By Any Means Necessary? How Black and White Americans Evaluate Protest Tactics in Response to a Police Killing

Abstract

The majority of protests in support of racial justice are peaceful. However, since the racial reckoning of 2020, there has been debate about when and how exposure to violent or disruptive protest activities can shift public opinion towards a social movement. Using the Black Lives Matter Movement as a lens, we design a survey experiment to test the causal effects of different protest tactics on support for protesters and the movement itself among Black and white Americans. We include a control condition with no protest and manipulate the level of disruption in each treatment condition, ranging from a simple march in response to the police killing of an unarmed Black man to a protest in which participants set fire to an empty police headquarters. We use OLS regressions to estimate average treatment effects. Overall, we find that both Blacks and whites react negatively to more disruptive protests but whites tend to react more negatively than Blacks. Conversely, we also find that whites overall report more confidence in the ability of Black Lives Matter to facilitate racial equality after exposure to a protest, even when that protest employs disruptive tactics. We also test for the moderating effects of racial identity and racial resentment. We conclude by discussing the implications of our findings for the broader literature on social movements and public opinion.

Introduction

In the summer of 2020, the killings of George Floyd and other Black Americans at the hands of police put the cracks in American Democracy on full display. As protesters took to the streets, policy experts wondered how, if at all, public attitudes would shift in response to this “racial reckoning.” More specifically, experts questioned how the types of protests and demonstrations in response to police killings would affect the viability and success of Black Lives Matter (BLM) as a social movement. While many have speculated negative outcomes for the sustainability of BLM as a result of these disruptive tactics, empirical analyses do not find consistent or conclusive evidence for backlash effects (Brown and Mourão 2021; Enos, Kaufman, and Sands 2019a; Shuman et al. 2022; Wasow 2020a). Furthermore, scholars have consistently underrepresented the presence of Black Americans’ voices in judgments about Black Lives Matter and how race may shape these judgments. This puzzle brings us to our current research question: How may disruptive protest tactics affect support for social movements? And more specifically, under what conditions is support for a movement diminished (or elevated) because of disruptive protest activity?

Protests associated with the BLM movement have employed a variety of tactics to achieve their political goals. While most BLM protests since its inception have been peaceful, the level of disruption has differed considerably from protest to protest, ranging from marching with signs to blocking traffic at major intersections and, in some rare instances, property damage (Chenoweth and Pressman 2020; Clayton 2018; Taylor 2016a; Taylor 2016b). We take this nuance seriously in assessing how violent protests may condition support for a movement and, as a result, provide two main contributions in this article. First, we offer a typology of violence in protest tactics; specifically, we propose a spectrum of disruptive activities that could be used during a protest. This typology is important because public protests are often unpredictable and pluralistic in nature. It can be difficult to draw a clean line between actions taken explicitly on behalf of the protesters and actions which occur adjacent to the protest but are not necessarily associated with the larger movement strategy. Additionally, it is problematic to group both damage to property and injury to human beings as equally “violent” outcomes, especially in predicting how the public responds to disruptive acts of differing magnitudes. Our proposed framework allows for cleaner analytical tests of the conditions under which violent and nonviolent protest tactics may actually elicit different responses and contributes to the growing literature that analyzes how the American public’s responses to BLM and protests aimed at ameliorating police brutality are a function of protest activity (Brown and Mourão 2021; Metcalfe and Pickett 2022; Newman, Reny, and Merolla 2023; Reny and Newman 2021; Shuman et al. 2022).

This typology lays the ground for our second contribution: a test of competing hypotheses about how race moderates the effect of disruptive protests on perceptions of a movement. The debate around Black Lives Matter’s controversial tactics also speaks to a larger argument in the literature about the relative effectiveness of nonviolent and violent protests. This article considers both how specific protests tactics and race matter for public perceptions of a social movement.

In our analyses, we test whether the magnitude of the effect of protest violence on support for the BLM movement differs among white and Black respondents. However, due to mixed findings in the literature on backlash in response to violent protets, we remain exploratory about the direction of these effects. We also argue that attitudes about the behavior and actions of protesters may not be indicative of attitudes about the overall effectiveness of the movement associated with the protesters (Shuman et al. 2022). Thus, we expect the magnitude of treatment effects to differ depending on whether respondents are asked to evaluate the actions of protesters in general or assess the overall efectiveness of the Black Lives Matter movement.

To test the effects of different methods of protest activity in response to a police killing on perceptions of protesters and evaluations of their movement’s legitimacy, we use a survey experiment on a nationally diverse sample of Black and white Americans. We are the first, to our knowledge, to experimentally test the effects of protest activity on support for BLM on a sufficient over-sample of Black Americans. We manipulate the level of disruption in each experimental condition, ranging from no protest in response to the police killing of an unarmed Black man to a protest in which participants set fire to an empty police station. We use OLS regressions to estimate average treatment effects across conditions. Overall, we find few racial differences (in direction) for the effect of more disruptive tactics on approval and evaluation of the actions taken by protesters, as both Black and white viewers evaluated protesters more negatively to more disruptive protest activity. There was, however, a racial difference for the effect of disruptive activity on emotional reactions to protesters and perceptions of the Black Lives Matter Movement. Disruptive tactics appear to increase whites’ perceptions of the Black Lives Matter Movement as politically effective but have no effect on Black respondents’ views of BLM’s legitimacy. We conclude by discussing the implications of our findings for the study of protests, social movements, and public opinion.

