Does protest help a group’s cause? Do cities with strong Black protest movements improve their policing practices in Black communities? Do police respond more repressively in places with strong Black movements? Does mass incarceration reduce the capacity for Black protest? To answer these questions, we need to know how to measure the frequency and intensity of protest.
When several hundred people holding “Black Lives Matter” block an Interstate, that is pretty clearly a protest. But is it a protest when one person stands on a street corner holding a “Black Lives Matter” sign? When a group of religious leaders calls a press conference to issue a statement calling for police reform? What if the religious leaders issue the statement without holding the press conference? What if the street corner sign or the religious leaders’ statement is urging people to pray to God for guidance or attend church more often? What if the sign is advertising a car wash?
Is it a protest if people start throwing rocks and burn cars in response to police killing an unarmed person? Is it a protest if people throw rocks and burn cars after their team loses a football game? Is it a protest when several thousand people gather around the state capitol building to hear speeches about police reform? What if the same number are sitting inside an arena? Does it matter whether the speeches are only informative or also include calls for change? What if the speeches are about fighting cancer? What if they are supporting a political candidate who is running for office?
If there are rallies in seven cities on the same day about police reform, is that one protest or seven? What if there are rallies in the same city for seven days in a row, is that one long protest or seven smaller ones? What if there is an Occupy encampment, where people stay overnight and don’t go home?
Protest researchers count protests so that we can study them. We want to know how other factors affect the rise and fall of protests, how protests affect public policy, and how protests affect each other in diffusion or influence processes. But this research requires that we be able to measure and count them. As the opening examples indicate, this is harder than it seems at first. The core idea of a protest is something nearly everyone agrees on: we are all sure that several hundred people holding “Black Lives Matter” signs while blocking an Interstate is one. And that so is a less-disruptive march or rally for Black Lives Matter. But as you move toward edge cases, it gets harder and harder to be sure.
A protest as usually understood inherently involves three elements: (1) a protest issue or message, (2) a protest actor, and (3) a protest action form. Each of these has a center that everyone agrees on and “edge cases” that are more borderline. Moreover, many events meet two of the three criteria.
Protests carry messages, or symbolic content. Nearly everyone who studies protests says their purpose is to make a claim or express a grievance. More generally, I would say they are “message events.” That is, their purpose is to call for some sort of action on the part of people who are not participating in the event. They are about something; they have a purpose beyond the event itself. A block party or concert or religious service is not a message event in this sense, because its purpose is internal to the people attending the event.
But even this sentence might be challenged, as sometimes protest organizations hold meetings or social events or religious services with the intent of building up internal solidarity and motivation for action and of educating their group members and planning strategy. These internally-focused events are often important parts of protest movements, but we do not generally call them protest events. We usually add further restrictions on the message being conveyed, usually saying that the message is directed toward promoting or resisting social change. We usually do not include commercial messages (buy X), routine political messages (vote for X), or messages about consensual charitable causes (give money to X). We also typically exclude purely personal messages (my ex-spouse is a louse).
Human beings reading accounts of events will often disagree about whether a particular message promotes or resists social change. Our research team argued for an hour about whether community protests about denying license renewal for a coffee shop was an individual issue involving only the shop owner or a community collective issue. Protests about the killing of one individual person by police might be coded as an individual issue, although many would see each of these individual protests as part of a broader movement. Rallies about breast cancer or heart disease are usually thought of as charitable events that are not protests, but some ways of construing these health issues can and do involve criticisms of social policies. Gay pride parades clearly began as protests (the first was held in 1970 on the anniversary of the Stonewall uprising) and have clearly been part of the gay movement as it gained victories, even though the parades have become more institutionalized and ritualized over the years. St. Patrick’s Day parades are usually not treated as protest events, although they are the expression of ethnic pride by a particular group. Juneteenth celebrations similarly are usually not treated as protests, but protest messages can often be heard at Juneteenth events.
Political rallies are particularly problematic. Candidates running for office hold rallies a lot, and these rallies look a lot like protest rallies in that there is often conflictual language and criticism of social policies. Some non-incumbent candidates run as protest candidates and many have strong ties to one or more social movements or have social movement members with their issue signs as prominent members of their audiences. But all protest research projects seek to exclude “routine” political rallies from our protest data. This is partly because there are so many routine political rallies that we would be spending all our efforts on them and have few resources left for what we see as “real” protests. And it is partly because of the assumption that protests are undertaken by “outsiders”.
