hypothesis on police brutality

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• Am J Public Health
• v.110(11); Nov 2020

Impact of Police Violence on Mental Health: A Theoretical Framework

J. DeVylder wrote the original draft of the article. L. Fedina and B. Link contributed substantially to the editing and revision of subsequent drafts of the article. All authors participated in the conceptual development and final editing of the article.

Police violence has increasingly been recognized as a public health concern in the United States, and accumulating evidence has shown police violence exposure to be linked to a broad range of health and mental health outcomes. These associations appear to extend beyond the typical associations between violence and mental health, and to be independent of the effects of co-occurring forms of trauma and violence exposure. However, there is no existing theoretical framework within which we may understand the unique contributions of police violence to mental health and illness.

This article aims to identify potential factors that may distinguish police violence from other forms of violence and trauma exposure, and to explore the possibility that this unique combination of factors distinguishes police violence from related risk exposures. We identify 8 factors that may alter this relationship, including those that increase the likelihood of overall exposure, increase the psychological impact of police violence, and impede the possibility of coping or recovery from such exposures.

On the basis of these factors, we propose a theoretical framework for the further study of police violence from a public mental health perspective.

A new public narrative around the prevalence and effects of police violence has emerged over the past several years in the United States, accompanied recently by a dramatic shift in public opinion following the deaths of George Floyd, Breonna Taylor, and Elijah McClain, and the related national civil uprising and protests. Although Black, Latinx, Native American, and sexual and gender minority communities have long perceived a culture of inequitable treatment, it is only with the widespread adaptation of smartphone technology and real-time dissemination of footage through social media that this has become part of the national consciousness. 1 Media attention has primarily focused on individual incidents of police killings rather than on broader population-level health effects and implications. Although death is certainly the most severe health outcome, it is just as certainly not the most common. The mental health effects of police violence may be less visible yet much more pervasive and, potentially, more impactful when considered across an entire community or population.

In this article, we place the emerging literature on the mental health correlates of police violence within the broader context of research on violence, and explore whether the “police” in “police violence” bestows a specific meaning that extends beyond violence itself—is police violence a form of violence just like any other? By describing potential factors that may distinguish police violence from other forms of violence and trauma exposure—either as factors that are unique to police violence or that vary in degree between police violence and other forms of violence—we propose a theoretical framework for the further study of police violence from a public mental health perspective.

RELEVANCE OF POLICE VIOLENCE TO MENTAL HEALTH

Stress has pervasive effects on one’s psychological well-being, straining one’s sense of role or purpose and affecting concepts of self-esteem and mastery, which contributes in turn to mental health difficulties. 2 Although there is not a single unifying theory linking stressful or traumatic social exposures to mental health symptoms, these factors play a prominent etiological role in leading theories on a broad range of disparate mental health conditions, such as the social signal transduction theory of depression 3 or the social defeat theory of psychosis. 4 Although the often-siloed research of each psychological outcome has led to uniquely labeled theories, these theories all point to a pathway in which trauma spurs biological or psychological changes that manifest over time as psychiatric symptoms, particularly when the trauma is sexually or physically violent. 5 Further, although theoretical work on stressful life events has attempted to provide a broader framework for how stress may translate to psychopathology, focusing particularly on the role of uncontrollable stressful events that affect one’s usual activities, goals, and values, this framework has not been directly applied toward understanding police violence. 6

We therefore explore the construct of police violence as a potential etiological factor for mental health conditions, based on the assumptions that (1) violence and trauma are associated with elevated risk for a broad range of mental health symptoms and (2) the contribution to risk may vary not only by severity of exposure, but also by type of exposure. Specifically, we explore whether police violence possesses a unique pattern of characteristics and mechanisms that distinguish it from other forms of violence exposure in its association with mental health symptoms.

For the purposes of this article, we refer generally to “police violence” and “mental health” because there is not yet sufficient research to confidently link specific subtypes of police violence to specific mental health outcomes. We therefore define police violence as acute events of physical, sexual, psychological, or neglectful violence, following the World Health Organization’s guidelines on defining violence and earlier work on the phenomenology of police violence exposure. 7 Mental health is intended to be inclusive of behaviors and psychological symptoms that would be considered indicators of clinical psychopathology, including but not limited to general psychological distress, posttraumatic stress symptoms, suicidal ideation and behavior, psychosis-like experiences, and depression. These definitions may need to be expanded as this literature develops, as currently it typically focuses on acute violent events (rather than chronic or vicarious exposures) and a psychopathology-oriented view of mental health (rather than a focus on functioning or quality of life), but they are being used here as a reflection of the variables typically employed in the literature at this point in time.

MENTAL HEALTH CORRELATES OF POLICE VIOLENCE

Recent public attention directed toward police violence has spurred an emerging literature on the health significance of police violence exposure, 1,8,9 addressing a long-unheeded call to conceptualize police violence as a public health issue in the United States. 7 Cross-sectional studies have consistently found clinically and statistically significant associations between police violence exposure and a range of mental health outcomes, 10-16 and community-level data have likewise demonstrated higher rates of mental health symptoms in neighborhoods or cities in which police abuse (e.g., “stop and frisk” practices, which are primarily used in neighborhoods predominantly composed of people of color) and killings of unarmed civilians are more common. 17,18 These associations have generally been found to remain statistically significant (and of sufficient effect sizes to support public health significance) even with adjustment for closely related forms of violence exposure, such as interpersonal violence or lifetime abuse exposure. 10,14 For example, exposure to assaultive forms of police violence (i.e., physical or sexual) has been found to be associated with 4- to 11-fold greater odds for a suicide attempt among adults across racial/ethnic groups, even with conservative adjustments. 12,14 Although most of this research has been conducted with adults, recent analyses suggest that this problem extends into adolescence as well. 19 A selective overview of recent work on this topic is provided in Table 1 , and has recently been reviewed elsewhere. 21

TABLE 1—

Selective Overview of Recent Studies of Police Violence and Mental Health: United States

Note. DSM-IV  =  Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (Washington, DC: American Psychiatric Association; 1994); PTSD = posttraumatic stress disorder.

WHY IS POLICE VIOLENCE DIFFERENT?

Overall, accumulated evidence consistently identifies moderate to strong associations between self-reported exposure to police violence and measures of mental health. Additionally, some evidence indicates that these effects operate independently of exposure to other forms of violence. It was this accumulation of evidence that led us to ask whether and to what extent police violence has unique features that lead it to be so impactful for mental health outcomes. Here, we propose 8 factors that may distinguish police violence from other forms of violence, some of which are unique to police violence and others that may vary by degree. Given the complexity of the issue, we see our conceptualization as a step toward a more complete understanding of this important issue, recognizing that it will need further development in the time ahead.

Police Violence Is State Sanctioned

A long tradition in social science theory suggests that the police play a critical role in disciplining the public, not just in terms of offenses and punishments but in the construction and maintenance of an established social order favoring dominant groups. In light of the use of the police in this regard, it follows that exposure to violence emanating from their actions would have distinct and pernicious features. 22,23 Police organizations in the United States are thus authoritative institutions legitimized to apply force—and potentially fatal force—to maintain a particular social and political order. 24 In interactions with civilians, police officers are in positions of relatively greater power because of both the symbolic and state-sanctioned status of their profession, and their immediate legal availability of means (e.g., guns, batons, tasers) to wield force, threat of force, and coercion, at their discretion. This distinguishes police violence from interpersonal forms of violence that are perpetrated by people who are not sanctioned to enact violence, such as caregivers, peers, or intimate partners.

This distinction is made not to downplay the seriousness of other forms of violence—such as child abuse, intimate partner violence, or sexual assault—but to assert that modern-day police violence is embedded in historical state-enforced practices that permitted cruel, unusual, and dehumanizing punishment of individuals deemed to be from threatening or “dangerous classes,” 25 particularly Blacks. Communities of color and lesbian, gay, bisexual, transgender, and queer (LGBTQ) communities have been historically subjected to racially motivated, discriminatory state-sponsored laws (e.g., Jim Crow laws, sodomy laws) enforced by police that permitted harassment, discrimination, and excessive and fatal force against individuals from these communities. As such, the processes and contexts in which police violence has been historically perpetuated are uniquely distinct from the perpetuation of interpersonal forms of violence by others. Furthermore, police violence is sanctioned not only by institutions in the United States but also by the American public, and is intentionally designed to uphold White supremacy. 26 Members of the dominant society thus contribute to police violence and the lack of police accountability.

The Police Are a Pervasive Presence

A core characteristic of many people’s reaction to violence is avoidance of reminders and triggers—especially of the perpetrators themselves. This common and adaptive response to a harmful situation is not available to people who have been exposed to police violence. It is simply not possible to avoid a system that inflicts racially motivated violence while staying within the country, even if one manages to avoid the specific offending officer, and the stress of this police avoidance has been shown to be directly related to severity of depressive symptoms among adult Black men. 20 Much in the same way that police violence by one officer generalizes to fear of all officers even if most officers do not perpetrate violence, intimate partner violence often generalizes to fear of all romantic partners, particularly of a given gender. However, this process may be exacerbated for victims of police violence and is in some ways different from what transpires when people are exposed to other forms of violence. For example, although victims of intimate partner violence and sexual assault who seek help and legal recourse face enormous barriers and challenges, the US justice system can separate victims from perpetrators through legal protection or restraining orders or through incarceration of the perpetrator. In the case of police violence, the presence of law enforcement in the US context is pervasive, and victims have few or no options to seek help and legal recourse, or to entirely avoid police officers in public places.

There Are Limited Options for Recourse

Victims of police violence have little legal recourse or opportunities for seeking help in the criminal justice system. The police have legal sanction to intervene in other crimes of violence (e.g., sexual assault, physical assault), making it much more difficult to prove that the violence was unjustly or excessively delivered. Additionally, the people reviewing disputed cases are often also police officers, and indicted police officers are tried by prosecutors who must otherwise work with police officers. These and other circumstances make contesting the perpetration of violence extremely difficult. Victims of other forms of violence, particularly intimate partner violence, indeed face enormous barriers in seeking help and legal recourse, including stigma in reporting intimate partner violence, poverty and other economic barriers, and other sociocultural and contextual factors. 27 Victims of police violence face many of these same barriers; because they have few if any options for reporting an incident, for legal recourse, or for advocacy services and referrals to mental health treatment, any mental health symptoms they have may worsen over time. 28

Police Culture Deters Internal Accountability

Police violence occurs within a larger, institutional context that is shaped by the organization’s culture. An organizational culture that upholds a “code of silence” surrounding police officers’ abusive behaviors toward civilians allows for the perpetuation of police abuse of power and can prevent police officers, particularly those from lower ranks, from reporting such abuses to their superiors. 29

Given that violence perpetuated through institutions (rather than interpersonal relationships) is supported by an organizational culture condoning harmful behaviors (e.g., harassment, coercion, psychological abuse, physical assault), particularly against those from historically marginalized and disadvantaged communities, experiencing abuse at the hands of police officers who wield such power and authority over civilians may lead to exacerbated mental health consequences. Past research suggests that exposure to sexual assault while serving in the military is associated with psychiatric disorders above and beyond symptoms associated with civilian sexual assault. 30 This suggests that contextual factors related to violence, particularly contexts defined by substantial power and authoritative differentials, may influence associations with mental health symptoms.

Police Violence Alters Deeply Held Beliefs

People feel more secure if they feel safe and protected in their day-to-day activities. Assumptive World Theory proposes that people’s deeply held beliefs about the world and themselves can be shaken by an event that forcefully disconfirms such beliefs. 31 Police violence is particularly likely to provide such disconfirming evidence in that the police represent a societal institution that many, though not all, have come to rely on deeply and implicitly for help when a threat emerges. When police perpetrate violence, this belief is shattered as the police are no longer protectors but rather the central threat that needs to be addressed. Additionally, police violence is normative, rather than an acute or singular event, which has led to the erosion of public trust in the police and favorable views of police seen as protective.

Theories of police legitimacy, which refers to the public’s perceptions and views of police as a legitimate authority that is trustworthy and upholds public safety, propose that legitimacy is in part formed through individual police–citizen interactions. 32 As such, it is plausible that individual and group experiences with police violence influence individual views and beliefs that police are not trustworthy sources of protection and safety. Of course, this sundering of assumptions occurs with other types of violence, such as when a believed-to-be-loving spouse hits a partner or a thought-to-be-protective parent engages in child abuse. However, the police have been described as a “last resort” for people when other remedies have been tried and failed. 33 A spouse might call the police as a last resort when other efforts to stop an abusive partner have failed, or a neighbor might make such a call if polite efforts to address enduring abuse of a child have failed. But to the extent that exposure to police violence intrudes, the “last resort” is gone and one may feel stuck in a brutal and frightening world with no recourse.

Racial and Economic Disparities in Exposure

Because police violence is disproportionately directed toward people of color, many of whom are poor, it can underscore a sense of diminished value within the US racial and class hierarchies. Accordingly, the media narrative around police violence has focused on incidents directed toward Black people, and has at times framed these incidents within the context of the legacy of racism and White supremacy in the United States. Data from the first and second Survey of Police–Public Encounters studies have confirmed that—at least in Baltimore, Maryland; New York City; the District of Columbia; and Philadelphia, Pennsylvania—police violence is more likely to be directed toward people of color, although it is notable that these studies have found Latinx groups to be at approximately the same level of risk as non-Latinx Blacks. 11,14 Although White respondents were also at some risk of exposure to police violence, the racial disparities were significant, even after adjustment for crime involvement and income. Similarly, the prevalence of police-inflicted shootings is approximately 3.5-fold greater among non-Latinx Black than non-Latinx White residents of the United States. 34 Perceptions of racism have been shown to magnify, and perhaps even overshadow, the effects of violent acts. 35 Given that police violence is perceived to be racially motivated in many cases, 34 it is likely that these same effects carry over to many victims of this form of violence.