Protest and the Political System

Protests are an important tool for social movement activists to raise awareness around issues and garner support for policy goals. Much of the work on social movements and protest has focused on assessing whether social movements are effective in achieving certain goals and, if so, identifying the mechanisms through which they “work.” Most scholars argue that protests work in part through raising the salience of an issue that directs public perception to a problem which, in turn, has downstream effects on political behaviors (Amenta et al. 2010; Enos, Kaufman, and Sands 2019b; Gause 2022b; Gillion and Soule 2018; Wasow 2020b). Some argue protests influence political behavior by making one issue appear more urgent to address than others (Carey, Branton, and Martinez-Ebers 2014). Wasow extends this further by arguing that the type of protest used by the movement shapes how media and political institutions interpret the issue and thereby respond to the movement’s demands (Wasow 2020b). If the choice of protest tactics (violent vs. nonviolent) affects a movement’s ability to achieve its political goals, it begs the question: under what conditions might it be advantageous to use one protest tactic over another? Answering this question is especially important because protesting is a costly form of political engagement, especially for minority communities (Gause 2022b). Despite their high costs of participation however, protests also have great potential to address racial, gender, and other inequalities (Banks, White, and McKenzie 2019; Davenport, Soule, and Armstrong 2011; Gause 2022b; Gillion and Soule 2018; Hinton 2021; Wouters and Walgrave 2017). The question of just how much cost a social movement or political group incurs from employing different protest tactics has prompted a scholarly debate with inconclusive results.

Violent Protests and Public Opinion

While there is some consensus that increased issue salience is one of the primary mechanisms through which protests garner support and influence political behaviors, there is less agreement on how the use of disruptive or violent tactics might alter the effectiveness of a protest. Scholars disagree about whether disruptive tactics are helpful in mobilizing constituencies and generating support for the movement’s goals. For instance, several argue that protest violence ultimately leads to negative outcomes for the movement. In the Black-led protests for civil rights in the 1960s, scholars find that counties in close proximity to nonviolent protests saw an increase in the Democratic vote share while those proximate to protest-initiated violence saw increases in Republican vote share, the party with policy goals antagonistic to the Civil Rights Movement (Wasow 2020b). Another example of violent protests eliciting a negative response is the January 6th insurrection led by Republican supporters of Donald Trump. According to one study, the insurrection prompted a significant decrease in user association with the Republican party on Twitter, ultimately diminishing outward displays of support for the riot (Eady, Hjorth, and Dinesen 2022). In the realm of public opinion, scholars argue that people tend to prefer nonviolent protests over violent ones, as violent tactics may increase perceptions of unreasonableness, and lower identification with the movement among out-group members (Bobo et al. 1994; Feinberg, Willer, and Kovacheff 2020; Orazani and Leidner 2018; Simpson, Willer, and Feinberg 2018).

Two recent studies examining the effect of BLM and Black-led protests empirically have arrived at similar results. For example, findings from an experiment that randomized the topic of the protest (BLM, pro-immigrant, pro-white Confederacy) and different protest characteristics (size, level of violence, duration) suggest that the public’s negative attitudes about protests are a function of both their racial attitudes and the extent to which more disruptive protests elevate their fear of protesters (Metcalfe and Pickett 2022). Similarly, an experimental study on civil rights protests finds that more disruptive and confrontational media frames that emphasize riots lowered support for the movement and police criticism (Brown and Mourão 2021). While these two studies suggest that increases in more disruptive protests lead to a decrease in support, they lack important nuance when it comes to the racial identity of the individual judging the protest, which may reverse this expected relationship. For instance, while Black Americans may not condone protest violence, they may have a higher disdain for how the problem of police brutality has affected their communities, thus allowing their desire to ameliorate police brutality to supersede their value of safety (Cooper 2014).

Other scholars, however, have found that, under certain conditions, disruptive or violent protests can be effective at engendering support and advancing a movement’s political goals. For instance, in the case of 1992 Los Angles riots in response to the beating of Rodney King, geographic proximity to a riot was associated with a liberal shift in policy preferences and overall increase in support for the Democratic party who were more in align with the policy goals of the movement (Enos, Kaufman, and Sands 2019b). Here the mechanism was mobilization, with the riots signaling the importance of political involvement for those in close proximity, particularly African American voters.

Consistent with these findings, others have found that exposure to BLM protests via living in a county with a mix of both violent and nonviolent protests predicted support for key policies and symbolic goals of the movement (Shuman et al. 2022). Even earlier analyses of the potential positive effects of disruptive protests find that under the right proportions of Black insurgent groups relative to white resistance, civil unrest or disruptive political behavior could lead to greater access to relief efforts or government programs (Fording 1997). Scholars have also found that the effects of a protest on political behavior depend on the race of the viewer. Perceptions of violence for example vary greatly according to race and identity. Protests that advocate for the issues of a dominant group, such as white Americans, may be perceived as less violent compared to protests that advocate for mostly minority groups and particularly Black Americans, even if they employ the same tactics (Manekin and Mitts 2022; Peay and Camarillo 2021).

Overall, existing literature is unsettled about whether violent protest tactics increase or diminish support for a movement and its policy goals. These inconsistencies could be for several reasons not yet addressed. For one, there are multiple scenarios where witnessing disruptive protests could either lead a respondent to be mobilized to support the movement and its goals or cause them to be alienated from a movement and want to diminish its activity. The effects of exposure to a violent or disruptive protest could vary by both racial identity of the person viewing the protest and the level of protest violence. Additionally, much of the social science literature has focused on how protests affect institutional outcomes such as policy, vote share, and candidate responsiveness. This leaves much to learn when it comes to the effects of specific protest tactics on public opinion (Amenta et al. 2010; Enos, Kaufman, and Sands 2019b; Gause 2022b; Gillion and Soule 2018; Wasow 2020b). It is likely that outcomes of increased or diminished support are both viable, and a methodological test that clarifies the conditions under which support for a certain protest tactics are effective and when they are ineffective may help explain public reactions to disruptive protest activity.

