach protests through three different lenses: nonviolent resistance, mass movement, or contentious politics. These are certainly overlapping categories, but each puts “protests” in the company of slightly different activities of dissent, from sit-ins and vigils to terrorism and civil war. (And these are good-faith efforts to understand political behavior; you can see how a not-so-good-faith actor might, hypothetically, label something most of us would call protest as “terrorism.”)

Instead of choosing one clean definition of protests, like “a gathering of 100 or more people within a specific amount of time, in a specific type of place, with a specific type of grievance, against a specific type of actor,” researchers on protests in political science tend to focus on the claims made in a protest-like activity and the extent and manner in which violence is used.

One reason for this is that the tactics by which we can express our demand for change are effectively infinite (I can boycott a product, blow a whistle, cast a “protest vote,” kick a taillight, drive a car, hold a sign, camp out in a city center, stand in front of a tank, or perform any number of other extremely small or extremely big activities). When we’re thinking about whether a protest “works,” it’s more useful to agree that a protest is a collective effort to change something, largely outside of formal political institutions. That something could range from a local policy to a national law to a regime itself.

As for whether a protest is violent: if you, as I once did, thought you could simply check a box labeled “yes” or “no” and move on with your protest-violence categorization, you were mistaken. One of the most comprehensive frameworks for characterizing the many ways violence may show up in a protest comes from Erica Chenoweth’s 2023 paper about the role of violence in nonviolent protests. (Chenoweth is also the scholar behind the “3.5%” rule, which we’ll get to in a moment.)

According to Chenoweth, resistance can be either “armed” or “unarmed.” Armed resistance is when individuals or groups physically attack opponents with lethal weapons as part of a collective contentious movement. Unarmed resistance, which is what most of us probably think of when we think of “protests,” is a form of collective contentious action in which people do not carry or display arms.

There are two types of unarmed resistance. One is “nonviolent resistance,” which can include marches, sit-ins, or other acts of collective resistance where nonviolence is “explicitly and deliberately maintained.” But another kind of unarmed resistance is “unarmed collective violence,” such as the destruction of property, vandalism, or fighting without the use of formal weapons (e.g., throwing a stone would be considered unarmed, while firing a gun would not).

Further, even when a movement is broadly considered to be unarmed and nonviolent, some actors or smaller subgroups might engage in either unarmed collective violence or armed resistance. Chenoweth and others refer to these as “violent flanks.”

And even further, the violent flanks can emerge from within the movement (for example, protesters spontaneously becoming violent, perhaps in response to an inciting behavior by their opponents) or from outside the movement (for example, violent groups external to the core movement might join in and bring arms or violence with them).

Making these distinctions might feel both tedious and trivializing (“Is that a chaise or a lounger here on this Titanic deck?”), but it’s crucial if we want to understand what sorts of protests are more likely to “work.”

What does it mean for a protest to be successful?

The first thing to know about what makes a protest successful is that there’s no single definition of “success.” The simplest way to think about it would be to ask whether the protest reached its specific goal (provided it has a specific goal). If we want to say that the protests and other acts of resistance in Minneapolis right now are specifically targeted at getting ICE out of the city, then “success” means ICE completely ceases operations in Minneapolis. Even then, we’d have to discuss whether it was the protests that led to ICE’s leaving, or it was the calls to representatives, or ICE’s inability to walk on actual ice, or some mix of these influences, or something else. But we could at least pretty cleanly check a box for “success.”

Whether something is a “success” can change over time, though. In the immediate aftermath of the Arab Spring, when massive demonstrations across many countries in the Middle East in 2011–12 led to regime changes in Tunisia, Egypt, Libya, and Yemen, these protests were widely considered to have been successes. After more than ten years of population displacement, backlash repression, ongoing violence, and other lingering effects, however, the protest movements are now considered to have been failures in some ways and mixed at best in others.

Anti-government demonstrators in Yemen in 2011

And many large protests widely thought to have had an impact did not have singular, clear, or “SMART” goals. Many important movements, such as the Women’s March on Washington in 2017 and the Million Man March in 1995, had many goals, from increasing awareness to building community to combatting stereotypes to preserving or expanding rights. It’s also not the case that failing to reach a specific goal means the movement has failed. Even if ICE doesn’t leave Minneapolis, there have already been successes in demonstrating solidarity and conviction, raising awareness, leading by example for other cities, protecting neighbors, and more.