Notably, there is insufficient prior data to allow a thorough discussion of police violence and mental health among indigenous populations, although the rate of police killings is extremely high among this group. Other potentially high-risk groups likely include people who identify as sexual or gender minorities, people who are homeless, or those who have a severe mental illness diagnosis, among others. Future research should focus on understudied sociodemographic groups that are disproportionately subject to police violence (e.g., indigenous populations, trans individuals), and the conceptual framework presented in this article will require modification as more data become available.

Police Violence Is Stigmatizing

Victims of violent incidents, such as intimate partner violence and community violence, often seek informal support from friends, family, and other social contacts, which has been shown to have a beneficial impact on mental health. 36 However, exposure to police violence carries the potential of inducing harmful stigma. Although stigma may be mitigated in some circumstances in which people distrust the police, a person may nonetheless face judgments from dominant groups who carry the power to discriminate in critical life domains such as educational opportunities, jobs, and housing. This stigma may in turn limit help-seeking behaviors if mental health problems emerge and if there is a perception that treatment providers may not be able to sufficiently understand the circumstances that led to the mental health problems. 37

The police are highly respected in some US communities, sometimes to the point of exaltation, and are supported by a labor union of more than 100 000 workers as well as significant and well-funded public image and advocacy groups such as Blue Lives Matter (which arose as a countermovement to Black Lives Matter and consequently contributes to rather than alleviates concerns of racism and lack of accountability around police violence). As such, there may be substantial stigma around reporting incidents of police violence to family members, friends, and acquaintances, some of whom may have some personal or ideological connection to the police force. Further, when there are major social movements or protests following prominent incidents of police violence, many in the public, particularly those who benefit from the dominant social order that the police help to maintain, take a “blaming the victim” mentality and highlight infractions by the victim that may have justified their injury or death (e.g., the alleged theft of cigarillos by Michael Brown cited as justification for excessive and fatal force). On a broader societal level, protests in Ferguson, Missouri were blamed for a subsequent supposed “war on cops” in which the rate of civilians killing police officers purportedly increased, although there is no actual evidence for any such increase. 38

Police Are Typically Armed

Unlike front-line police officers in some other countries, police officers patrolling neighborhoods in the United States are typically armed, which makes civilians’ interactions with the police potentially more threatening. As a result of several landmark Supreme Court decisions, police officers in the United States have a great deal of legal latitude in determining when to use force, and even fatal force. Additionally, the militarization of police in the United States, largely as a result of “War on Drugs” and “War on Terror” policies, has equipped police departments with firearms and military-grade equipment and expanded their capacity to use force if officers believe their lives or the lives of others are in danger. 39 Thus, the perceived threat of police victimization in civilians’ interactions with police may lead to unique mental health implications for communities most affected by police violence. Further, in addition to the threat of immediate violence through the use of weapons, police encounters also can lead to a more sustained form of exposure to violence and coercion through imprisonment. This threat may be compounded in geographical (e.g., low-income urban areas) and demographic communities (e.g., Black, Latinx, and Native American) with high rates of incarceration.

PROPOSED CONCEPTUAL MODEL

Figure A (available as a supplement to the online version of this article at http://www.ajph.org ) portrays a conceptual model illustrating points at which the influence of police violence on mental health may be different from processes that produce associations between other types of violence exposure and mental health. Specifically, we highlight the 8 potentially influential factors that were described in the previous section, which provides a valuable starting point from which the construct of police violence can be further explored from a public health standpoint. The assumption that police violence is violence like any other would require that the net effect of all of these 8 factors would sum to zero (i.e., have no total effect on mental health). This assumption is highly unlikely, particularly since some of these pathways are now supported by epidemiological evidence (e.g., stress of police avoidance has been recently linked to depressive symptoms). 20 Many (but not all) of these features are present in other forms of violence, although the unique intersection of these features may make police violence a specific type of violence and one worthy of study as a separate construct, similar to the intersection of common and specific elements as determinants of the health impact of other life events. 6 In fact, the literature on stressful life events may provide a useful framework for determining the potential mental health salience of these various features of police violence. Table 2 outlines the primary dimensions of stressful life events based on work by Dohrenwend, 6 a widely used framework for understanding and interpreting the relationship between uncontrollable stressors and mental health outcomes, and it applies these dimensions to our model of police violence.

TABLE 2—

Police Violence Within the Life Events Dimensions Proposed by Dohrenwend

Source. Dohrenwend. 6

To provide a preliminary framework for subsequent work, we have also developed a more complex hypothetical model that illustrates potential mechanisms through which the discussed factors may influence the pathways from police violence to mental health (Figure B, available as a supplement to the online version of this article at http://www.ajph.org ). Although it is speculative because of the limited prior empirical research, we are proposing this model to provide potential conceptual pathways that can be tested in future research. Specifically, 4 of the factors (i.e., access to a weapon, state-sanctioned violence, perceived racial and class biases, and risk of incarceration) are likely to increase the immediate impact of violent incidents and therefore may have the most direct effects on mental health, as they are characteristics of the acute incident itself. Three of the factors (i.e., pervasive presence, lack of recourse, and stigma of reporting police violence) pertain more to the time following an exposure to violence, and therefore may have an effect on mental health by impeding coping and recovery. Finally, police culture, in combination with the proliferation of firearms in the US general population and the American legacy of racism, 24 may have an impact on the overall likelihood or prevalence of police violence. 40 Future studies can confirm whether these pathways provide a feasible explanation for the link between police violence and mental health. It is our intention that this preliminary framework may be modified and updated as research evidence accumulates that may confirm or disconfirm these proposed pathways.

CONCLUSIONS AND NEXT STEPS

In this article, we aimed to determine whether it is reasonable to consider police violence exposure to be a unique risk factor for mental distress, independent and conceptually separable from other forms of violence, or whether such a distinction is unjustified and insufficiently parsimonious. We highlighted several features of police violence that may conceptually distinguish it from other forms of violence. For police violence to be considered effectively similar to other forms of violence exposure, regarding its impact on health, the net effect of these distinguishing features would need to sum to zero, or at least have a clinically insignificant effect. Albeit speculatively, we are confident in stating that this seems highly unlikely. There is now substantial and growing evidence that police violence exposure is associated with a broad range of mental health outcomes, independent of other forms of violence and stress exposure. To test the proposed model, subsequent studies will need to examine the mechanisms underlying this risk and map those mechanisms onto these proposed features of police violence.

ACKNOWLEDGMENTS

Thank you to Leslie Salas-Hernández for contributing to the selective overview of recent studies on the mental health implications of police violence.

CONFLICTS OF INTEREST

The authors have no conflicts of interest to disclose.

HUMAN PARTICIPANT PROTECTION

Institutional approval was not required for this conceptual article, which did not directly involve human participants.

See also Alang, p. 1597 .

Featured Topics

Featured series.

A series of random questions answered by Harvard experts.

Explore the Gazette

Read the latest.

Footnote leads to exploration of start of for-profit prisons in N.Y.

Should NATO step up role in Russia-Ukraine war?

It’s on Facebook, and it’s complicated

Protesters take a knee in front of New York City police officers during a solidarity rally for George Floyd, June 4, 2020.

AP Photo/Frank Franklin II

Solving racial disparities in policing

Colleen Walsh

Harvard Staff Writer

Experts say approach must be comprehensive as roots are embedded in culture

“ Unequal ” is a multipart series highlighting the work of Harvard faculty, staff, students, alumni, and researchers on issues of race and inequality across the U.S. The first part explores the experience of people of color with the criminal justice legal system in America.

It seems there’s no end to them. They are the recent videos and reports of Black and brown people beaten or killed by law enforcement officers, and they have fueled a national outcry over the disproportionate use of excessive, and often lethal, force against people of color, and galvanized demands for police reform.

This is not the first time in recent decades that high-profile police violence — from the 1991 beating of Rodney King to the fatal shooting of Michael Brown in 2014 — ignited calls for change. But this time appears different. The police killings of Breonna Taylor in March, George Floyd in May, and a string of others triggered historic, widespread marches and rallies across the nation, from small towns to major cities, drawing protesters of unprecedented diversity in race, gender, and age.

According to historians and other scholars, the problem is embedded in the story of the nation and its culture. Rooted in slavery, racial disparities in policing and police violence, they say, are sustained by systemic exclusion and discrimination, and fueled by implicit and explicit bias. Any solution clearly will require myriad new approaches to law enforcement, courts, and community involvement, and comprehensive social change driven from the bottom up and the top down.

While police reform has become a major focus, the current moment of national reckoning has widened the lens on systemic racism for many Americans. The range of issues, though less familiar to some, is well known to scholars and activists. Across Harvard, for instance, faculty members have long explored the ways inequality permeates every aspect of American life. Their research and scholarship sits at the heart of a new Gazette series starting today aimed at finding ways forward in the areas of democracy; wealth and opportunity; environment and health; and education. It begins with this first on policing.

Harvard Kennedy School Professor Khalil Gibran Muhammad traces the history of policing in America to “slave patrols” in the antebellum South, in which white citizens were expected to help supervise the movements of enslaved Black people.

Photo by Martha Stewart

The history of racialized policing

Like many scholars, Khalil Gibran Muhammad , professor of history, race, and public policy at the Harvard Kennedy School , traces the history of policing in America to “slave patrols” in the antebellum South, in which white citizens were expected to help supervise the movements of enslaved Black people. This legacy, he believes, can still be seen in policing today. “The surveillance, the deputization essentially of all white men to be police officers or, in this case, slave patrollers, and then to dispense corporal punishment on the scene are all baked in from the very beginning,” he  told NPR  last year.

Slave patrols, and the slave codes they enforced, ended after the Civil War and the passage of the 13th amendment, which formally ended slavery “except as a punishment for crime.” But Muhammad notes that former Confederate states quickly used that exception to justify new restrictions. Known as the Black codes, the various rules limited the kinds of jobs African Americans could hold, their rights to buy and own property, and even their movements.

“The genius of the former Confederate states was to say, ‘Oh, well, if all we need to do is make them criminals and they can be put back in slavery, well, then that’s what we’ll do.’ And that’s exactly what the Black codes set out to do. The Black codes, for all intents and purposes, criminalized every form of African American freedom and mobility, political power, economic power, except the one thing it didn’t criminalize was the right to work for a white man on a white man’s terms.” In particular, he said the Ku Klux Klan “took about the business of terrorizing, policing, surveilling, and controlling Black people. … The Klan totally dominates the machinery of justice in the South.”

When, during what became known as the Great Migration, millions of African Americans fled the still largely agrarian South for opportunities in the thriving manufacturing centers of the North, they discovered that metropolitan police departments tended to enforce the law along racial and ethnic lines, with newcomers overseen by those who came before. “There was an early emphasis on people whose status was just a tiny notch better than the folks whom they were focused on policing,” Muhammad said. “And so the Anglo-Saxons are policing the Irish or the Germans are policing the Irish. The Irish are policing the Poles.” And then arrived a wave of Black Southerners looking for a better life.

In his groundbreaking work, “ The Condemnation of Blackness: Race, Crime, and the Making of Modern Urban America ,” Muhammad argues that an essential turning point came in the early 1900s amid efforts to professionalize police forces across the nation, in part by using crime statistics to guide law enforcement efforts. For the first time, Americans with European roots were grouped into one broad category, white, and set apart from the other category, Black.

Citing Muhammad’s research, Harvard historian Jill Lepore  has summarized the consequences this way : “Police patrolled Black neighborhoods and arrested Black people disproportionately; prosecutors indicted Black people disproportionately; juries found Black people guilty disproportionately; judges gave Black people disproportionately long sentences; and, then, after all this, social scientists, observing the number of Black people in jail, decided that, as a matter of biology, Black people were disproportionately inclined to criminality.”

“History shows that crime data was never objective in any meaningful sense,” Muhammad wrote. Instead, crime statistics were “weaponized” to justify racial profiling, police brutality, and ever more policing of Black people.

This phenomenon, he believes, has continued well into this century and is exemplified by William J. Bratton, one of the most famous police leaders in recent America history. Known as “America’s Top Cop,” Bratton led police departments in his native Boston, Los Angeles, and twice in New York, finally retiring in 2016.

Bratton rejected notions that crime was a result of social and economic forces, such as poverty, unemployment, police practices, and racism. Instead, he said in a 2017 speech, “It is about behavior.” Through most of his career, he was a proponent of statistically-based “predictive” policing — essentially placing forces in areas where crime numbers were highest, focused on the groups found there.

Bratton argued that the technology eliminated the problem of prejudice in policing, without ever questioning potential bias in the data or algorithms themselves — a significant issue given the fact that Black Americans are arrested and convicted of crimes at disproportionately higher rates than whites. This approach has led to widely discredited practices such as racial profiling and “stop-and-frisk.” And, Muhammad notes, “There is no research consensus on whether or how much violence dropped in cities due to policing.”

Gathering numbers

In 2015 The Washington Post began tracking every fatal shooting by an on-duty officer, using news stories, social media posts, and police reports in the wake of the fatal police shooting of Brown, a Black teenager in Ferguson, Mo. According to the newspaper, Black Americans are killed by police at twice the rate of white Americans, and Hispanic Americans are also killed by police at a disproportionate rate.

Such efforts have proved useful for researchers such as economist Rajiv Sethi .

A Joy Foundation Fellow at the Harvard  Radcliffe Institute , Sethi is investigating the use of lethal force by law enforcement officers, a difficult task given that data from such encounters is largely unavailable from police departments. Instead, Sethi and his team of researchers have turned to information collected by websites and news organizations including The Washington Post and The Guardian, merged with data from other sources such as the Bureau of Justice Statistics, the Census, and the Centers for Disease Control and Prevention.

A Joy Foundation Fellow at the Harvard Radcliffe Institute, Rajiv Sethi is investigating the use of lethal force by law enforcement officers,

Courtesy photo

They have found that exposure to deadly force is highest in the Mountain West and Pacific regions relative to the mid-Atlantic and northeastern states, and that racial disparities in relation to deadly force are even greater than the national numbers imply. “In the country as a whole, you’re about two to three times more likely to face deadly force if you’re Black than if you are white” said Sethi. “But if you look at individual cities separately, disparities in exposure are much higher.”