Black Lives Matter and the 2020 Racial Reckoning

To assess under what conditions support for a movement is diminished (or elevated) because of disruptive protest activity, we turn to the example of the 2020 Black Lives Matter protests movement and the so-called “racial reckoning.” Anxieties about the maintenance of “law and order” led to substantial increases in local policing of race-class subjugated Black and Brown communities since the 1960s (Soss and Weaver 2017). While the BLM movement shares lineage with previous Black-led social movements, like the Civil Rights movement, scholars have noted that 21st-century protests against police brutality among Black Americans have been characterized by a need to disrupt social norms (Bonilla and Tillery 2020; Harris 2015; Tillery 2019). In the years preceeding the 2020 racial reckoning, respectability politics had constrained the ways in which both Black and non-Black audiences expected Black-led movements to fight and comment on racial inequality. Respectability politics refers to the worldview that Black Americans should comport themselves in “acceptable” social behaviors to rebuff negative stereotypes of Black culture in America (Kerrison, Cobbina, and Bender 2018). While scholars have found age cohort moderates belief in respectability politics (e.g. Kerrison, Cobbina, and Bender 2018 finds Black Millennials tend to reject the ideology) respectability politics remains an important constraint on how Black-associated movements are evaluated and expected to perform. Scholars have found that Black politicians before 2020 and the killing of George Floyd often distanced themselves from direct discussions of police brutality to avoid being perceived as “anti-police” in front of white audiences (Stephens-Dougan 2020; Wade 2023). BLM throughout the 2020 racial reckoning and after has been seen as an open display of unmodified Black rage, and for better and worse, a direct disruption to this norm of respectability (Cooper 2014). As a political movement focused on ending police brutality however, BLM has maintained a hold on public consciousness and influenced demands for the equal treatment of Black Americans both domestically and abroad (Taylor 2016a). Though it would come to embody both a formal organization and an informal political commitment to ending police violence, Black Lives Matter started as a viral social media campaign in 2013. Founders Alicia Garza, Opal Tometti, and Patrice Cullors tweeted the hashtag “BlackLivesMatter” in response to the acquittal of George Zimmerman for the shooting deaths of Trayvon Martin, an unarmed Black teenager in Sanford, Florida, and the phrase spread across the web. Although members of the American public have demonstrated against police brutality in the United States prior to 2013 (i.e. in response to the beating of Rodney King), the term “BLM” soon became the organizing mantra and principle used by activists to rally support in the fight against police brutality. While BLM initially gained an immense online presence, this social movement soon found footing in real life, as massive demonstrations erupted in major American cities like New York City, Chicago, and Oakland sparking media attention and scholarly inquiry (Bonilla and Tillery 2020; Harris 2015; Merseth 2018; Peay and Camarillo 2021; Tillery 2019). BLM has even gained traction internationally, inspiring similar movements for racial justice and equality among other ethnic groups including Black Britons in the UK and Aboriginal communities in Australia (Allam 2020; Siddique and Skopeliti 2020). While the overwhelming majority of BLM protests have been peaceful, some have incorporated acts of disruption ranging from burning public buildings to looting local shopping centers (Khazan 2020; Stockman 2020; Tacopino 2020).

As a result of the different tactics employed by some participants in BLM protests, and BLM’s choice to disrupt norms around Black protests in the United States, the movement itself has received some criticism from Black political leaders and public intellectuals. For example, a columnist in the Washington Post details how as an activist in the Civil Rights Movement, many like her “admire the cause and courage of these young activists but fundamentally disagree with their approach.” She goes on to say that her generation, that encompasses the baby boomers, often see BLM’s tactics as “confrontational and divisive” (Reynolds 2015). Similarly, former president Barack Obama said of the movement in 2012 that activists “can’t just keep on yelling” if they want to create meaningful change (Shear and Stack 2016). Despite the criticism BLM has received for some of its less common but controversial tactics, the academic literature on the response to this movement is limited. While still informative, most of the scholarly literature on BLM does not analyze these contrasting opinions in response to BLM empirically (Harris 2015; Lindsey 2015; Rickford 2016).

While these analyses are integral for contextualizing the significance of BLM and what it means for American democracy, more nuanced empirical tests are needed to clarify the conditions under which support for BLM may be a function of the types of protest activity members take and how an individual’s racial identity may condition their responses. While the overwhelming majority of Black Americans support political action to address police violence, recent polls detail that support for BLM post the 2020 racial reckoning has declined, though not as much among Blacks as it has among other racial groups, particularly whites (Chudy and Jefferson 2021). This may indicate variation in the opinion Black Americans have about Black Lives Matter and whether this movement is the best means through which to advocate for the end of police brutality.

Theoretical Expectations: A Proposed Typology of Protest Violence

There are strong reasons we may expect to see an increase in support for a social movement in response to disruptive tactics, but there is also substantial evidence to the contrary. These expectations are further complicated by existing literature’s opaque theorization about what constitutes a “violent” or “nonviolent” protest. For instance, (Brown and Mourão 2021) described their conception of violent protest as a civil rights-related riot that grew disruptive and unruly with a picture of a burning police car with no comparison to other activities that may viewed as disruptive. Similarly, (Metcalfe and Pickett 2022) defined violence in protests as whether civilians were hurt and property was damaged and again did not provide a range of other activities that may be considered disruptive. It could be the case that some activities are viewed as both more violent and more justifiable than others, making a binary categorization of “violent/nonviolent” incomplete. For example, a Black mother who is worried about the safety of her Black child may view looting or property damage affiliated with a BLM protest as no more “violent” than police brutality itself (Osterweil 2020). We argue that clarity on the conditions under which we should expect more or less violent protests to cause an increase or decrease in support for the movement comes from, in part, how we define and classify violent or disruptive protests. First, public protests are often unpredictable and pluralistic in nature. It can be difficult to draw a clean line between actions taken explicitly on behalf of the protesters and actions which occur adjacent to the protest but are not necessarily associated with a larger movement strategy. Additionally, it is problematic to group both damage to property and injury to human beings as equally “violent” protest outcomes.