This takes us to the second criterion, protest actors. Protest researchers generally assume that protests are conducted by people who do not already have the power to create social policy. Protesters are people outside the power system, trying to get in. They are defined by most researchers as inherently being outside “ordinary” or “routine” political processes. Speeches, statements, and press conferences by elected officials, business managers, university officials, and others in power are typically defined as inherently not protests. As noted above, people running for electoral office are routinely excluded, although this is partly for the practical reasons that not making this exclusion would swamp our research. But there are elites that are included as protest actors, particularly the leaders of large movement organizations, such as the large labor, feminist, environmental and civil rights organizations that have substantial budgets and professional paid leaders that promote causes that are generally considered to be “movement” causes.
Movements that begin as volunteer decentralized uprisings do sometimes win victories that permit them to gain resources and become larger organizations that have more routine access to power and policy making. Labor unions, in particular, present an issue, as strikes and picket lines are “ordinary” features of labor struggle. Some research projects include actions by labor unions as protests, while other research projects categorically exclude them. Further, sometimes political elites stage protest actions. Most notably, within a few weeks of this writing, Democrats in Congress staged a sit-in about gun control. Trying to decide whether these elite actors are exercising their “routine” political power or are acting outside routines in protesting is an exercise in hair-splitting that research teams have to deal with.
The third criterion is protest forms. There are a standard set of forms that are usually understood as protests, including rally, march, demonstration, boycott, picket, vigil, blockade, strike, sit-in, petition. But most of these words in English are also used to refer to many different things that have nothing at all to do with protest, and even when they are used to refer to events, every one of these event forms describes a type of event that often has non-protest content. It has already been mentioned that election rallies are common. Holiday parades and protest marches are exactly the same form varying only in content. Vigils and blockades of roads occur for reasons other than protest. And many symbolic protests use creative novel tactics that do not fall into one of these standard forms.
Additionally, some of the forms are hard to pin down as events, especially boycotts and petitions. The famous bus boycotts of the Civil Rights movement occurred in specific cities and could be identified by the absence of Black people on buses, but would be hard to pin down as specific events. The famous Montgomery Bus Boycott lasted 381 days. Was that one event or 381? Other boycotts are felt only as declines in sales and can be very diffuse. Calling for a boycott can be a time-place located event, but the boycott itself is typically harder to locate in time and space, especially if it is a national-level boycott. Petitions and petition campaigns are similarly inherently diffuse in time and space, although delivering the petition can be pinned down as an event.
The standard forms don’t even include editorials, articles, books, and advertising campaigns that frequently undergird the efforts to promote change. Not to mention social media campaigns. These symbolic communications clearly convey messages, but are they “actions”? Most protest researchers say no, partly because it is hard to pin written words down to a time or place. Some researchers (such as Ruud Koopmans and Myra Marx Ferree) argue that we should focus our attention on words or “claims,” regardless of whether they are uttered at protests or in editorials, and follow these claims as they move from actor to actor or action to action. But this approach only works if you are following one specific set of words such as immigrants or abortion.
There is also the matter of collective violence and crime. The word “riot” encompasses collective fights and brawls, rock-throwing, arson and looting, and may include gunfire. Some “riots” are closely tied to protests, either spinning off from protests or arising around clear-cut grievances such as police violence. Others are understood to reflect underlying grievances about inequality and oppression even if their specific precipitating incident is unclear. Some “riots” involve communal attacks in which one group (historically Whites, in the US) collectively attacks and burns down another group’s housing. But some “riots” are entirely celebratory, most notably sports riots after victories. Some research teams exclude all forms of collective violence entirely from their protocols, while others select those incidents that seem to have protest or inter-group conflict content. If a team decides to code “riots,” then other decisions have to be made about boundaries. Most teams limit consideration to larger or collective events, and exclude fights between two people or individual acts of violence or property damage, regardless of the stated motivations of the perpetrators.
Our research team has had human coders deciding whether news stories describe protests so that these coded data can be used as training data to feed into an automated system that identifies protests in news sources. (This project is described at http://www.ssc.wisc.edu/~oliver/protest-research/mpeds/.) What we have learned is that cases that pose problems for humans also pose problems for the machine. Even though there is high agreement about the clear-cut cases of things that are protests, there is inherent ambiguity in around the edges of the concept in deciding whether something is a protest. Our review of the literature finds that different research teams have put different boundaries around the protest concept when they measure it. This means that it is important to check the “fine print” about what rules and boundaries researchers use in identifying protests. In addition, it is important to pay attention to the sampling procedures, that is, to what news sources were used and to what procedures were used to find stories about protests in those sources.