Examining the characteristics associated with police departments that experience high numbers of lethal encounters is one way to better understand and address racial disparities in policing and the use of violence, Sethi said, but it’s a massive undertaking given the decentralized nature of policing in America. There are roughly 18,000 police departments in the country, and more than 3,000 sheriff’s offices, each with its own approaches to training and selection.

“They behave in very different ways, and what we’re finding in our current research is that they are very different in the degree to which they use deadly force,” said Sethi. To make real change, “You really need to focus on the agency level where organizational culture lies, where selection and training protocols have an effect, and where leadership can make a difference.”

Sethi pointed to the example of Camden, N.J., which disbanded and replaced its police force in 2013, initially in response to a budget crisis, but eventually resulting in an effort to fundamentally change the way the police engaged with the community. While there have been improvements, including greater witness cooperation, lower crime, and fewer abuse complaints, the Camden case doesn’t fit any particular narrative, said Sethi, noting that the number of officers actually increased as part of the reform. While the city is still faced with its share of problems, Sethi called its efforts to rethink policing “important models from which we can learn.”

Fighting vs. preventing crime

For many analysts, the real problem with policing in America is the fact that there is simply too much of it. “We’ve seen since the mid-1970s a dramatic increase in expenditures that are associated with expanding the criminal legal system, including personnel and the tasks we ask police to do,” said Sandra Susan Smith , Daniel and Florence Guggenheim Professor of Criminal Justice at HKS, and the Carol K. Pforzheimer Professor at the Radcliffe Institute. “And at the same time we see dramatic declines in resources devoted to social welfare programs.”

“You can have all the armored personnel carriers you want in Ferguson, but public safety is more likely to come from redressing environmental pollution, poor education, and unfair work,” said Brandon Terry, assistant professor of African and African American Studies and social studies.

Kris Snibble/Harvard file photo

Smith’s comment highlights a key argument embraced by many activists and experts calling for dramatic police reform: diverting resources from the police to better support community services including health care, housing, and education, and stronger economic and job opportunities. They argue that broader support for such measures will decrease the need for policing, and in turn reduce violent confrontations, particularly in over-policed, economically disadvantaged communities, and communities of color.

For Brandon Terry , that tension took the form of an ice container during his Baltimore high school chemistry final. The frozen cubes were placed in the middle of the classroom to help keep the students cool as a heat wave sent temperatures soaring. “That was their solution to the building’s lack of air conditioning,” said Terry, a Harvard assistant professor of African and African American Studies and social studies. “Just grab an ice cube.”

Terry’s story is the kind many researchers cite to show the negative impact of underinvesting in children who will make up the future population, and instead devoting resources toward policing tactics that embrace armored vehicles, automatic weapons, and spy planes. Terry’s is also the kind of tale promoted by activists eager to defund the police, a movement begun in the late 1960s that has again gained momentum as the death toll from violent encounters mounts. A scholar of Martin Luther King Jr., Terry said the Civil Rights leader’s views on the Vietnam War are echoed in the calls of activists today who are pressing to redistribute police resources.

“King thought that the idea of spending many orders of magnitude more for an unjust war than we did for the abolition of poverty and the abolition of ghettoization was a moral travesty, and it reflected a kind of sickness at the core of our society,” said Terry. “And part of what the defund model is based upon is a similar moral criticism, that these budgets reflect priorities that we have, and our priorities are broken.”

Terry also thinks the policing debate needs to be expanded to embrace a fuller understanding of what it means for people to feel truly safe in their communities. He highlights the work of sociologist Chris Muller and Harvard’s Robert Sampson, who have studied racial disparities in exposures to lead and the connections between a child’s early exposure to the toxic metal and antisocial behavior. Various studies have shown that lead exposure in children can contribute to cognitive impairment and behavioral problems, including heightened aggression.

“You can have all the armored personnel carriers you want in Ferguson,” said Terry, “but public safety is more likely to come from redressing environmental pollution, poor education, and unfair work.”

Policing and criminal justice system

Alexandra Natapoff , Lee S. Kreindler Professor of Law, sees policing as inexorably linked to the country’s criminal justice system and its long ties to racism.

“Policing does not stand alone or apart from how we charge people with crimes, or how we convict them, or how we treat them once they’ve been convicted,” she said. “That entire bundle of official practices is a central part of how we govern, and in particular, how we have historically governed Black people and other people of color, and economically and socially disadvantaged populations.”

Unpacking such a complicated issue requires voices from a variety of different backgrounds, experiences, and fields of expertise who can shine light on the problem and possible solutions, said Natapoff, who co-founded a new lecture series with HLS Professor Andrew Crespo titled “ Policing in America .”

In recent weeks the pair have hosted Zoom discussions on topics ranging from qualified immunity to the Black Lives Matter movement to police unions to the broad contours of the American penal system. The series reflects the important work being done around the country, said Natapoff, and offers people the chance to further “engage in dialogue over these over these rich, complicated, controversial issues around race and policing, and governance and democracy.”

Courts and mass incarceration

Much of Natapoff’s recent work emphasizes the hidden dangers of the nation’s misdemeanor system. In her book “ Punishment Without Crime: How Our Massive Misdemeanor System Traps the Innocent and Makes America More Unequal ,” Natapoff shows how the practice of stopping, arresting, and charging people with low-level offenses often sends them down a devastating path.

“This is how most people encounter the criminal apparatus, and it’s the first step of mass incarceration, the initial net that sweeps people of color disproportionately into the criminal system,” said Natapoff. “It is also the locus that overexposes Black people to police violence. The implications of this enormous net of police and prosecutorial authority around minor conduct is central to understanding many of the worst dysfunctions of our criminal system.”

One consequence is that Black and brown people are incarcerated at much higher rates than white people. America has approximately 2.3 million people in federal, state, and local prisons and jails, according to a 2020 report from the nonprofit the Prison Policy Initiative. According to a 2018 report from the Sentencing Project, Black men are 5.9 times as likely to be incarcerated as white men and Hispanic men are 3.1 times as likely.

Reducing mass incarceration requires shrinking the misdemeanor net “along all of its axes” said Natapoff, who supports a range of reforms including training police officers to both confront and arrest people less for low-level offenses, and the policies of forward-thinking prosecutors willing to “charge fewer of those offenses when police do make arrests.”

She praises the efforts of Suffolk County District Attorney Rachael Rollins in Massachusetts and George Gascón, the district attorney in Los Angeles County, Calif., who have pledged to stop prosecuting a range of misdemeanor crimes such as resisting arrest, loitering, trespassing, and drug possession. “If cities and towns across the country committed to that kind of reform, that would be a profoundly meaningful change,” said Natapoff, “and it would be a big step toward shrinking our entire criminal apparatus.”

Retired U.S. Judge Nancy Gertner cites the need to reform federal sentencing guidelines, arguing that all too often they have been proven to be biased and to result in packing the nation’s jails and prisons.

Stephanie Mitchell/Harvard file photo

Sentencing reform

Another contributing factor in mass incarceration is sentencing disparities.

A recent Harvard Law School study found that, as is true nationally, people of color are “drastically overrepresented in Massachusetts state prisons.” But the report also noted that Black and Latinx people were less likely to have their cases resolved through pretrial probation ­— a way to dismiss charges if the accused meet certain conditions — and receive much longer sentences than their white counterparts.

Retired U.S. Judge Nancy Gertner also notes the need to reform federal sentencing guidelines, arguing that all too often they have been proven to be biased and to result in packing the nation’s jails and prisons. She points to the way the 1994 Crime Bill (legislation sponsored by then-Sen. Joe Biden of Delaware) ushered in much harsher drug penalties for crack than for powder cocaine. This tied the hands of judges issuing sentences and disproportionately punished people of color in the process. “The disparity in the treatment of crack and cocaine really was backed up by anecdote and stereotype, not by data,” said Gertner, a lecturer at HLS. “There was no data suggesting that crack was infinitely more dangerous than cocaine. It was the young Black predator narrative.”

The First Step Act, a bipartisan prison reform bill aimed at reducing racial disparities in drug sentencing and signed into law by President Donald Trump in 2018, is just what its name implies, said Gertner.

“It reduces sentences to the merely inhumane rather than the grotesque. We still throw people in jail more than anybody else. We still resort to imprisonment, rather than thinking of other alternatives. We still resort to punishment rather than other models. None of that has really changed. I don’t deny the significance of somebody getting out of prison a year or two early, but no one should think that that’s reform.”

 Not just bad apples

Reform has long been a goal for federal leaders. Many heralded Obama-era changes aimed at eliminating racial disparities in policing and outlined in the report by The President’s Task Force on 21st Century policing. But HKS’s Smith saw them as largely symbolic. “It’s a nod to reform. But most of the reforms that are implemented in this country tend to be reforms that nibble around the edges and don’t really make much of a difference.”

Efforts such as diversifying police forces and implicit bias training do little to change behaviors and reduce violent conduct against people of color, said Smith, who cites studies suggesting a majority of Americans hold negative biases against Black and brown people, and that unconscious prejudices and stereotypes are difficult to erase.

“Experiments show that you can, in the context of a day, get people to think about race differently, and maybe even behave differently. But if you follow up, say, a week, or two weeks later, those effects are gone. We don’t know how to produce effects that are long-lasting. We invest huge amounts to implement such police reforms, but most often there’s no empirical evidence to support their efficacy.”

Even the early studies around the effectiveness of body cameras suggest the devices do little to change “officers’ patterns of behavior,” said Smith, though she cautions that researchers are still in the early stages of collecting and analyzing the data.

And though police body cameras have caught officers in unjust violence, much of the general public views the problem as anomalous.

“Despite what many people in low-income communities of color think about police officers, the broader society has a lot of respect for police and thinks if you just get rid of the bad apples, everything will be fine,” Smith added. “The problem, of course, is this is not just an issue of bad apples.”

Efforts such as diversifying police forces and implicit bias training do little to change behaviors and reduce violent conduct against people of color, said Sandra Susan Smith, a professor of criminal justice Harvard Kennedy School.

Community-based ways forward

Still Smith sees reason for hope and possible ways forward involving a range of community-based approaches. As part of the effort to explore meaningful change, Smith, along with Christopher Winship , Diker-Tishman Professor of Sociology at Harvard University and a member of the senior faculty at HKS, have organized “ Reimagining Community Safety: A Program in Criminal Justice Speaker Series ” to better understand the perspectives of practitioners, policymakers, community leaders, activists, and academics engaged in public safety reform.

Some community-based safety models have yielded important results. Smith singles out the Crisis Assistance Helping Out on the Streets program (known as CAHOOTS ) in Eugene, Ore., which supplements police with a community-based public safety program. When callers dial 911 they are often diverted to teams of workers trained in crisis resolution, mental health, and emergency medicine, who are better equipped to handle non-life-threatening situations. The numbers support her case. In 2017 the program received 25,000 calls, only 250 of which required police assistance. Training similar teams of specialists who don’t carry weapons to handle all traffic stops could go a long way toward ending violent police encounters, she said.

“Imagine you have those kinds of services in play,” said Smith, paired with community-based anti-violence program such as Cure Violence , which aims to stop violence in targeted neighborhoods by using approaches health experts take to control disease, such as identifying and treating individuals and changing social norms. Together, she said, these programs “could make a huge difference.”

At Harvard Law School, students have been  studying how an alternate 911-response team  might function in Boston. “We were trying to move from thinking about a 911-response system as an opportunity to intervene in an acute moment, to thinking about what it would look like to have a system that is trying to help reweave some of the threads of community, a system that is more focused on healing than just on stopping harm” said HLS Professor Rachel Viscomi, who directs the Harvard Negotiation and Mediation Clinical Program and oversaw the research.

The forthcoming report, compiled by two students in the HLS clinic, Billy Roberts and Anna Vande Velde, will offer officials a range of ideas for how to think about community safety that builds on existing efforts in Boston and other cities, said Viscomi.

But Smith, like others, knows community-based interventions are only part of the solution. She applauds the Justice Department’s investigation into the Ferguson Police Department after the shooting of Brown. The 102-page report shed light on the department’s discriminatory policing practices, including the ways police disproportionately targeted Black residents for tickets and fines to help balance the city’s budget. To fix such entrenched problems, state governments need to rethink their spending priorities and tax systems so they can provide cities and towns the financial support they need to remain debt-free, said Smith.

Rethinking the 911-response system to being one that is “more focused on healing than just on stopping harm” is part of the student-led research under the direction of Law School Professor Rachel Viscomi, who heads up the Harvard Negotiation and Mediation Clinical Program.

Jon Chase/Harvard file photo

“Part of the solution has to be a discussion about how government is funded and how a city like Ferguson got to a place where government had so few resources that they resorted to extortion of their residents, in particular residents of color, in order to make ends meet,” she said. “We’ve learned since that Ferguson is hardly the only municipality that has struggled with funding issues and sought to address them through the oppression and repression of their politically, socially, and economically marginalized Black and Latino residents.”

Police contracts, she said, also need to be reexamined. The daughter of a “union man,” Smith said she firmly supports officers’ rights to union representation to secure fair wages, health care, and safe working conditions. But the power unions hold to structure police contracts in ways that protect officers from being disciplined for “illegal and unethical behavior” needs to be challenged, she said.

“I think it’s incredibly important for individuals to be held accountable and for those institutions in which they are embedded to hold them to account. But we routinely find that union contracts buffer individual officers from having to be accountable. We see this at the level of the Supreme Court as well, whose rulings around qualified immunity have protected law enforcement from civil suits. That needs to change.”

Other Harvard experts agree. In an opinion piece in The Boston Globe last June, Tomiko Brown-Nagin , dean of the Harvard Radcliffe Institute and the Daniel P.S. Paul Professor of Constitutional Law at HLS, pointed out the Court’s “expansive interpretation of qualified immunity” and called for reform that would “promote accountability.”