Within the category of property damage alone, there may be substantial variation. In the case of Black Lives Matter, looting (which typically involves the destruction of local business property) may signal broader economic deprivation and unrest in addition to demands for police accountability (Osterweil 2020). On the other hand, damage to property with a specific symbolic meaning like a police station may be a clearer signal of frustration with the police (Stockman 2020). Additionally, actions typically categorized as nonviolent might also vary in the extent to which they disrupt and draw public attention. In an effort to account for the inherent imprecision of categorizing protests as discretely “violent” or “nonviolent,” we refer to a spectrum of disruptive activities that could be used during a protest. Figure 1 visualizes the spectrum of disruptive actions that have been reported in association with protests against police violence (affiliated with the Black Lives Matter Movement) since 2020. ¹ We leverage both low-level disruption tactics like marching as well as high-level tactics like property damage to assess how, if at all, these varied tactics affect public opinion differently. This typology helps explain why there may be a myriad of reactions to protests broadly categorized as “violent” or “nonviolent.”

Figure 1. Spectrum of Disruptive Protest Tactics

Our typology also offers a more inclusive approach to the types of actions respondents may interpret as politically relevant. At times, less traditional acts not defined as political behaviors within the literature may be nevertheless interpreted as legitimate by audiences who are personally impacted by the subject of the protests. Conceiving of violence in protest activity as a spectrum rather than a discrete, binary category allows future research to build upon our example of the Black Lives Matter movement and update the scale to include other tactics that groups employ over time. This typology may also be useful to isolate certain features of protest disruption while keeping others constant to understand which factors may be interpreted as violent and relevant politically, versus others that are violent but matter less for political judgments.

Racial Identity and Protest Perceptions

In addition to conceptualizing protest activity as a spectrum, we also contend that racial identity is an important force that determines reactions to disruptive protests. Overall, it is hard to disentangle perceptions of violence from racial identity. This is particularly true for BLM as a Black-associated organization that is focused on racial justice in response to the killings of Black Americans. We argue that the racially provocative nature of BLM should stimulate differences in reactions to the movement’s protests among Black and white Americans. Across many domains of public opinion, there exists a stark racial divide between Black and white Americans, and this applies to perceptions of BLM (Hutchings and Valentino 2004). A large portion of this divide is explained by partisan polarization and sorting, as scholars find responses to BLM protests are a function of preexisting political orientations such as liberal ideology and Democratic affiliation (Reny and Newman 2021; Stout 2020). Nonetheless, the racial differences in perception of BLM have been well documented. Pew Research data show that at baseline, 65% of Black Americans support BLM, while only 40% of white Americans express support (Horowitz and Livingston 2016). Other data also confirm these differences in perceptions of crime, which may be extended to perceptions of violent protests (Gramlich 2019). Additionally, scholars have found that believing a protest will end in violence is a function of the amount of minorities visibly seen as a part of the protest (Peay and Camarillo 2021).

From this literature, two possible outcomes emerge for how race may moderate the effects of a disruptive protest on perceptions of BLM. First, it could be the case that there is a racial divide in perceptions of protest violence and public opinion about BLM, where Black Americans are more tolerant of protest violence than white Americans. Data show that whites are less supportive of BLM at baseline, making their support for more disruptive BLM protests less likely (Horowitz and Livingston 2016). White Americans’ support of BLM, as with other political beliefs, is often a function of their racial attitudes. For instance, white Americans’ attitudes about whether white Americans are a disadvantaged group in the United States, and whether they fare worse in the job market compared to Black Americans, were associated with increased support for Donald Trump in 2016 (Bunyasi 2019). White Americans also have a wider range of responses to racial appeals depending on their partisan and ideological affiliations (Stout 2020), and their racial prejudices are significantly associated with support for counter-movements to BLM such as Blue Lives Matter (Newman, Reny, and Merolla 2023). Thus, white Americans may be more predisposed to oppose more disruptive protests from BLM than Black Americans, especially if violent protests prime their racial attitudes.

In comparison to white Americans, Black Americans may view less disruptive tactics as frustrating, especially among those who feel it is a more passive approach and prefer more disruptive displays of protest that send dramatic signals of issue salience to the American public. Black Americans have a history of participating in counter-movement politics in response to threatening political environments (Towler and Parker 2018). In the current case, many Black Americans have argued that the situation with George Floyd is another reminder of how resisting passively allows police brutality to continue to have violent outcomes on their communities (ODonnell 2020). These arguments are also present in media discussions of BLM. The heightened salience of BLM created one of the most prolific debates about the legitimacy of disruptive and even violent protests. Some argue, for example, that BLM’s association with controversial tactics can be likened to the Stonewall riots, which succeeded in securing some LGTBQ rights (ODonnell 2020). This may also be the case among Black Americans, given that the George Floyd protests did ultimately lead to the conviction of Derek Chauvin, the police officer who was recorded taking George Floyd’s life. In addition, BLM has also appeared to have other victories, such as several cities across the United States restricting police use of chokeholds (Menjivar 2020; Senter and Dewan 2022).

On the other hand, it is possible that the racial divide over disruptive protest actions runs in the opposite direction. That is, Black Americans might have a lower tolerance for disruptive actions than white Americans. For example, some Black Americans may believe that BLM needs to follow the peaceful and incremental change approach that, arguably, lead to widespread support for the Civil Rights Movement (Reynolds 2015). Concern about the decorum of protesters on behalf of political issues affecting Black people may also be a manifestation of respectability politics in the Black community (Jefferson 2022). Further, it is possible that white Americans may not feel the same need to police how BLM performs protests as Black Americans do. Even though white Americans support BLM less than Black Americans, their overall support was at its peak in 2020, right after the “summer of racial reckoning” which encompassed one of the most the dramatic increases in both violent and nonviolent BLM protests (Chudy and Jefferson 2021).