One way to think about “success” is to identify the level of the movement’s goal. Are the protesters aiming to increase awareness, change public perception, alter a policy, or remove a leader? Different participants might have different goals: during the 2020 protests in the wake of the killings of Ahmaud Arbery, Breonna Taylor, and George Floyd, some participants might have considered success to mean fully defunding and abolishing the police, others might have considered success to mean eradicating white supremacy, and others still might have considered success to mean anything that raised awareness about the systemic injustice and violence inflicted on Black communities in America.

In general, while scholars do sometimes consider meeting stated goals a definition of success, most political science researchers in this space tend to think more broadly about the consequences of a protest as part of an interactive dynamic. Did the movement play a role in nudging or moving the issue in the direction(s) broadly preferred by the participants? For example, even if there was no policy change, were politicians who supported the movement more likely to get reelected? Do opinion polls suggest people in the areas of the protest changed their views in a way consistent with the goals of the protesters? Is the movement ongoing? Is progress still being made, however subtly or inconsistently, in the direction of the aims of the protesters?

It can be useful to think of protests as waxing and waning movements that are part of an ongoing “conversation”about oppression, repression, and resistance, as opposed to discrete events with clear start and end dates. Just as Egyptians are still living in the world created by the Arab Spring, we are still living in the America created by the Civil Rights Movement, the BLM protests, the pro-Palestinian protests, and more.

Okay, but what works?

I go through all of the above detail to emphasize two things. One, given how difficult it is to measure protests and define success, it’s remarkable that we have findings at all. And even if we could clearly and non-controversially define both protests and success, there would still be massive challenges to inferring from “A protest happened and then this outcome happened” that the protest caused the outcome. And, two, all findings must be interpreted with caution and care, as different studies will draw different lines around protests and success. It’s not realistic to expect sweeping, broad generalizations that will hold across all protests all the time.

With those cautions in place, here are some findings.

(1) The 3.5% rule is powerful, but there’s also more to it

The 3.5% rule is a 2013 research finding by our above hero, Erica Chenoweth, who studied a century of data about protest movements for regime change around the world and concluded that no government had withstood a challenge when 3.5% of the population was involved in it. This has led to hope that if only we could get 3.5% of the population involved in a protest in the US, it could lead to major changes from the Trump administration.

As Chenoweth later noted, however, the 3.5% rule refers specifically to protests aiming for regime change in authoritarian countries. It also does not refer to cumulative participation over time — it means that at least 3.5% of the population participates at the peak of the protest. And the organizers of these movements did not deliberately recruit 3.5% of the population in the expectation that doing so would lead to a tipping point; rather, if 3.5% of the population is openly participating in mass resistance, it’s likely a sign that the issue is really important and unlikely to go away.

Reaching 3.5% participation is also not necessarily the cause of regime change on its own; to have reached 3.5%, movement leaders probably also have strong messaging, organization, and discipline, which may be what causes the regime change. Chenoweth also clarifies that 3.5% of the US population would mean 11 million people in the country participating. The 2017 Women’s March, the largest nonviolent mass movement in American history, had an estimated four million participants, which was between 1% and 1.6% of the population at that time.

The good news in all this, however, is that it means movements that don’t reach 3.5% can also have successes.

(2) Nonviolence seems to be more effective than violence

In the universe of broadly nonviolent protests, the addition of violence (whether internal or external flank violence, as described above) tends on balance to harm the movement rather than help it. To note this is not to assign blame; violence can of course be provoked or initiated by an opponent. Nonetheless, at least in the empirical record in the US, the addition of violence to a protest is associated with less sympathy and support for the protest movement. And that relationship appears to hold beyond the US. Chenoweth finds that between 1900 and 2006, nonviolent protests around the world were twice as likely to be successful as violent protests.

(3) Protests are very good at increasing awareness, and in some cases can increase sympathy for the cause

Protest movements are public, involve lots of people, and are often disruptive by design. Seeing large numbers of individuals leave their homes, show up in the (perhaps freezing) streets, and demonstrate that they care can do a lot to shape a country’s conversations about politics. The political scientist Omar Wasow coined the phrase “agenda seeding” to describe how Civil Rights protesters between 1960 and 1972 used nonviolent, disruptive tactics to capture media attention, increase the frequency of conversations about race and civil rights, and help elect Democrats who were in line with the protest messaging at the time.