“This nation is devoted to freedom, to combating racial discrimination, and to making government accountable to the people,” wrote Brown-Nagin. “Legislators today, like those who passed landmark Civil Rights legislation more than 50 years ago, must take a stand for equal justice under law. Shielding police misconduct offends our fundamental values and cannot be tolerated.”

Share this article

You might like.

Historian traces 19th-century murder case that brought together historical figures, helped shape American thinking on race, violence, incarceration

National security analysts outline stakes ahead of July summit

‘Spermworld’ documentary examines motivations of prospective parents, volunteer donors who connect through private group page 

Six receive honorary degrees

Harvard recognizes educator, conductor, theoretical physicist, advocate for elderly, writer, and Nobel laureate

Everything counts!

New study finds step-count and time are equally valid in reducing health risks

Bridging social distance, embracing uncertainty, fighting for equity

Student Commencement speeches to tap into themes faced by Class of 2024

June 4, 2020

A Civil Rights Expert Explains the Social Science of Police Racism

Columbia University attorney Alexis J. Hoag discusses the history of how we got to this point and the ways that researchers can help reduce bias against black Americans throughout the legal system

By Lydia Denworth

Protester holds sign during a demonstration in honor of George Floyd on June 2, 2020, in Marin City, Calif.

Justin Sullivan Getty Images

In a now infamous event captured on video, on May 25, 2020, George Floyd, a 46-year-old Black man, was killed by a Minneapolis police officer outside of a corner store. Derek Chauvin kneeled on Floyd’s neck for nine minutes and 29 seconds while two other officers helped to hold him down and a third stood guard nearby. Nearly a year later, in April 2021, a jury convicted Chauvin of second-degree murder, third-degree murder and second-degree manslaughter. He could face decades in prison (sentencing was expected on June 25). In a highly unusual development, other police officers, including the Minneapolis chief of police, testified against Chauvin.

The three other officers involved, Thomas Lane, J. Alexander Kueng and Tou Thao, were indicted on a range of state and federal charges, including violating Floyd’s constitutional rights, failing to intervene to stop Chauvin, and aiding and abetting second-degree murder and second-degree manslaughter. Their trial is scheduled for March 2022.

The 2014 shooting death of Black teenager Michael Brown in Ferguson, Mo., sparked a renewed emphasis on racism and police brutality in the U.S.’s political and cultural conversation. In the past few years many names have been added to the list of Black people killed by police. Despite some efforts to acknowledge and grapple with systemic racism in American institutions, anger and distrust between law enforcement and Black Americans have remained high. But Floyd’s death sparked a new level of outrage. Protests erupted in hundreds of cities around the U.S. in the summer of 2020. Most demonstrations were peaceful. But some turned violent, with police using force against protesters and a small percentage of people setting fire to police cars, looting stores, and defacing or damaging buildings. By July the demonstrations were thought to be the largest protest movement in American history, with some 15 million to 26 million people estimated to have taken part.

In addition to the criminal charges against the officers, Floyd’s death has prompted U.S. Justice Department investigations into the practices of the Minneapolis Police Department. And Democrats in Congress are hoping to pass criminal justice reform legislation named for Floyd. Both reflect the interests of the new administration since Joe Biden took office in January 2021.

In June 2020, at the height of the protests, Scientific American  spoke with civil rights attorney Alexis J. Hoag. Hoag is the inaugural practitioner in residence at the Eric H. Holder, Jr., Initiative for Civil and Political Rights at Columbia University. She works with both undergraduates and law school students at Columbia to introduce them to civil rights fieldwork (which she describes as “real issues, real clients, real cases”). Hoag was previously a senior counsel at the NAACP Legal Defense and Educational Fund. Scientific American asked her to share her perspective on the history that has brought the U.S. to a breaking point—and her ideas for how to make substantive improvements in how law enforcement and courts treat Black people in the country.

Why are we seeing this level of protest now?

I think it’s a combination of things. COVID-19 [has had a] disproportionate impact on Black people because of long-standing structural inequalities. Black people are more likely to live in hypersegregated low-income areas that are underresourced. And Black people are more prone to the very preexisting conditions that make people vulnerable to COVID-19 because of structural inequality and lack of access to health care. We’ve all been cooped up for 10 to 11 weeks. Forty million people [in the U.S.] are unemployed. And there was something egregious about the video that circulated of George Floyd being executed for the suspicion of tendering a counterfeit $20 bill. And I want to stress “suspicion” because we still don’t know. That became a death sentence for him.

The violence that has been rendered against Black bodies has gone on for centuries. Now it’s out there for everyone to see. And the response, which is hopeful and heartening to me, is that people—not just Black Americans—in this country are really disturbed and appropriately so.

What are the important historical factors that have led up to this point?

I lean so heavily on the unique history of this country and the fact that we enslaved people, Black people. To hold people in bondage as property, you had to look at them as less than human. You see that continuing to happen today in [what] I refer to as the criminal legal system, not the justice system, because it is not just. We are not there yet. As an appellate attorney, I read a lot of transcripts of trials. And the level of dehumanization that prosecutors use to refer to Black criminal defendants is striking. It’s the verbiage used, that the defendant was “circling” and “hunting” the victim. What hunts and circles? Animals. When you can dehumanize an individual, of course, you can put the person away for a long time, you can sentence him or her to death. And of course, you can put your knee on somebody’s neck for nine minutes because you see them as less than human. It’s a combination of the dehumanization of Black people with the presumption of dangerousness and criminality.

Is racism getting worse? Or has the ubiquity of cell phones and video recordings simply made us more aware of it?

These issues are getting amplified; they’re getting recorded. I think back to the early 1990s and Rodney King’s videotaped beating. That really galvanized people around this issue—an issue that many Black Americans were intimately aware of already—and put it out there for the world to see. Then the response after those officers were acquitted was public demonstrations in 1992 in Los Angeles. I think people would not have been as engaged if we didn’t have that image. Now we walk around with [cameras] in our pockets.

How does the seeming increase in white nationalism fit in?

I don’t know that I would call it an increase. White nationalists, known earlier as white supremacists, first rallied [more than] 150 years ago to violently limit the freedom of newly emancipated Black Americans. Despite federal legislation extending the benefits of citizenship to Black people, white supremacists passed state laws codifying inequality and used violence and intimidation to curtail any Black exercise of freedom. What’s happening now [in June 2020] is that we have [a presidential] administration that welcomes and encourages white nationalist views and activities.

Have events in Ferguson and other cities, and the Black Lives Matter movement as a whole, had any effect on policing?

Ferguson was a massive wake-up call. There was a brief glimmer of hope. There was a mechanism in place: the Law Enforcement Misconduct [Statute]. It [is] a federal law the Department of Justice could rely on to investigate Ferguson, to investigate police misconduct in Baltimore [where Freddie Gray, a 25-year-old Black man, died while being transported in a police vehicle in what was ruled to be a homicide]. That law was grossly underutilized by Attorney General William Barr. Who the administration is and who the chief law-enforcement officer of this country is—the attorney general of the U.S.—makes a difference. We’ve seen a massive rollback in the responsiveness of the [Trump] administration [in taking] a hard look at injustice and at rampant police misconduct.

The other step back that the country has taken is to characterize officers involved in misconduct as “a few bad apples.” I think we all need to admit that it’s not a few bad apples; it’s a rotten apple tree. The history of policing in the South [was driven in part by] slave patrols that were monitoring the movement of Black bodies. And in the North, law enforcement was privately funded [and often involved protecting property and goods]. The police got started targeting poor people and Black people.

What would you like to see happen now?

I think there needs to be a really hard conversation nationally and within law enforcement. To use force, police officers have to reasonably believe that their lives are in danger. What is it about Black skin that makes law enforcement feel threatened for their lives? In addition, there are legal mechanisms that need to be examined. “Qualified immunity” as a defense to police misconduct was judicially created in 1982. It shields government officials from being sued for discretionary actions that are performed within their official capacities unless the action violates clearly established federal law. Somebody who is suing an officer for tasing someone while they’re handcuffed has to find a case from the U.S. Supreme Court or the highest court of appeals in their jurisdiction that says that exact act—being handcuffed and tased—is unconstitutional. This is a massive hurdle for a plaintiff.

What are social scientists and researchers doing to help?

Data are currency. We can create a national database of officer misconduct. You have officers such as Derek Chauvin, who had 18 complaints against him and [was] still allowed to operate within the [Minneapolis Police] Department.

The data collection that happens within police departments enabled experts in the stop-and-frisk litigation [against] the [New York City Police Department] to shine a spotlight on gross disparities: the rate of stops and searches of Black and brown men and boys [coupled with] the low rate of actually acquiring contraband. They found that the rate of securing contraband from white individuals who had been stopped and frisked was so much higher because the police were actually using discretion.

There’s powerful data collection that happens in our criminal courts. Studies show that, all factors being equal, judges are rendering longer and harsher sentences for Black defendants. These judges are setting higher bail. You can isolate all these other factors, but race is the difference. That’s very powerful—to be able to document and publish those findings.

There has also been some really good social science research on implicit bias and the way that it operates. We could all take [implicit association tests] on our computers. You could do a training with your employees. To start with, there is this recognition, this acknowledgment, that we all have implicit bias.

And how do we use that information and not just let people off the hook?

Let’s talk about it. Social science research shows that when there’s recognition that we harbor implicit bias, that awareness can help mitigate [such] bias impacting our daily interactions and decisions.

What about people’s decision to protest during the pandemic? Are you worried that protesters will get sick and spread COVID-19?

Of course. I worry that there will be a second wave of infections. But I think that also speaks to how pressing the issue is and how strongly people feel about it—that they are risking their lives to bring attention to the rampant and lethal mistreatment of Black and brown bodies at the hands of law enforcement.

Numbers, Facts and Trends Shaping Your World

Read our research on:

Full Topic List

Regions & Countries

• Publications
• Our Methods
• Short Reads
• Tools & Resources

Read Our Research On:

10 things we know about race and policing in the U.S.

Days of protests across the United States in the wake of George Floyd’s death in the custody of Minneapolis police have brought new attention to questions about police officers’ attitudes toward black Americans, protesters and others. The public’s views of the police, in turn, are also in the spotlight. Here’s a roundup of Pew Research Center survey findings from the past few years about the intersection of race and law enforcement.

How we did this

Most of the findings in this post were drawn from two previous Pew Research Center reports: one on police officers and policing issues published in January 2017, and one on the state of race relations in the United States published in April 2019. We also drew from a September 2016 report on how black and white Americans view police in their communities. (The questions asked for these reports, as well as their responses, can be found in the reports’ accompanying “topline” file or files.)

The 2017 police report was based on two surveys. One was of 7,917 law enforcement officers from 54 police and sheriff’s departments across the U.S., designed and weighted to represent the population of officers who work in agencies that employ at least 100 full-time sworn law enforcement officers with general arrest powers, and conducted between May and August 2016. The other survey, of the general public, was conducted via the Center’s American Trends Panel (ATP) in August and September 2016 among 4,538 respondents. (The 2016 report on how blacks and whites view police in their communities also was based on that survey.) More information on methodology is available here .

The 2019 race report was based on a survey conducted in January and February 2019. A total of 6,637 people responded, out of 9,402 who were sampled, for a response rate of 71%. The respondents included 5,599 from the ATP and oversamples of 530 non-Hispanic black and 508 Hispanic respondents sampled from Ipsos’ KnowledgePanel. More information on methodology is available here .

Majorities of both black and white Americans say black people are treated less fairly than whites in dealing with the police and by the criminal justice system as a whole. In a 2019 Center survey , 84% of black adults said that, in dealing with police, blacks are generally treated less fairly than whites; 63% of whites said the same. Similarly, 87% of blacks and 61% of whites said the U.S. criminal justice system treats black people less fairly.

Black adults are about five times as likely as whites to say they’ve been unfairly stopped by police because of their race or ethnicity (44% vs. 9%), according to the same survey. Black men are especially likely to say this : 59% say they’ve been unfairly stopped, versus 31% of black women.

White Democrats and white Republicans have vastly different views of how black people are treated by police and the wider justice system. Overwhelming majorities of white Democrats say black people are treated less fairly than whites by the police (88%) and the criminal justice system (86%), according to the 2019 poll. About four-in-ten white Republicans agree (43% and 39%, respectively).

Nearly two-thirds of black adults (65%) say they’ve been in situations where people acted as if they were suspicious of them because of their race or ethnicity, while only a quarter of white adults say that’s happened to them. Roughly a third of both Asian and Hispanic adults (34% and 37%, respectively) say they’ve been in such situations, the 2019 survey found.

Black Americans are far less likely than whites to give police high marks for the way they do their jobs . In a 2016 survey, only about a third of black adults said that police in their community did an “excellent” or “good” job in using the right amount of force (33%, compared with 75% of whites), treating racial and ethnic groups equally (35% vs. 75%), and holding officers accountable for misconduct (31% vs. 70%).

In the past, police officers and the general public have tended to view fatal encounters between black people and police very differently. In a 2016 survey  of nearly 8,000 policemen and women from departments with at least 100 officers, two-thirds said most such encounters are isolated incidents and not signs of broader problems between police and the black community. In a companion survey of more than 4,500 U.S. adults, 60% of the public called such incidents signs of broader problems between police and black people. But the views given by police themselves were sharply differentiated by race: A majority of black officers (57%) said that such incidents were evidence of a broader problem, but only 27% of white officers and 26% of Hispanic officers said so.

Around two-thirds of police officers (68%) said in 2016 that the demonstrations over the deaths of black people during encounters with law enforcement were motivated to a great extent by anti-police bias; only 10% said (in a separate question) that protesters were primarily motivated by a genuine desire to hold police accountable for their actions. Here as elsewhere, police officers’ views differed by race: Only about a quarter of white officers (27%) but around six-in-ten of their black colleagues (57%) said such protests were motivated at least to some extent by a genuine desire to hold police accountable.