Given the evidence for both outcomes, we offer a test of two competing hypotheses. In other words, we expect the magnitude of the effect of protest violence on support for BLM to differ among white and Black respondents. It is possible that Black Americans will be more positive and approving of disruptive protests than white Americans ( ${H_{1a}}$ ), or the reverse, white Americans may be more positive and approving of disruptive protests than Black Americans( ${H_{1b}}$ ).

Further, while there is some evidence that protest violence signals how a movement should be interpreted (Wasow 2020b), it is not clear which outcome variables might be affected by this signaling. It is possible that disruptive actions only affect judgments of the protesters themselves and have no effect on perceptions of the overall movement’s legitimacy. The differential effects of exposure to protests on public opinion are documented in other work about the Black Lives Matter movement which finds being in proximity to violent BLM protests in liberal areas resulted in an increase in white conservatives’ support for policy goals but had no effect on their prejudice towards Black people (Shuman et al. 2022). We leave room for the possibility that observers may recoil at the thought of a protester committing arson during a Black Lives Matter protest but, nonetheless, may not view this incident as reflective of the BLM movement as a whole by examining two different outcomes related to protest effectiveness. In determining the effect of disruptive protests on perceptions of protest effectiveness, we again remain exploratory due to the literature’s mixed findings (Brown and Mourão 2021; Enos, Kaufman, and Sands 2019b; Shuman et al. 2022; Wasow 2020a). Thus, we expect exposure to the treatments to either increase ( ${H_{2a}}$ ) or decrease ( ${H_{2b}}$ ) perceptions of the effectiveness of protesting generally and the Black Lives Matter movement specifically.

Data and Methods

Data

In order to test our predictions, we conducted a pre-registered survey experiment on Lucid Marketplace in December 2022 ² . The purpose of our empirical test is to provide clarity about the conditions under which disruptive protest tactics may significantly change support for a social movement. We designed the survey experiment to help establish causality while controlling for confounding factors. The survey was administered online to a nationally diverse non-probability sample of N = 3004 American adults. Respondents were acquired from and paid by Lucid Marketplace for their participation. Roughly half of the sample (n = 1435) identified as white and the other half of the sample (N = 1563) identified as Black. The mean age for the entire sample was 44 and the mean level of education was “some college.” 54% of the final sample identified as a Democrat, 21% as Republican and the remainder as independents or with a third party. Table 1 summarizes the demographics of the sample. While our sample is by no means nationally representative (e.g. our study has a larger proportion of white Democrats than the general population), non-representative convenience samples are nonetheless useful for hypothesis testing in experimental designs (Mullinix et al. 2015). For a full breakdown of the sample demographics by racial group, see Appendix D.

Table 1. Summary Statistics

Survey Design and Treatments

After completing a demographic questionnaire, respondents were randomly shown one of five mock news articles modeled after the online reporting of the Associated Press. We created a fictitious news organization, The St. Louis Daily News, as the source of the reporting. For all analyses except protester evaluations and emotional reactions, the reference or control condition is an article about climate change that includes picture of a polar bear floating on a block of ice. The four treatment conditions each describe the killing of an unarmed Black man by a white police officer in St. Louis, MO, and an accompanying protest in reaction to the police shooting. Table 2 summarizes the manipulations across conditions.

Table 2. Summary of Manipulations

In order to isolate the effects of protest tactics on perceptions of the protest, we heightened the level of disruption across treatment. In the marching condition, protesters gathered and marched at a park in downtown St. Louis to raise awareness about the shooting and demand police accountability The blocking condition described a similar protest where, instead of ending the march at the park, participants went on to block a major highway exit. In the looting condition, the protesters stole items from the convenience store where the shooting took place in addition to their march. Finally, in the burning condition, the article describes protesters setting fire to the St. Louis Metropolitan Police Department headquarters in addition to the march. All three disruptive actions used in these treatments are based on reporting from actual Black Lives Matter Protests that took place between 2020 and 2021 (Khazan 2020; Stockman 2020). However, we emphasize that these disruptive protests (while heavily reported) are outliers. Violence is rarely used as a tactic for racial justice protests in the United States, especially among protests associated with the Black Lives Matter Movement (Taylor 2016a). Our hypotheses, however, concern the variation in reactions to protests according to protest tactics. To this end, it is necessary for our treatments to include some violent conditions in order to make comparisons to less disruptive or nonviolent conditions. This decision means that our violent conditions are not indicative of a typical BLM protest and should not be interpreted to suggest that violence of any kind is a common tactic employed at protests affiliated with the Black Lives Matter Movement. Nonetheless, the treatments provide us with an opportunity to test if and how different levels of violence elicit different evaluations of the protesters specifically and the movement as a whole.

Across all four treatment conditions, no person is injured or harmed by the protesters’ actions. To enhance external validity, each article text was paired with a photograph. In the control condition, the photo was a still of a polar bear floating on a block of ice. In the treatment conditions, each photograph depicted the protest activity (e.g. marching, blocking interstate). We intentionally selected photographs that obscure the race of the protesters as scholarship has found race of the protesters to significantly alter responses (Peay and Camarillo 2021). Each of the photographs used in the treatments was taken from a far distance, obscuring the racial identity of the protesters (see Appendix 1). After reading the article, respondents completed a post-test questionnaire which measured our dependent variables as well as a manipulation check. For our dependent variables, we measured two dimensions of public support: evaluations of the protesters’ actions and perceptions of protests and the Black Lives Matter movement as a whole. This includes questions on the extent to which respondents felt proud, hopeful, ashamed, or embarrassed of the protesters and felt their actions were justified or reasonable. Additionally, we asked whether they felt protests were effective at achieving political goals and if the Black Lives Matter Movement would be effective at advancing racial equality. After these questions, respondents were debriefed and informed that the article they read was fictitious.