Further research suggests that counties that had more Tea Party protests in 2009 may have had more Republican votes in the 2010 midterm elections, as well as more conservative policymaking. And it’s possible these kinds of effects can last. Observational research suggests that white Americans in counties that experienced Civil Rights Movement protests are to this day more Democratic and more likely to support affirmative action. Scholars have also found that even the 1992 (violent) Los Angeles riot was associated with a measurable liberal shift in the next local election, and the authors suggest that shift also persists today.

(4) But protests (especially if violent) can reduce sympathy for the cause among its political opponents

While a fairly robust finding overall is that protests and related movements can increase support for an issue among co-partisans (e.g., protests for a liberal cause will increase support for that cause among other liberals), the opposite tends to be true for the opposition — meaning that liberal protests make conservatives even less sympathetic to the cause, and vice versa.

For example, Wasow’s finding above about agenda-seeding is exciting, but his research also shows that when nonviolent Civil Rights protests veered into violence, the movement lost sympathy among white Republicans and might even have tipped the presidential election in favor of Nixon.

Other research since the BLM protests of 2020 also finds that while Democrats became less favorable toward the police and had a greater perception of anti-Black discrimination, conservative-leaning Americans either remained unchanged or moved only slightly. The researchers concluded that the George Floyd protests may have further divided America and politicized attitudes about policing and race.

(5) “Success” can change over time

Further complicating things, however, is that attitudes can also shift over time. Early research after the January 6 insurrection (of course a term that is also debated) suggested that while it was largely considered a conservative movement, co-partisans seemed to distance themselves from the Republican Party and Donald Trump, at least as of the study’s publication in 2022. Since then, however, polling suggests Republican disapproval of January 6 has softened. Similarly, evidence suggests that in the years since the BLM protests, a “white grievance” movement has taken hold in backlash against the raised awareness of white supremacy and Black oppression.

To sum up our findings so far, then, protests can raise awareness, but they’re more likely to increase support on the side already inclined to agree with the protesters. They risk undermining potential support or inviting backlash among people inclined to disagree. The presence of violence, however sporadic, seems to increase that risk. This thinking about backlash is relatively new, however, and we’re working from just a few recent cases in the US, so it’s hard to draw definitive conclusions.

(6) Lightning-round findings

There is a ton of other research on protests, how they are perceived, who participates, and other things that can help or hurt. The findings below tend to be supported in fewer studies, mainly because they’re either newer or more niche, but nonetheless I find them interesting and hope you will, too:

1. Protests predominantly involving people who present as women tend to be perceived as less violent overall. But individual protesters who are women tend to be perceived as more deserving of repression if they behave in a “patriarchy-defiant” manner, such as by appearing to be feminists, to be independent, or otherwise to occupy roles other than wife and mother. (Believe it or not, this study came out two years before the killing of Renée Nicole Good.)

2. Protests may have effects well beyond the arena of politics. One study found that liberal-leaning parents, particularly those located near protest centers, were more likely to support initiatives to help children learn about their race or ethnicity.

3. Perceptions of outside interference can make things worse. One study argues that accusations of foreign intervention made protesters seem less credible, eroding support among both sympathizers and non-sympathizers, and inflamed nationalist fears. While it’s too early for formal studies, I’d imagine that current efforts to discredit the protests in Minneapolis as a result of a fraud network paying people to demonstrate would also fall under this category.

4. We’ve mostly been talking about external outcomes — what the protest caused or didn’t. But protests can also affect protesters themselves. Researchers have found that participation in a protest can increase our sense of political efficacy. And, while we tend to focus on the power of big movements, the feeling of efficacy can actually decrease as movements grow. (But I’m not saying to keep your movements small; I’m saying that even small movements can create lasting change, though perhaps not in the way we normally think about it.)

5. We are in what many scholars consider an “age of protest,” both in the US and in many parts of the world. Particularly during the second Trump administration, Chenoweth and colleagues have found that protests are occurring even in areas that did not see much protest activity before, including parts of “Trump country.”

Where does all of this leave us? Protests are hard to define. Success is hard to define. Causality is extremely difficult to pin down even with the best definitions. Yet scholarship all over the world, and from lots of different angles, seems to suggest that nonviolent, disciplined, large protests can make a meaningful difference in many ways, including through raising awareness, changing political attitudes, and causing disruptions that force conversations and change. But backlash is real, and one of the biggest risks is that even if the protest is overwhelmingly peaceful and not at all orchestrated by an outside influence, the mere perception that it’s not peaceful or not organic can, unfortunately, cause many people to dismiss, or even activate against, the sacred things we’re so diligently protesting for.