White police officers and their black colleagues have starkly different views on fundamental questions regarding the situation of blacks in American society, the 2016 survey found. For example, nearly all white officers (92%) – but only 29% of their black colleagues – said the U.S. had made the changes needed to assure equal rights for blacks.

A majority of officers said in 2016 that relations between the police in their department and black people in the community they serve were “excellent” (8%) or “good” (47%). However, far higher shares saw excellent or good community relations with whites (91%), Asians (88%) and Hispanics (70%). About a quarter of police officers (26%) said relations between police and black people in their community were “only fair,” while nearly one-in-five (18%) said they were “poor” – with black officers far more likely than others to say so. (These percentages are based on only those officers who offered a rating.)

An overwhelming majority of police officers (86%) said in 2016 that high-profile fatal encounters between black people and police officers had made their jobs harder . Sizable majorities also said such incidents had made their colleagues more worried about safety (93%), heightened tensions between police and blacks (75%), and left many officers reluctant to use force when appropriate (76%) or to question people who seemed suspicious (72%).

• Criminal Justice
• Discrimination & Prejudice
• Racial Bias & Discrimination

Drew DeSilver is a senior writer at Pew Research Center .

Michael Lipka is an associate director focusing on news and information research at Pew Research Center .

Dalia Fahmy is a senior writer/editor focusing on religion at Pew Research Center .

What the data says about crime in the U.S.

Fewer than 1% of federal criminal defendants were acquitted in 2022, before release of video showing tyre nichols’ beating, public views of police conduct had improved modestly, black americans differ from other u.s. adults over whether individual or structural racism is a bigger problem, violent crime is a key midterm voting issue, but what does the data say, most popular.

1615 L St. NW, Suite 800 Washington, DC 20036 USA (+1) 202-419-4300 | Main (+1) 202-857-8562 | Fax (+1) 202-419-4372 |  Media Inquiries

Research Topics

• Email Newsletters

ABOUT PEW RESEARCH CENTER  Pew Research Center is a nonpartisan fact tank that informs the public about the issues, attitudes and trends shaping the world. It conducts public opinion polling, demographic research, media content analysis and other empirical social science research. Pew Research Center does not take policy positions. It is a subsidiary of  The Pew Charitable Trusts .

© 2024 Pew Research Center

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• My Account Login
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 07 February 2024

Understanding the role of media in the formation of public sentiment towards the police

• Rayan Succar   ORCID: orcid.org/0000-0003-0927-3147 1 , 2 ,
• Salvador Ramallo 1 , 3 ,
• Rishita Das   ORCID: orcid.org/0000-0001-9785-5109 1 , 2 ,
• Roni Barak Ventura   ORCID: orcid.org/0000-0001-8186-3710 1 , 2 &
• Maurizio Porfiri   ORCID: orcid.org/0000-0002-1480-3539 1 , 2 , 4  

Communications Psychology volume  2 , Article number:  11 ( 2024 ) Cite this article

2816 Accesses

43 Altmetric

Metrics details

• Complex networks
• Computational science
• Human behaviour

Public sentiment towards the police is a matter of great interest in the United States, as reports on police misconduct are increasingly being published in mass and social media. Here, we test how the public’s perception of the police can be majorly shaped by media reports of police brutality and local crime. We collect data on media coverage of police brutality and local crime, together with Twitter posts from 2010-2020 about the police in 18 metropolitan areas in the country. Using a range of model-free approaches building on transfer entropy analysis, we discover an association between public sentiment towards the police and media coverage of police brutality. We cautiously interpret this relationship as causal. Through this lens, the public’s sentiment towards the police appears to be driven by media-projected images of police misconduct, with no statistically significant evidence for a comparable effect driven by media reports on crimes.

Similar content being viewed by others

Determinants of behaviour and their efficacy as targets of behavioural change interventions

Toolbox of individual-level interventions against online misinformation

Mechanisms linking social media use to adolescent mental health vulnerability

Introduction.

In recent years, public sentiment towards the police has been inflamed by events involving excessive use of force by law enforcement officers. Heated debates and law enforcement actions have fueled social movements in the United States (US) that play a key role in shaping public discourse, such as “Black Lives Matter" 1 , 2 , 3 . These movements gained significant traction following the death of George Floyd, who was murdered by police officers in Minneapolis on May 25, 2020 4 , 5 . Within these movements, voices to reform the police were often raised. In response to these calls, other movements that demand higher funding for the police were formed, such as “Blue Lives Matters". The social rupture regarding the police was recently reflected in a survey by the Associated Press of the National Opinion Research Center 6 , which pointed to a deep divide in public view regarding the judicial trials of police officers who were allegedly responsible for individuals’ injuries or deaths.

The public’s response to police brutality and its impact on police-community relations particularly influence police agencies in large metropolitan areas 7 , where events of police misconduct are most prevalent and police actions are subjected to greater scrutiny 8 , 9 , 10 . As evidence, Weitzer 11 found that high-profile incidents of police misconduct in Los Angeles and New York City had adverse effects on the public’s perception of the police departments implicated in the respective incidents. Likewise, Lasley 12 showed that Rodney King’s incident in Los Angeles in 1991 deeply impacted citizens’ attitude towards the Los Angeles Police Department. Recently, Hagan et al. 13 proposed that legal cynicism, defined by Sampson and Bartusch as “general beliefs about the legitimacy of laws and social norms” 14 , is exacerbated following events of police misconduct that lead to profound distrust in the police, especially among members of economically and racially isolated communities.

Similar to other contentious societal topics, the public’s perception of police misconduct is likely not shaped by facts alone but rather by how events are portrayed by mass media and more recently by social networks – the so-called “framing effect” 15 . For example, Kapuściński and Richards 16 determined that the choice of words used in reporting the news of terrorism could modify readers’ perceptions of risk, thereby creating a cognitive bias. Specifically, the authors observed that using the term “Al-Qaeda” strokes more fear in readers than the term “Domestic rebel separatist group”. In controlled studies with human participants, Boudreau et al. 17 and Mullinix et al. 18 studied the response to different forms of information about police brutality, including statistics, newspaper articles, and videos. From the analysis of the administered surveys, the authors concluded that people form a stronger, negative perception towards the police after witnessing such information.

While media coverage of police misconduct is an important actor in the formation of public attitude towards the police, there exists another potential driver: media coverage of crimes 19 , 20 . As it is the responsibility of police departments to maintain safety and order, frequent reports on crime may negatively affect the reputation of local police. Several studies have shown that local crime per se could have a strong impact on the public’s perception of the police whereby people who perceive their environment as unsafe tend to be less satisfied with the police than those who perceive their environment as safe 19 , 20 , 21 , 22 . However, the relationship between media coverage of local crime and the public’s perception of the police is generally understudied and partially dislinked from local crimes themselves, as it is often the job of the media to sensationalize events to draw an audience 23 . News reports may exaggerate stories of criminal events, either by highlighting the blameworthiness of offenders, aggrandizing criminal situations, or by emphasizing the police’s ineffectiveness in combating them 24 , 25 , 26 . On several occasions, research has demonstrated that the amount of media coverage of criminal events is largely uncorrelated with the actual crime rates, thereby creating the perception that crime rates continuously soar 27 , 28 .

The important role of media in forming the public’s perception towards law enforcement has not escaped the police. Police agencies widely acknowledge that police-community relationships are paramount to cultivating partnerships with citizens and creating a safer environment for them 29 . Through positive police-community relationships, the police could disseminate and receive information quickly and efficiently, recruit people and resources, and improve problem-solving processes 30 , 31 , 32 . In the end, the police strategically engage in symbiotic relationships with media outlets where they provide reporters with exclusive information about their activities and gain some control over their representation in the media 33 , 34 . Such a relationship could be used to open up the police to media inquiries and promote transparency and trust.

Whether it is driven by media coverage of police misconduct or local crime, public attitude towards the police is likely to change over time and in response to sensational events. Pinpointing the dynamic drivers of the formation of public attitude towards the police calls for overcoming the limitations of traditional survey-based research to create richer, time-resolved, multi-dimensional datasets. Such datasets could help formulate statistics-based approaches to objectively quantify how media coverage of both police brutality and local crime shape public attitude towards the police. Hillygus and Snell 35 pointed out that surveys often suffer from a lack of continuity and inconsistencies in the questionnaire design over the years. Similarly, Blossfeld et al. 36 raised practical concerns regarding recruiting participants and sampling responses.

In the past few years, social networks have emerged as a useful tool for gauging how public attitude towards various topics changes over time. It is estimated that more than 70% of the US population uses social media, whether to be informed about current events or to interact with each other 37 . Among the several social media platforms used to measure public attitude, Twitter is quickly becoming the platform of choice, offering an ultrafast “thermometer" of public attitude 38 , 39 . Twitter (now X) is a micro-blogging platform, where users can post visual and textual content with up to 280 characters, known as tweets. Unlike traditional survey instruments that were used in the past to assess public opinion regarding timely matters 40 , 41 , scraping Twitter posts allows for the collection of massive datasets over long time intervals, until the very minute an event of interest may have occurred. Moreover, Twitter posts are not constrained to the language of a survey so that “respondents” can express more complex thoughts. Twitter data have been proven useful in addressing a number of problems. Without trying to be exhaustive, Buntain et al. 42 demonstrated the successful use of Twitter to gauge public attitude regarding the 2013 Boston Marathon Bombing, Lampos and Cristianini 43 used Twitter data to track the Flu pandemic, Surano et al. 44 utilized them to understand people’s sentiment about the COVID lockdown in 2020, and Kraaijeveld and De Smedt 45 showed their power in forecasting cryptocurrency prices. In the context of police brutality events, Oglesby-Neal et al. 29 examined how the attitude towards the police changed after the death of Freddie Gray on April 19, 2015. Likewise, Mayes 46 used Twitter to reconstruct the public’s image of the police and compare it against the image the police intended to project onto social media.

In this study, we investigated the role of the media in the formation of “public sentiment” towards the police, measured from Twitter posts. Within the context of natural language processing, sentiment is understood as a classification of a piece of text in terms of the positivity and polarity of the emotions it reflects 47 . Thus, we operationally defined public sentiment as a collective measure of the positivity and polarity of the emotions of the public, measured by the sentiment of Twitter posts. In this vein, public sentiment serves as an approximation of the aggregate public mood towards a topic 48 . The concept of public sentiment is related but not equivalent to that of public opinion, which, according to Britannica 49 , refers to the aggregate views, attitudes, and beliefs of the public toward a particular topic. In criminal justice, public opinion may entail emotional, behavioral, and cognitive aspects 50 , whereas public sentiment does not necessarily reflect judgment and evaluation.

With these definitions in mind, we assembled an original dataset, consisting of 10 years of daily-resolved Twitter posts containing the public’s sentiment towards the police, media coverage of police brutality, and media coverage of local crime in 18 metropolitan areas in the US, for a total of more than two and a half million tweets and nearly two hundred thousand articles. Based on the technical literature 11 , 20 , 21 , 29 , we formulated the following hypotheses (Fig.  1 a):

H1-Public discourse about the police on Twitter will be influenced by media reports of local crime. Specifically, we expected that media coverage of local crime will influence both positive (H1a) and negative (H1b) public sentiment as perceived from Twitter posts: public sentiment towards the police will become more favorable when media coverage of local crime decreases, and conversely, public sentiment towards the police will become unfavorable when media coverage of local crime increases.

H2-Public discourse about the police on Twitter will be influenced by exposure to instances of police misconduct everywhere in the US through the media. Specifically, we anticipated that Twitter users will express distrust in the police as news about police brutality breaks, so that negative sentiment will proliferate in response to increased media coverage of police brutality.

a H1a and H1b refer to the influence of media coverage of local crime on public sentiment towards the police. H2 suggests that media coverage of police brutality, occurring anywhere in the US, drives negative public sentiment. b Each metropolitan area is represented with a circle, whose radius reflects the number of positive and negative tweets collected; areas were selected based on population size. Photo: 4zevar, Andrey-Kuzmin, ART.ICON, Black Creator 24, imagewriter, Oasis World, StockAppeal, sumberarto, Tatiana Garanina/Shutterstock.

To overcome the known limitations of linear correlation analyses in the discovery of causal associations 51 , 52 , 53 , we relied on an information-theoretic approach for hypothesis testing 54 , 55 . Specifically, we inferred a causal association from a source variable to a target variable in a Wiener-Granger sense, through the reduction of uncertainty in the prediction of the future of the target variable from its history due to additional knowledge about the history of the source. Within an information-theoretic approach, the uncertainty encoded in a variable is measured in terms of its entropy and causal associations are determined through transfer entropy 54 . An association that is discovered using transfer entropy does not mean that manipulation of the source variable will inevitably lead to a change in the target variable. Instead, such a link implies that the source variable helps in the prediction of the target variable, without ruling out the interference of unobserved variables 56 . Transfer entropy has been used to infer causality between two or more stochastic dynamical systems and the direction of information flow in several applications, ranging from financial markets 57 to the study of climate change 58 , 59 .

We chose to focus our study on the 18 most populated metropolitan areas in the US (Fig.  1 b), based on the following rationale. As one would expect, reliable inference of associations relies on large amounts of data 60 , which can be accessed if working with metropolitan areas. Not only are rates of police-related fatalities higher in more densely populated areas 8 , 9 , but also the attention of media to police brutality and protests about police violence gain more resonance when they take place in large cities 61 . Working with an original dataset, we put forward a mathematically backed, statistical framework that begets three main advantages with respect to the current state of knowledge. First, in contrast to the standard practice that relies on descriptive surveys of public opinion 11 , 21 , our analysis is performed on rich, multi-dimensional time series, whose coupled evolution is indicative of their interdependencies. Second, our approach allows for investigating the influence of media coverage of local crime on the public’s perception of the police across many US metropolitan areas. Third, the information-theoretic tools we employed to understand the driving factors of public sentiment towards the police overcome classical correlational analyses 62 .