Analysis

We proposed two sets of competing hypotheses concerning the effects of race and disruptive activity on perceptions of the protests. Here, we may expect Black Americans will be more positive and approving of disruptive protests than white Americans ( ${H_{1a}}$ ), or the reverse, that white Americans will be more positive and approving of disruptive protests than Black Americans( ${H_{1b}}$ ). Additionally, we hypothesized that disruptive protests may increase ( ${H_{2a}}$ ) or decrease ( ${H_{2b}}$ ) perceptions of the effectiveness of protesting generally and the Black Lives Matter movement specifically. We used OLS regression to estimate average treatment effects for each condition, regressing a treatment dummy variable (which indicated whether or not a respondent was assigned to each of the treatment groups) onto each of our dependent variables. All dependent variables were re-coded to range from 0 to 1 with 1 indicating stronger support. Because respondents in the control condition (which discussed climate change) were not asked to evaluate protester actions, the marching condition is used as the reference category for the analyses of protester evaluations. Using the marching condition as the reference category allows us to compare how respondents evaluate protests across our proposed typology of disruptive tactics from a march (which involves a relatively low level of public disruption) to burning a police building (a relatively high level of public disruption). This analytical approach sheds light on how viewers evaluate protest actions across the spectrum of disruption. For the remaining dependent variables (protest effectiveness and Black Lives Matter effectiveness) we compare the treatment conditions to the polar bear control. This allows us to test the effect of exposure to a protest on support for the movement relative to having no exposure to a protest at all.

Results

Emotional Reactions and Evaluations of Protesters

After exposure, respondents in the protest conditions (marching, blocking, looting, and burning) were asked how proud, hopeful, ashamed, and embarrassed the actions of the protesters made them feel. Respondents in these protest conditions were also asked if they believed the protesters were justified and reasonable and if they approved of the protesters’ actions. Figure 2 shows the effects of race interacted with the blocking, looting, and burning conditions on the emotional reactions to the protesters, where the baseline is the marching condition. We converted the point estimates for each regression into predicted probabilities for readability. First, we can see from Figure 2 that overall, the burning and looting conditions significantly decreased pride and hope and increased shame and embarrassment across the sample relative to the marching and blocking conditions.

Figure 2. Effects of Race and Treatment on Emotional Reactions to Protesters

Among Black Americans, the predicted levels of pride in the marching and blocking conditions was around 0.6 on a 0 to 1 scale. Levels of hope were similarly high for Black Americans in these conditions, between 0.55 and 0.6. In contrast, feelings of hope and pride hovered around 0.4 for Blacks in the more disruptive conditions, indicating a roughly 10% decrease in positive emotions for Blacks exposed to the looting or burning conditions. In comparison, whites also showed less pride and hope in the more disruptive conditions. However, the decreases in pride and hope between whites in the highly disruptive conditions compared to whites in the less disruptive conditions is smaller than the differences for Black Americans. In other words, the disruptive treatments had a much weaker effect on whitess’ levels of pride and and hope than they did on Blacks.

In contrast however, white Americans reported more shame and embarrassment, than their Black counterparts. Among Black Americans, the two most disruptive conditions elicited about 10 percentage points more shame and embarrassment realtive to marching, with the looting condition eliciting the most shame and embarrassment. The differences between Black and white levels of shame and embarrassment in the least disruptive conditions is negligible. However, in the highly disruptive conditions, whites showed a larger increase in shame and embarrassment. Compared to the marching condition for example, whites reported 20% more shame and nearly 30% more embarrassment in the looting conditions.

Overall, the looting and burning conditions elicited significantly more negative emotional reactions among Blacks and whites, but the magnitude of the these effects were greater among white Americans. While both groups reported more shame and embarrassment in the more disruptive conditions, the effects sizes for whites were larger. In contrast, for feelings of hope and pride, whites had much weaker emotional reactions to the treatments, showing little change in perceptions between the least and most disruptive conditions. Black respondents however made a greater distinction between non-disruptive tactics like blocking and marching and the disruptive tactics (looting and burning) that involved property damage. These results concerning emotional reactions to more disruptive protests show partial support for ( ${H_{1a}}$ ) that Black Americans have more positive (or in this case less negative) reactions to disruptive protests than white Americans.

Black and white Americans also differ in their perceptions of the validity of the tactics used by the protesters. However, the effects of race are much smaller on these outcomes. Figure 3 shows the effects of the treatments on approval of the protesters’ actions and belief that the protesters were reasonable and justified. Across the entire sample, respondents were significantly less likely to approve of burning or looting than they were to approve of marching and blocking. Both Blacks and whites viewed the action of blocking an interstate exit as equally justifiable and reasonable as marching in a zoned-off area. The interaction between race and treatment is significant only for the reasonable actions outcome in the burning and looting conditions. That is, race significantly moderated perceptions of the protester’s actions as reasonable when the protesters employed the most disruptive tactics. Here, whites are significantly less likely to view the more disruptive protest tactics as reasonable. This is exemplified by whites’ 20% decrease in the view that the protester’s actions are reasonable in the burning condition relative to marching. There was a similar decrease among whites in perceptions of protester reasonableness in the looting condition, again relative to marching. For the rest of the outcomes, the coefficient on race is not significant, indicating that the Blacks and whites were equally likely to approve of protester actions and view them as justified (see Appendix G for full regression tales of these results). The results concerning protester evaluations again provide partial support for ${H_{1a}}$ , suggesting that when there is a significant racial difference in perception, white Americans view more disruptive protests as less valid than Black Americans.