Since predictions from an information-theoretic analysis come with assumptions on the dynamics of the variables at hand (such as stochasticity and separability) and are prone to the presence of unobserved (latent) variables, we validated our claims through three independent analyses. First, we performed statistical analysis with highly resolved data collected in the wake of George Floyd’s murder, a unique event in terms of impact and media coverage 5 . Specifically, we collected data from the most followed Twitter accounts of newspapers that were active during that period to unveil the chain of posting events between the accounts of newspapers and general users, adapting methodologies that are often used in the study of climate networks and extreme events 63 , 64 . Second, we applied an alternative methodology borrowed from the study of causal structures in ecology 52 , which does not employ a Wiener-Granger notion of causality like the one that underlies information-theoretic schemes 54 , 55 . Third, we employed another causal discovery approach developed by Gerhardus and Runge 65 based on the conditional independence 66 framework, to account for the possibility of the presence of unobserved variables.

Ethical standards

This study was not pre-registered. It was administratively reviewed by New York University (NYU)’s Institutional Review Board (IRB), determining that it does not meet the criteria for NYU’s engagement in research involving human participants as defined by 45CFR46.102. This determination was made because the study does not engage with human participants (that is, obtaining information about living humans through interaction or intervention or obtaining identifiable private information). The data collected in May 2022 for this study does not contain information that identifies the individuals who posted materials on Twitter, such as their name, age, gender, sex, race, or ethnicity.

Data collection and sentiment analysis

We collected data to generate time series for four variables between October 1, 2010 and December 31, 2020: media coverage of police brutality, media coverage of local crime, positive Twitter posts about the police, and negative Twitter posts about the police.

Data on media coverage of police brutality and local crime were collected using the ProQuest search engine. While news can be consumed through multiple media sources, including television, radio, and the internet, only print media can be collected in a reliable manner 34 . Databases that systematically record data on the content of news in media other than newspapers do not exist and scraping such large-scale information from the internet is not feasible. In the present study, media coverage of police brutality constitutes a “global variable”, as police brutality events have a reach that extends beyond city and state boundaries, spanning the entire US. That is, while local crimes are not typically of interest to people living elsewhere, incidents of police violence and their aftermath draw public attention at a national level. This effect was previously demonstrated by Weitzer 11 , who examined the cases of Abner Louima (sexually assaulted by a police officer in 1997) and Amadou Diallo (shot by a police officer who mistakenly thought he was pulling a gun out of his pocket in 1999), and by Fridkin et al. 67 , who studied the arrest of Ersula Ore (arrested for alleged assault after a confrontation with a campus police officer in 2014). Therefore, media coverage of police brutality was acquired by searching on ProQuest for the words “police” and “brutality" appearing together in sequence in the text or the headline of printed articles, published in the 20 most-circulated newspapers in the country 68 . We recorded the number of articles every day from October 1, 2010, to December 31, 2020. Since printed articles capture events that took place the day before, the time series was shifted backward by one day.

For media coverage of local crime, we focused our search on local newspapers. We collated data for the top 20 most-populated metropolitan areas in the US 69 . For each metropolitan area, we searched for the word “crime,” appearing together with the name of any of the cities with a population of more than 100,000 residents that form the metropolitan area under scrutiny. For example, for New York City metropolitan area, we searched for the word “crime” together with “New York City”, “Yonkers”, “Woodbridge”, “New Haven”, “Stamford”, “Smithtown”, “Paterson”, “North Hempstead”, “Newark”, “Huntington”, “Hempstead”, “Elizabeth”, “Waterbury”, “Oyster Bay”, “Edison”, “Brookhaven”, “Jersey City”, “Islip”, “Bridgeport”, or “Babylon.” Similar to the search on media coverage of police brutality, we searched for printed articles, published in the 20 largest newspapers by circulation in the corresponding state available on ProQuest 70 . For example, for the metropolitan area of New York City, we searched in the following newspapers: Philadelphia Inquirer, Pittsburgh Post-Gazette, New York Times, New York Daily News, Newsday, Rochester Democrat and Chronicle, Asbury Park Press, Press of Atlantic City, Hartford Courant, Buffalo News, Pittsburgh Tribune-Review, and Home News Tribune. The daily count of news articles that met these criteria between October 1, 2010 and December 31, 2020 was registered to generate a time series for each metropolitan area; see Supplementary Note  1 for further details on the ProQuest search, including detailed queries of all the metropolitan areas.

Next, we collected data on the public’s sentiment regarding the police from Twitter. Specifically, using the Python Library snscrape 71 , which relies on Twitter’s dedicated application programming interface (API), we scraped geo-located tweets for all the 20 metropolitan areas. The keywords used in the search included “police," “cop," and the abbreviation of the police department of the main city of the area (for example, “NYPD” for New York Police Department of the main city in the New York City metropolitan area). To identify the confined geographical boundaries of each metropolitan region, we approximated its area with circles of different radii. Tweet scraping was performed for each of these circular regions covering the complete metropolitan area. The search was conducted in May and June of 2022 and in compliance with Twitter’s terms of service at the time.

Given the time series of tweets in each metropolitan area, we performed aspect-based sentiment analysis to disentangle positive and negative sentiment from tweets. While other methods for sentiment analysis do not refer to any specific word (target), aspect-based sentiment analysis classifies the sentence with respect to a target based on the rest of the sentence (context). To this end, we made use of the DeBERTa model by Yang et al. 72 through the Pytorch package, a powerful state-of-the-art pre-trained language model with exemplary performance in context modeling.

In addition to people’s sentiment towards the police, we used Twitter to collect data on media coverage of police brutality with a minute-by-minute resolution. Specifically, we scraped for Twitter posts containing the term “police brutality" from Twitter accounts of the top ten most-followed newspapers in the country. Searching for tweets between October 1, 2010 and December 31, 2020 yielded a sparse time series, hence, we focused on the period around the murder of George Floyd, during which Twitter activity on the topic was intense. To avoid ceiling effects in the analysis, we began the collection five days after the murder and recorded 14 consecutive days from 00:00 on May 29, 2020 until 23:59 on June 13, 2020. This highly resolved time series was examined in conjunction with the time series of negative tweets about the police during the same two-week time window.

Two metropolitan areas (Seattle and San Diego) were excluded from the analysis due to limited data. San Diego had too few observations of media coverage of local crime (less than 0.8 per day), while Seattle’s data on media coverage of local crime was concentrated in a small time window (more than one third of the total values occurred within a couple of days), thereby challenging the processes of symbolization and detrending that are needed for the implementation of the methods.

Seasonal adjustment and detrending

One prerequisite to transfer entropy analysis is the stationarity of the processes under inspection. Therefore, all our daily time series were detrended and seasonally adjusted. An additive decomposition of the series using Loess regression 73 was performed with the statsmodels package in Python 74 , specifying weekly, monthly, and yearly seasons. The resulting trend and seasonal components were subtracted from the original time series.

Transfer entropy analysis

We quantified causal influence within an information-theoretic approach, based on transfer entropy. Within this framework, influence is measured as the enhanced ability to predict the future of a variable from its present due to the knowledge of the present of another variable. Entropy of a random variable X is its average information content (or uncertainty), defined as

where χ is the sample space of the variable X ,  x is any possible realization of X , and P refers to the probability of an event x . When entropy is measured in bits, the base of the logarithm is set to be 2. This quantity can not be negative by construction, and zero entropy refers to a deterministic variable.

For two processes X and Y , the joint entropy is given by

which can be understood as the global uncertainty of both variables. Similarly, the conditional entropy can be defined as

which measures the amount of information in variable X given knowledge of Y .

Assuming X and Y are a pair of discrete-time stationary stochastic processes, transfer entropy from Y (source) to X (target) is defined as

In this formulation, if knowledge of Y t does not improve the prediction of X t +1 , then both entropy terms will be identical and transfer entropy from Y to X will be zero. Alternatively, if Y encodes information that reduces the amount of entropy X and helps predict it, transfer entropy will have a positive value.

In the multivariate case where more than two variables are considered, there might be indirect coupling between processes that can lead to detection of spurious interactions between two variables 55 . In such cases, conditional transfer entropy can be used to mitigate this issue. Let Z i , for i  = 1, . . . ,  m , be other m stationary processes in addition to X and Y . Then, conditional transfer entropy from Y to X , with Z i conditional processes is defined as

Through this formulation, we control for the possibility of a process Z i acting as a common driver of X and Y , and ensure that we do not incorrectly infer an interaction between X and Y due to presence of a common driver. Moreover, conditional transfer entropy also accounts for spurious information flow by a possible cascade effect where X and Y are indirectly coupled through Z i .

From a practical point of view, there is seldom knowledge of the exact probability distributions for computation of entropy, as they are estimated from a finite time series sample. Therefore, we followed the standard practice of symbolizing our time series 75 . The symbolization was accomplished by binning the values of the time series into three equally sized quantiles, representing high, medium and low values of the time series. The length of the time series poses a constraint on both the maximum number of processes that can be examined at once and the number of symbols used 76 . In fact, there is a trade-off between the number of symbols used and the number of processes to include, since for s symbols we need to estimate s ( m +3) conditional probability values, where m is the number of confounding variables. Since the time series of media coverage contained the number of articles in daily print editions, it was shifted backwards by one day. Therefore, at a given time step t , a tuple of the four time series would represent the amount ("Low," “Medium," or “High") of positive tweets, and negative tweets, as well as the amount of media coverage of police brutality and crimes from printed outlets the following day.

Once the time series were symbolized, conditional transfer entropy could be estimated from Eq. ( 5 ). To determine whether the computed conditional transfer entropy value is significantly different from zero, we adopted the approach used by Runge et al. 77 and Porfiri et al. 78 Specifically, a surrogate distribution was created by shuffling the source variable while maintaining the structure of the dynamics of the target and conditioning variables. We first grouped the target variable and the conditioning variables together for every possible joint realization, and then we shuffled the symbols of the source within these groups. We computed the conditional transfer entropy for the shuffled time series, and repeated this procedure 20, 000 times to create a surrogate distribution. This distribution would represent values of transfer entropy obtained by chance. Finally, the p -value of the transfer entropy of the observed time series is given as the percentile of this estimated value within the ordered surrogate distribution. A transfer entropy value will be considered significantly greater than chance if it is larger than the 95 th percentile of the surrogate distribution.

We computed three values of conditional transfer entropy, corresponding to the hypothesized associations under H1a, H1b, and H2. Specifically, to assess the role of media coverage of local crime (MLC) on negative tweets (NT) and positive tweets (PT) about the police posited by H1a and H1b, we computed transfer entropy from MLC to NT and from MLC to PT, conditioned on media coverage of police brutality (MPB): T E MLC→NT ∣ MPB and T E MLC→PT ∣ MPB , respectively. To study the effect of MPB on NT underlying H2, we calculated T E MPB→NT ∣ MLC . The rationale for conditioning transfer entropy computations on a third variable was to mitigate common driver and cascading effects 55 , 79 ; for example, by using T E MLC→NT ∣ MPB we control for the possibility that MPB could simultaneously drive MLC and NT (common driver) or that MLC influences MPB, which, itself, influences NT (cascade).

Transfer entropy alone does not offer insight into the type of association between two variables; that is, transfer entropy does not indicate whether the association between two variables is positive (both variables are increasing or decreasing together) or negative (one variable increases while the other decreases). As such, another analysis is required for a complete hypothesis testing. Specifically, for the significant links, we computed the partial correlation between the past of the source variable and the present of the target variable, controlling for the past of the target variable and the past of the conditioning variable. To compute partial correlation, we used the Python pingouin package 80 , which implements the method developed by Kim 81 . All correlation analyses were performed with the non-parametric Spearman correlation 82 .

Analysis through convergent cross mapping

To offer independent backing to the transfer entropy analysis, we performed an additional causal analysis using a dynamical systems method called convergent cross mapping (CCM) 52 , which is not based on Wiener-Granger causality. CCM is a causal discovery framework developed for coupled dynamical systems. It is one of the most widely used causal inference techniques alongside Wiener-Granger causality, whereby it can be used for the detection of coupling in non-linear, non-stationary time series 83 , as well as in very short time series 84 . The method has been demonstrated in animal behavior 85 , neuroscience 86 , ecology 87 , and earth and climate science 88 , 89 .

CCM is built on Takens’ theorem for state-space reconstruction 90 , which states that a time delay embedding of a time series of a dynamical system is sufficient to reconstruct a diffeomorphic attractor of the original one, also known as a shadow manifold. It proposes that the successful reconstruction of time series X from a manifold of the time-delayed embedding of another time series Y implies coupling between X and Y . More precisely, if Y can be used to reconstruct X and not vice versa, we conclude that X is driving Y . To reconstruct one time series from a manifold, Sugihara et al. 52 proposed the use of the simplex projection method 91 , which is a linear weighted average of the nearest neighbors. Given two time series, x t and y t , the manifold M y ( t ) of the time delay embedding of delay τ and dimension E of y t is M y ( t ) = [ y t ,  y t − τ ,  y t −2 τ , …,  y t −( E −1) τ ] for t  = 0, …,  N where N is the length of the time series. The n nearest neighbors of M y ( t ), in the E -dimensional space are T 1 ( t ),  T 2 ( t ), …,  T n ( t ), from nearest to farthest. Using M y ( t ), the reconstructed time series

where w i is a weighting based on the distance between M y ( t ) and its i -th nearest neighbor M y ( T i ( t )),

with d [  ⋅  ] being the Euclidean distance.