Figure 3. Effects of Race and Treatment on Evaluations of Protesters. Note: Points are predicted probabilities. Horizontal bars are 95% confidence intervals

While the aggregate results for Blacks and whites were compelling, we also considered how racial attitudes could moderate the effects of treatments. For whites, we tested the effect of the interaction between racial resentment and the treatment conditions on evaluations of protesters. To do this, we performed the moderation analysis separately on the white subsample. We found that whites low on racial resentment were significantly less likely to approve of burning and looting tactics and less likely to see these actions as justified (p  $ \lt $  .01). However, whites high on racial resentment did not vary significantly in their evaluations of these tactics. That is, whites high on racial resentment saw no significant differences between looting and burning in terms of reason or justification. Importantly, whites high on racial resentment were much less likely to approve of the baseline protest activity of marching. As we move across the spectrum of protest disruption, racially resentful whites see very little difference between anti-police brutality protesters gathering for a march and protesters burning down a police station (see Appendix F). We also tested for moderating effects among Black Americans, specifically for their level of linked fate and their support for respectability politics. We found little variation in reactions to the protesters across levels of respectability and linked fate (see Appendix E). Overall, Black Americans’ reactions to the protests were not conditional on the indicators of identity salience or intra-group attitudes.

Effectiveness of Protests and the BLM Movement

Exposure to different protest tactics also shifted perceptions of the effectiveness of the Black Lives Matter Movement but did not shift general attitudes about protests as a political tool. For these analyses, we compared each of the treatment conditions to the control condition which did not discuss racial justice protests at all. Here we found significant differences between whites and Blacks on the question of BLM’s effectiveness. From the left columns of Figure 4, we can see exposure to the treatments did not affect Black Americans’ perceptions of the BLM movement as effective or the perceptions of protests in general as an effective political tool. We can further see from the right-hand side columns of Figure 4 that the predicted level of perceived effectiveness of BLM for whites was overall lower than Blacks across all conditions.

Figure 4. Treatment Effects on Perceived Effectiveness of Protesting and BLM

However, relative to the control condition where no racial justice protest was cued, whites were significantly more likely to believe the Black Lives Matter movement would be effective at advancing racial equality in America in all four treatment conditions as indicated by the positive and significant (p  $ \lt \;$ .05) coefficients on the interaction terms between race and treatment (a tabular summary of the results in Figure 4 is available in Table 17 of Appendix G). Here we see the treatment effect was largest ( $\beta $  = .08) in the burning condition. While the predicted level of perceived BLM effectiveness for whites was overall lower than Blacks across treatment conditions, more disruptive protests did significantly increase whites’ perception of Black Lives Matter as politically effective. These results lend partial support to ${H_{2a}}$ just among whites.

Discussion

Overall, our study demonstrates that reactions to racial justice protests depend on the tactics employed by the protesters and the racial identity of the observer. First, we find significant differences in the effect of disruptive protest tactics on negative emotional reactions to protesters among Blacks and whites. Tactics like looting a store and burning a building elicited significantly more negative emotional reactions from both Blacks and whites relative to the less disruptive action of marching. The treatment effects for whites, however, were larger and significantly more negative than Black Americans’ reactions. We find similar results for the perceived validity of protesters (perceived approval, justified actions) between white and Black reactions to more disruptive protests. In combination, these results indicate support for ( ${H_{1a}}$ ) that Black Americans have more positive and approving (or in this case less negative) reactions to disruptive protests than white Americans.

Furthermore, race moderated the treatment effects on perceptions of the effectiveness of protesting generally and the success of the Black Lives Matter movement specifically. This finding is especially interesting in light of the 2020 racial justice protests and the “racial reckoning.” Despite whites being overall less likely to view BLM as effective, all four treatment conditions significantly increased whites’ perceptions that BLM would be successful at facilitating racial equality. These effects are modest ( $p\; \lt \;.05$ ) but they indicate that disruptive protest tactics in and of themselves do not necessarily incur a penalty from whites, indicating support for ( ${H_{2a}}$ ) among white Americans. Concerns that such tactics might isolate potential allies to BLM are legitimate but, given the data, may be overstated. At best, this finding suggests that whites may see the BLM movement as an effective means of raising the salience of racialized police violence as a political issue.

This finding is also consistent with Gause (2022b)’s work which shows that protests associated with disadvantaged groups are more effective at catalyzing policy changes than protests on behalf of privileged groups. Our findings indicate that whites, as members of the dominant group, may be aware of the “advantage of disadvantage” and react to disruptive or violent protests with this in mind. Conversely, Black Americans’ perceptions of the effectiveness of the Black Lives Matter movement are not significantly affected by exposure to the treatments. That is, even though Black Americans may disapprove of more disruptive actions, this disapproval does not necessarily translate into negative evaluations of the BLM’s ability to effect political changes.

While again we emphasize the peaceful nature of the overwhelming majority of these protests, reporting on the summer of 2020 gave disproportionate airtime to the handful of protests that involved damage to private and public property, including businesses and police stations (Khazan 2020; Stockman 2020). Further, reporting on protests affiliated with Black Americans is more likely to draw attention to violence, regardless of the protesters’ actions (Gause, Moore, and Ostfeld 2023). Thus, the ways in which protesters engage in political activity are politically relevant and may have the possibility to shape how the movement is perceived.

In addition to our main results, we also find that whites’ responses to the protest were moderated by their levels of racial resentment. Further analyses of the interaction between treatment and racial resentment for whites revealed that the treatments were most effective for whites low on racial resentment. In contrast, racially resentful whites viewed protests against police killings as unreasonable and unjustified whether they employed disruptive tactics or not. Among Blacks, support for protesters was similar among those with low and high levels of linked fate and low and high levels of support for respectability politics. Blacks overall were more supportive of the protesters, but had strong emotional reactions to the more disruptive tactics (See Supplementary Material for regression tables of these results).

Overall, these findings complicate the media narrative that disruptive BLM protests will generate increased opposition to the movement, especially from non-Black observers. It seems that the individuals we would anticipate to be most sympathetic to disruptive protest tactics (Blacks and white Americans low on racial resentment or “racial liberals”) are also likely to view increasingly disruptive actions more negatively. This is not to suggest that Black and white racial liberals are more antagonistic to racial justice protests than conservatives; after all, baseline support for the protests is much higher for Blacks and whites low on racial resentment. The findings do however suggest that concerns that the news coverage of the small number of Black Lives Matter protests that lead to violence would increase conservative animosity towards the issue of police brutality may have been overblown.

Future work should further investigate this finding by experimenting with different protest scenarios. For example, researchers may want to examine if and how negative affect for protesters translates to weakened support for policies associated with the protesters’ demands. Future work should also consider what drives the negative responses to disruptive protest tactics among the groups who should be most sympathetic towards the protesters. It is possible that the decreased feelings of pride and hope among white racial liberals and Blacks in the highly disruptive conditions reflect their fear of backlash from white conservatives. Finally, a non-trivial weakness of this study’s design is that it cannot account for how violence enacted against protesters, either at the hands of police or by other bystanders, might affect support for the protesters’ tactics. Black Lives Matter protesters have historically been confronted with disproportionately aggressive policing even when the protesters are not breaking laws or causing major disruption. This was especially the case in many of the Black Lives Matter protests during 2020.

Generally, the public’s evaluations of Black liberation movements like Black Lives Matter are complicated. Our findings demonstrate the importance of applying nuanced analyses to the study of the tactics employed by activists in these movements. The binary categorization of protests as either violent or nonviolent obscures the varied and complex ways that protesters advocate for their demands. Our proposed typology further interrupts the dichotomy between violent and nonviolent tactics by defining property damage as an escalation of nonviolent tactics with the goal of creating disruption to draw attention to a cause. Our findings also complicate the debate about the political utility of violence in protests. The condition in which protesters looted a convenience store consistently elicited more negative reactions than the burning condition which was also disruptive (perhaps even moreso than looting). This finding shows that viewers distinguish protest activities that implicate broader socioeconomic inequities from protest activities narrowly targeted at police. While the former may elicit more negative reactions, it has little effect on perceptions of the movement as politically effective.

Our analyses also include important limitations to note for future research. There is a chance that respondents may be interpreting the more disruptive tactics as those being taken by actors who are not affiliated with organized protest activities. This could explain why Black Americans were less supportive of protesters in the conditions that described more disruptive actions. While we attempted to address this by explicitly stating in the treatment vignette that the disruptive actions were initiated by participants in the protest, we recognize that we do not directly test for the perception that “outside actors” were involved with the disruptive actions. Future research should take more extensive measures to clarify who is responsible for disruptive actions associated with a protest and to assess how much observers attribute disruptive actions to the organizers of the protest. Nonetheless, we do not believe this limitation weakens the external validity of our results. Television and print media reporting on the Black Lives Matter Movement tends to characterize its participants as riotous or violent regardless of the actions they take (Banks, White, and McKenzie 2019). While this conflation of BLM with riots is likely a reflection of racial bias in reporting (Gause, Moore, and Ostfeld 2023), our paper is not concerned with giving power to these news biases. Instead, we are primarily concerned with how people interpret exposure to this information. However, we do believe that identifying the extent to which disruptive actions are attributed to the main protesters or to a radical offshoot of the main protest is an important question that should be investigated in future research.

Lastly, while our findings point to an important racial difference in perceptions of disruptive protest, they do contradict those of Kaufman, and Sands (2019b) who find proximity to the Rodney King riots had a mobilizing effect, primarily driven by African Americans who shared an identity with the rioters. This is because overall support and perception of BLM among Black Americans in our sample did not increase positively as disruption increased. It could be the case that the racial ambiguity of the protesters performing the disruptive tactics in our experiment dwindled the extent to which Black Americans identified with the protesters. While other work has examined the effect of racial composition of BLM protesters on white American attitudes of BLM (Peay and Camarillo 2021), future research may want to investigate this among Black Americans to see if an increase of Black protesters changes their reactions to protesters when their tactics take more disruptive turns. Additionally, our theoretical predictions are heavily tied to the fact that BLM is a racially salient protest that may activate racial thinking among white and Black Americans in unique ways. Our typology of how salient identities like race inform how respondents interpret disruptive protests may be useful for other categories of protests, such as labor strikes, so long as the movement primes salient social identities.

Conclusion

Most people do not physically participate in protests, marches, or demonstrations against racial injustices like police brutality. However, nearly all are exposed to reporting on these protests. We sought to understand under what conditions support for a movement increased or decreased because of exposure to disruptive protest activity. With the impetus of the BLM movement, police brutality is and will continue to be a leading subject of protests against state, local, and national government in the United States and beyond. While previous work has focused on white and other non-Black responses to protests like those associated with BLM, we hope to add nuance to the conversation and reevaluate the link between disruptive protest activity and negative support for a movement among multiple populations. Our findings suggest that generally, public reactions to violent and nonviolent protests are complicated.

While this study is a good first step, our mixed results highlight the need for more research on how the means of protesting impact political goals. The results of this study complicate the debate about the decorum of Black Lives Matter protesters. Protests about racialized police violence may always be met with backlash from certain segments of the population, regardless of the level of disruption. These findings may help to explain instances of white backlash to passive acts of protest like former NFL player Colin Kapernick’s decision to kneel during the national anthem at football games to honor the victims of police brutality (Niven 2021). But for Black Americans, reactions to various levels of disruption in protest are understandably more nuanced. Negative emotional reactions among Black Americans to specific acts of protest may not signal disapproval of the goals associated with the movement. Given that racial disparities in policing and incarceration are unlikely to change anytime soon, we should expect public reactions to racial justice protests to be a function of both the viewer’s racial identity and the protests’ location on the spectrum of disruption.