The accuracy of the reconstruction is tested by inspecting the correlation between the reconstructed time series and the original time series. While any measure of correlation can be used, we chose to apply the Pearson correlation coefficient 92 . As explained by Sugihara et al. 52 , the larger the library length (that is, the number of data used for reconstruction), the denser the manifold becomes, thereby improving the correspondence of such nearest neighbors on the two manifolds. The three parameters of CCM are the embedding dimensions E and τ , and the number of the nearest neighbors n . According to Sugihara et al. 91 , the minimum number of nearest neighbors that can be used is E  + 1, which is also the one commonly used 84 . The embedding dimensions should be picked such that the attractor used in the reconstruction is best unfolded, which is accomplished with a simplex projection 84 . Herein, CCM analysis was performed using a Python code by Delforge et al. 93

Detection of latent variables

Transfer entropy cannot dismiss the presence of confounding variables 56 as it would require conditioning on all sources in the set of causal information contributors 94 , a virtually unfeasible task. To address the possibility of latent variables that could confound potential cause-and-effect relationship between MPB and NT, we employed the Latent Peter-Clark Momentary Conditional Independence (LPCMCI) framework 65 . LPCMCI is a state-of-the-art causal discovery algorithm for large temporal datasets based on the conditional independence framework 66 . The algorithm begins with a complete causal graph; through an iterative process of conditional independence tests that entails the removal and orientation of causal links, the algorithm removes non-significant edges until convergence. Importantly, for any pair of variables in the causal graph, the algorithm classifies the nature of their relationship and detects whether there is a latent variable that is driving both of them at the same time 65 , 95 . We used the publicly available Python library Tigramite for the detection of latent variables in all the 18 metropolitan areas using MPB, NT, and MLC. Consistent with the rest of the study, the significance level was set to α  = 0.05. The Pearson partial correlation was used as the conditional independence test, similar to CCM.

High-resolution analysis in the wake of George Floyd’s murder

For each metropolitan area and each Twitter post by newspapers on police brutality, we recorded NT about the police by users in a time window centered about the newspaper post. Each tweet from the media was assigned two numbers: the average number of users’ tweets per minute before the post and the average number of users’ tweets per minute after the post. As we sought to quantify the potential increase in the NT after a media tweet about police brutality, we performed a one-tailed Wilcoxon signed-rank test comparing the average NT per minute before and after each of the 205 instances of media tweets on police brutality. The null hypothesis of the test was that the two distributions share the same mean, and the alternative hypothesis was that the mean of the distribution of the average NT after media posts is greater than the mean of the other distribution. The same procedure was performed with several time intervals whose semi-width was systematically varied from 30 to 720 minutes, with 30-minute increments.

Reporting summary

Further information on research design is available in the  Nature Portfolio Reporting Summary linked to this article.

Data collection

Time series were generated for the four variables (MPB, MLC, PT, and NT), each consisting of 3745 measurements. MPB was a single daily time series for the entire country that aggregates the number of newspaper articles from 17 out of the 20 most-circulated newspapers in the country (Fig.  2 ). MLC assembled 18 daily time series of the number of newspaper reports on local crime, one for each metropolitan area. For the sake of illustration, we present time series for the metropolitan area of New York City (Fig.  3 a) and aggregated statistics for all the metropolitan areas (Fig.  4 ); all other times series are included in Supplementary Note  2 (Supplementary Figs.  1 – 17) . For PT and NT, we report the daily sampled time series of the New York City metropolitan area as an example (Fig.  3 b, c, respectively). For other metropolitan areas, we present aggregated statistics (Fig.  4 ); the complete set of times series is included in Supplementary Note  2 (Supplementary Figs.  1 – 17) . Finally, Fig.  5 illustrates the highly resolved time series of MPB after George Floyd’s murder.

Daily number of news articles mentioning “police brutality" in the 20 most-circulated newspapers in the US. A peak of 113 articles was recorded on June 13, 2020, 19 days after the death of George Floyd; activity in the wake of George Floyd’s murder is highlighted in red.

a Daily media coverage of local crime (MLC), with a peak of 44 articles registered on May 14, 2016. b Daily number of positive tweets (PT) about the police, with a peak of 104 registered on April 5, 2013. c Daily number of negative tweets (NT) about the police, with a peak of 1896 registered on May 31, 2020; activity in the wake of George Floyd’s murder is highlighted in red. d Zoomed-in view at the resolution of one minute of the number of negative tweets about the police in the wake of George Floyd’s murder period, from May 29, 2020 until June 13, 2020.

Daily means of positive tweets (green, PT), negative tweets (red, NT), and media coverage of local crime (blue, MLC) in the chosen metropolitan areas. The error bars represent one standard deviation from the mean.

Number of tweets posted by the 10 most-followed Twitter accounts of newspapers mentioning “police brutality" in the wake of George Floyd’s murder, from May 29, 2020 until June 13, 2020. A total of 209 tweets were recorded; the average time lapse between two consecutive tweets is 104.3 minutes.

Results of transfer entropy analysis for all metropolitan areas (Table  1 ) offer strong support in favor of H2 (the null hypothesis of independence was always rejected). On the contrary, partial support in favor of H1a was gathered for the metropolitan areas of Dallas and Washington, D.C. (the null hypothesis of independence was not rejected for all the other 16 metropolitan areas), while hypothesis H1b was only supported for the metropolitan area of Dallas, the null hypothesis of independence was rejected for all the other 17 metropolitan areas. For each of the significant associations, we computed the partial correlation and found a positive association (positive correlation coefficient) from MPB to NT in every metropolitan area, indicating that an increase in MPB would translate into an increase of NT throughout the country (Table  2 ).

In three supplementary analyses, we ensured that our results are robust. First, we ensured the robustness of transfer entropy with respect to memory effects, by performing further computations using longer time histories that accounted for more than one time step in the past (see Supplementary Note  3) . Second, we conducted transfer entropy analysis with time series of NT and PT generated with sentiment analysis using the classical natural language toolkit (NLTK) package of Bird et al. 96 , which relies on the Vader dictionary 97 (see Supplementary Note  4 for details). Finally, motivated by a recent survey by Gallup 98 suggesting that people are capable of gauging local crime rates (in spite of the biased depiction of crimes by the media), we conducted analogous analyses by utilizing local crimes rather than their media coverage (see Supplementary Note  5 for details). In all three studies, the conclusions of our transfer entropy analysis are equivalent. These analyses were not part of the initial hypotheses of the study, yet, they help understand the perception of the police by the general public as they experience crime in their neighborhoods.

CCM was used to test the hypotheses put forward (H1a, H1b, and H2). Similar to transfer entropy, CCM captured the coupling between MPB and NT through all metropolitan areas studied (Fig.  6 ). With respect to H1a and H1b hypotheses, MLC was poorly reconstructed from the Twitter time series in most cases (except for Houston), thereby confirming the conclusions of the transfer entropy analysis. Although we registered non-converging predictions in New York, MPB was better reconstructed. Hence, it can be stated that even if H1a and H1b should not be rejected in a few cases, MPB is the stronger driver of public sentiment towards the police.

. Each line corresponds to one of the four hypotheses (H1a, H1b, and H2). For each library length, a total of 30 estimations were performed. The error bars represent one standard deviation of the Pearson correlation coefficient estimated for each library length.

Based on the results in Supplementary Note  3 , the maximum lag was set to one and we focused on the synchronous (within one day) association between MPB and NT. For nine of the 18 metropolitan areas, the algorithm excludes the presence of any latent variable that would confound the association from MPB to NT (Table  3 ). For seven metropolitan areas, the algorithm yields inconclusive predictions (a latent variable may or may not exist), for one metropolitan area (Los Angeles) the presence of a link was deemed unlikely, and only in the Minneapolis metropolitan area it finds evidence of a latent variable (Table  3 ). Which is the exact latent variable that simultaneously drove MPB and NT is not an output of LPCMCI. It could be argued that it is no chance that Minneapolis, the location of George Floyd’s killing, is the singular metropolitan area in which a latent variable is present, but we have no means to pinpoint which latent variable may be in effect.

This analysis on the highly resolved time series (Fig.  3 d and Supplementary Figs.  1 – 17) was performed with various widths of the time window about the media post (Fig.  7 a), ranging from 60 minutes (30 before and 30 after) to 1440 minutes (720 before and 720 after). Statistical comparison for all the metropolitan areas (one-tailed Wilcoxon signed-rank test, n = 205) indicates that for the vast majority of metropolitan areas (at least 12 out of 18 over time windows between 240 and 1080 minutes, Fig.  7 b, c), there is an increase in the public’s expression of negative attitude towards the police on Twitter in response to media coverage of police brutality by newspapers on Twitter. This independent analysis offers further support to H2, while indicating that the time scale for the public to respond to the media is more than one hour and it can be as much as almost half a day.

a Sketch of the procedure used in the analysis; an artificial time series is utilized for better visualization. Purple thick vertical lines represent a tweet about police brutality from a newspaper and red lines are the numbers of negative tweets by the public. Smaller arrows identify 120-min time window (60 before and 60 after) and larger arrows 240-min period (120 before and 120 after), during which average numbers of negative tweets are calculated. b P -values of one-tailed Wilcoxon signed-rank test ( n  = 205) as a function of the size of the chosen time window around the posting of a tweet about police brutality by newspapers. The test compares the average number of negative tweets per minute about the police that are posted by the public before and after the posting of a media tweet mentioning “police brutality". The horizontal black line represents the chance at a significance level α  = 0.05. The null hypothesis is that there is no difference between the means of the distributions of the average numbers of negative tweets by the public before and after a post in a newspaper on police brutality. c Number of metropolitan areas for which the Wilcoxon test rejects the null hypothesis at α  = 0.05 as a function of the size of the chosen time window around the posting of a tweet about police brutality by newspapers -- number of metropolitan areas below the significance threshold in ( b ).

Mass media and social networks have been home to a growing debate regarding police brutality events 99 , 100 , 101 , suggesting that media coverage of such events can shape public sentiment towards the police. Another factor that could potentially drive the formation of public sentiment towards the police is media coverage of local crime, whereby the public will express more or less appreciation for the police as a function of the extent they feel safe. While several studies investigated the influence of local crime on one’s perception of the police, literature on the influence of media coverage of local crime is limited 23 —our study is the first to measure and investigate it directly.

In contrast to previous studies that employed surveys to assess the public’s opinion towards the police at a single point in time 11 , 62 , we employ a model-free, information-theoretic approach to study associations between media coverage of police brutality, media coverage of local crime, and public sentiment towards the police. We provide quantitative evidence for testing two hypotheses: H1) public sentiment towards the police (positive or negative) is influenced by media coverage of local crime, and H2) negative public sentiment towards the police is influenced by national media coverage of police brutality, which offers a proxy of police misconduct.

Our results do not offer evidence in favor of H1, whereby a role of media coverage of local crime on either positive (H1a) and negative (H1b) sentiment was not captured in almost any of the conditional transfer entropy analyses. The same conclusion was obtained by pursuing an alternative analysis with CCM, which tests a different notion of causality based on dynamical systems theory. This outcome is in agreement with the findings of Jackson and Sunshine 102 and Jackson and Bradford 103 . We cannot exclude the possibility that our analyses were not able to capture a relationship and we acknowledge that failing to reject a null hypothesis of independence between two variables is not sufficient to argue for their independence. However, since we were able to successfully detect associations with media coverage of police brutality with both information-theoretic and dynamical systems tools that aim to unveil causality, it is tenable that media coverage of local crime is not as strong of a driver of public sentiment of the police as media coverage of police brutality.

While not part of our original set of hypotheses about media coverage, we conducted additional analyses using objective measures of local crimes (number of local crimes and severity-weighted number of local crimes). Results of additional analyses for New York City, where such a data is made available to the public, yielded similar conclusions to those obtained with their local media coverage. Although limited to the case of New York City, the similarity between causal inferences obtained with number of crimes, number of crimes weighted by their severity, and their media coverage suggests that the public sentiment towards the police is only marginally affected by local crimes (whether gauged directly by residents or apprehended through the media). The absence of a causal effect of local crimes on public sentiment may find a basis in negativity bias theory, which states that humans tend to make sense of their environment based on negative information rather than positive information 104 . For example, Li et al. 105 and Miller et al. 106 showed that a negative encounter between citizens and the police has a stronger influence on their attitude towards the police than a positive encounter. In this vein, people would react less to successful crime prevention by law enforcement officers than they would to reports of police brutality.

In support of H2, our results revealed that media coverage of police brutality could be a key driver of negative sentiment of the public towards the police, whereby increased media coverage of police brutality affects the volume of the public discourse on Twitter (increased number of negative tweets). Furthermore, in agreement with the literature suggesting that news of police brutality reverberates throughout the country 11 , we unraveled the role of media reporting of police brutality at a national level. That is, all urban populations responded to instances of police misconduct on social media, even if the incidents took place elsewhere. Were the effect of police brutality on public sentiment towards the police only local (residents responding only to police brutality in their city), our analysis would unlikely yield a link from national media coverage to public sentiment. In such a hypothetical case, aggregating media at a national level would have “diluted” the causal role of the media and mitigated associations in some of the 20 metropolitan areas, in contrast with our findings.

Our finding is in line with the observations made by Miller et al. 106 , who proposed that media coverage of police routine has a limited effect on the public’s opinion towards the police, in contrast with reporting of misconduct and vicarious experiences of police abuse that shape our view of law enforcement. Such a response is also in agreement with previous accounts that found negativity to spread more than positivity on Twitter. For example, Schöne et al. 107 demonstrated that negative language regarding political matters tends to spread more than positive language; likewise, Hansen et al. 108 determined that the extent to which news become viral is affected by their negative content, such negative news content is more likely to be re-tweeted. Although our focus is on the appraisal of information through social media rather than real interactions between people and the police, there are similarities between our findings and the literature on the role of personal interactions between citizens and police. In particular, personal interactions with the police can leave a profound mark on citizens’ opinion about the police. For example, Skogan 109 surveyed 3005 citizens in Chicago and showed that a bad experience with the police significantly affects the public’s assessment of the police. Likewise, in controlled experiments, Mazerolle et al. 110 and Weisburd et al. 111 showed that policing that is carried out with dignity and respect is conducive to cooperation with citizens and may even lead to a reduction in crime incidents. Not only will negative interactions with the police hinder one’s confidence in the police, but they also spread to family and friends 106 . The role of cultural intelligence and other intercultural models on police-community work are reviewed by Louis and Grantham 112 .

Transfer entropy is based on the premise that causal associations do not instantaneously unfold, such that only the past of a source variable will contribute to the present uncertainty of a target variable. Such a condition is difficult to guarantee when working with Twitter data, whereby it is tenable that a sizable portion of the public will respond within the same day to a breaking news on police brutality 113 . To better detail the process by which the public reacts to these news, we performed an independent analysis at the resolution of one minute in the two-week period following the murder of George Floyd (starting four days after his death). In agreement with transfer entropy conclusions, we found a significant increase in the number of negative tweets from the public about the police following a tweet from a newspaper regarding police brutality across most of the metropolitan areas. Over the chosen two weeks, during which George Floyd’s death had become known to most of the country 5 , the public reacted to the media by expressing their negative sentiment to news that were being posted on Twitter accounts of newspapers. Had the association between media coverage of police brutality and negative public sentiment about the police been an artifact of the transfer entropy analysis, we should have not registered a clear sequence of posting events like the one we demonstrated; rather, posts of newspapers and general users should have been independently spaced in time. Even during this dire period of discourse on police brutality, the time scale of public response was on the order of several hours, thereby offering indirect backing to the use of daily data in the transfer entropy analysis for our observation window of more than 10 years.

Interestingly, the same behavior was observed throughout all the metropolitan areas under investigation. That is, irrespective of political leaning, population density, police budget, and other socio-demographic characteristics, people tweeted negative sentiment about the police following the outbreak of news about police brutality, paying limited attention to media about local crime. The prevalence of this consistent pattern throughout the US points at some key similarities in how we, as a society, appraise law enforcement. We cannot argue that such a consistency was expected at the beginning of our study, given the wide variation of human response to controversial, political topics similar to police brutality, such as the responses to the recent COVID-19 pandemic 114 and gun control 115 .

We presented multiple direct and indirect evidences that latent variables are unlikely to play a role on the observed associations. First, the analysis with LPCMCI 65 showed that, for the vast majority of metropolitan areas, there are no latent variables that would confound the causal link from police brutality to negative tweets. Second, in CCM, the values of ρ (Fig.  6 ) were as high as 90% for the majority of the metropolitan areas, indicating that media coverage of police brutality drives Twitter response nearly perfectly and ruling out the presence of any latent variable 116 . Finally, the high-resolution minute-by-minute analysis around the time of George Floyd’s murder offers compelling evidence that, throughout the country, people responded to news about police brutality on Twitter within a few hours, leaving limited room for alternative explanations in terms of latent variables. Overall, these results support the notion that a causal effect of media coverage of police brutality on the number of negative tweets discovered through multiple means (transfer entropy, CCM, and high-resolution, minute-by-minute analysis) is not an artifact of latent variables, but rather an instance of direct causation.

Limitations

The data acquisition and analysis of this study are not free of limitations. In particular, we identify five main limitations.

Sentiment analysis of Twitter data may not accurately represent the sentiment of the entire public. Certain sectors of the population may not use this social media platform and among those who use Twitter, likely not all express their views towards the police through tweets 117 . It is also tenable that a number of potentially useful tweets were not collected in this study as they were not geo-located. Although performed with state-of-the-art technique and validated with two different methods, we acknowledge that sentiment analysis is susceptible to noise in the detection of slang words, icons, and sarcastic expressions 118 .

The conclusions of the study may not generalize to every urban community, as we limited our analysis to large metropolitan areas that may experience more policing 119 . Differences in the exposure to police work may lead to different attitudes towards the police 120 . On top of that, people’s appraisal of the police may be shaped by their specific experience with law enforcement in the past 106 , 109 , 110 , 111 , as well as socio-demographic factors 121 . For example, it is known that legal cynicism and distrust in the police are higher in disadvantaged neighborhoods that are susceptible to crimes 13 , 14 . These measures cannot be disentangled from media coverage of police brutality or crime and their consumption by those communities. For example, Laniyonu 122 showed that people’s association with a cultural, ethnic, or racial group (black diaspora, in this instance) may instigate a stronger sentiment within those communities towards the police in response to brutality incidents.

Our analysis does not allow for teasing out the specific role of media coverage of local police brutality events on public sentiment. Due to the infrequency of these events, causal inference is difficult to perform. A transfer entropy analysis on time series with such a low information content may be prone to false inferences 75 .

The notion of causality that was discussed throughout the paper should be interpreted with caution as we only investigated observational data with a method that is not free of assumptions. Even though we overcame many of the limitations of transfer entropy by applying different causal inference techniques, these methods still have their own assumptions and limitations 123 . For example, LPCMCI assumes faithfulness, no selection bias, and acyclic causality in the time-series graphs, and it does not rule out the possibility of indirect causal links. For example, the association between media coverage of police brutality and public sentiment could be mitigated by the police’s own response to media coverage of police brutality; Through this lens, the police would react to news about police brutality and change their behavior accordingly, whether by adoption of new policies or by natural reaction of the officers to the news. In turn, this response could affect the public’s sentiment. While the high-resolution analysis in the wake of George Floyd’s murder hints at a direct effect due to the fast response, the aforementioned mechanism may not be ruled out as a general causal mechanism for longer reaction time, calling for similar research involving police behavior data.

While our results feature strong statistical significance, we warn care with respect to sampling of sentiment across the political aisle, whereby the composition of Twitter users in each metropolitan area may not fully replicate the make-up of Republican and Democratic voters at any point in time. The data presented in Yeung et al. 124 point to a representation of Democratic and Republican affiliations on Twitter in 2020 (56% and 44%, respectively) that is consistent with the party affiliation of US citizens reported in Gallup 125 data of the same period (52% and 48%). In contrast, the survey conducted by Pew Research Center between November and December 2018 (around mid-term elections) suggests an over-representation of Democratic users on Twitter 117 . Such differences may introduce a bias to the measure of public sentiment in a study, which varies by political ideology and affiliation 5 . Beyond the political composition of Twitter users, another factor to consider is the extent to which tweets by Democrats and Republicans vary in their positivity or negativity.

Conclusions

This study provides quantitative evidence of directional interdependencies between media coverage of local crime, media coverage of police brutality, and public sentiment towards the police, expanding on the state of knowledge that has largely relied on correlation analyses and descriptive statistics on single response, “snapshot” surveys 126 . This study examines the complete ecosystem of public-police interaction rather than some of its individual components. Through its original, information-theoretic approach, this study confirms our prediction of a strong influence of media coverage of police brutality on the public’s sentiment towards the police. At the same time, media reports of local crime, as well as local crimes, were not identified as salient drivers of public sentiment. This observation is likely to be rooted deep in our society, as recognized by Block 127 50 years back, when he wrote “citizen support for the police is constructed out of good and respectful policework. The negative effect of fear of the police on support for the police was far stronger than the positive effect of fear of crime". This notion brings to question a need for mechanisms to highlight the contributions of the police to society in addition to publicizing their misconduct, towards a debate of police reformation that is less biased and based on a holistic viewpoint of the role and need of law enforcement. Given the critical weight of negative news, media coverage should consider deliberate efforts towards reporting a more balanced projection of police-related news.

The study of public opinion is a long venture that can take on many forms. Childs 128 writes about 50 different definitions of public opinion. Taking a further step, Noelle 129 discusses the lack of consensus on a common operational definition and suggests abandoning the phrase “public opinion” entirely. This study does not seek to shed light on how people might rationally assess the political topics around the police and form an opinion about them. Rather, it tackles the spontaneous, impulsive, and emotional response of the public towards the actions of law enforcement, one of the many aspects of public opinion that bears a critical role in our hyper-connected society consuming and creating knowledge through mass and social media.

Violence and abuse of power seriously damage police reputation in the eyes of the public. In the context of urbanization, addressing police brutality and mitigating police-community tensions become even more critical. As cities become larger, they experience more crimes per capita at a super-linear scale 130 . In stark contrast, the size of the police force and budget that scale sublinearly in bigger cities 130 . That is, larger cities have less police funding and patrolling per resident. Given that events of police brutality stem from police officers’ fear of crime and injury 131 , 132 , we anticipate a bidirectional amplification of a negative trend where police and community grow further and further apart.

Data availability

Partial data needed to evaluate the conclusions in the paper are available on Github 133 . Raw Twitter data can not be shared, hence only the processed time series are made publicly available.

Code availability

All codes needed to evaluate the conclusions in the paper are available on Github 133 .

Rickford, R. Black Lives Matter: Toward a Modern Practice Of Mass Struggle Vol. 25, 34–42 (SAGE Publications, 2016). https://doi.org/10.1177/1095796015620171 .

Köksoy, E. & Demir, S. The impact of social media on agenda setting and public opinion formation: the Black Lives Matter movement as an example of social activism. Social Sci. Stud. J. 7 https://doi.org/10.26449/sssj.3130 (2021).

Dunivin, Z. O., Yan, H. Y., Ince, J. & Rojas, F. Black Lives Matter protests shift public discourse. Proc. Natl Acad. Sci. 119 , e2117320119 (2022).

Article   CAS   PubMed   PubMed Central   Google Scholar  

New York Times. How George Floyd Was killed in police custody (2020). https://www.nytimes.com/2020/05/31/us/george-floyd-investigation.html .

Reny, T. T. & Newman, B. J. The opinion-mobilizing effect of social protest against police violence: evidence from the 2020 George Floyd protests. Am. Political Sci. Rev. 115 , 1499–1507 (2021).

Article   Google Scholar  

AP NORC Center for Public Affairs Research. Growing disapproval of protests against police. (2020). https://apnorc.org/?post_type=project&p=2761 .

Lawrence, R. G. The Politics of Force: Media and the Construction of Police Brutality (Oxford University Press, 2022).

Edwards, F., Esposito, M. H. & Lee, H. Risk of police-involved death by race/ethnicity and place, United States, 2012–2018. Am. J. Public Health 108 , 1241–1248 (2018).

Article   PubMed   PubMed Central   Google Scholar  

Schwartz, G. L. & Jahn, J. L. Mapping fatal police violence across us metropolitan areas: Overall rates and racial/ethnic inequities, 2013-2017. PLoS ONE 15 , e0229686 (2020).

Premkumar, D. Public scrutiny, police behavior, and crime consequences: evidence from high-profile police killings. Police Behavior, and Crime Consequences: Evidence from High-Profile Police Killings (September 15, 2019) https://doi.org/10.2139/ssrn.3715223 (2019).

Weitzer, R. Incidents of police misconduct and public opinion. J. Crim. Justice 30 , 397–408 (2002).

Lasley, J. R. The impact of the Rodney King incident on citizen attitudes toward police. Policing Soc.: Int. J. 3 , 245–255 (1994).

Hagan, J., McCarthy, B., Herda, D. & Cann Chandrasekher, A. Dual-process theory of racial isolation, legal cynicism, and reported crime. Proc. Natl Acad. Sci. 115 , 7190–7199 (2018).

Article   CAS   PubMed   PubMed Central   ADS   Google Scholar  

Sampson, R. J. & Bartusch, D. J. Legal cynicism and (subcultural?) tolerance of deviance: the neighborhood context of racial differences. Law Soc. Rev. 32 , 777–804 (1998).

Tversky, A. & Kahneman, D. in The Framing of Decisions and the Psychology of Choice 25–41 (Springer, 1985).

Kapuściński, G. & Richards, B. News framing effects on destination risk perception. Tourism Manag 57 , 234–244 (2016).

Boudreau, C., MacKenzie, S. A. & Simmons, D. J. Police violence and public perceptions: an experimental study of how information and endorsements affect support for law enforcement. J. Politics 81 , 1101–1110 (2019).

Mullinix, K. J., Bolsen, T. & Norris, R. J. The feedback effects of controversial police use of force. Political Behav. 43 , 881–898 (2021).

Skogan, W. G. Concern about crime and confidence in the police: Reassurance or accountability? Police Q. 12 , 301–318 (2009).

La Vigne, N. G., Fontaine, J., Dwivedi, A. & Center, J. P. How Do People in High-crime, Low-income Communities View the Police? (Urban Institute Washington, DC, 2017).

Maxson, C. L., Hennigan, K. & Sloane, D. C. Factors that Influence Public Opinion of the Police (US Department of Justice, Office of Justice Programs, National Institute of Justice, 2003).

Skogan, W. G. Citizen satisfaction with police encounters. Police Q. 8 , 298–321 (2005).

Cheng, H. Factors influencing public satisfaction with the local police: a study in Saskatoon, Canada. Policing: Int. J. Police Strategies Manag 38 , 690–704 (2015).

Dixon, T. L. & Linz, D. Overrepresentation and underrepresentation of African Americans and Latinos as lawbreakers on television news. J. Commun. 50 , 131–154 (2000).

Beckett, K. & Sasson, T. The Politics of Injustice: Crime and Punishment in America (Sage Publications, 2003).

Gutsche Jr, R. E. Building boundaries: A case study of the use of news photographs and cultural narratives in the coverage of local crime and in the creation of urban space. Visual Commun. Q. 18 , 140–154 (2011).

Roche, S. P., Pickett, J. T. & Gertz, M. The scary world of online news? internet news exposure and public attitudes toward crime and justice. J. Quantitative Criminol. 32 , 215–236 (2016).

Dorfman, L. & Schiraldi, V. Off Balance: Youth, Race & Crime in the News (Building Blocks for Youth, 2001).

Oglesby-Neal, A., Tiry, E. & Kim, K. Public perceptions of police on social media. Urban Inst. 12 (2019).

Linders, D. From e-government to we-government: defining a typology for citizen coproduction in the age of social media. Gov. Inform. Q. 29 , 446–454 (2012).

Khan, G. F., Swar, B. & Lee, S. K. Social media risks and benefits: a public sector perspective. Social Sci. Comput. Rev. 32 , 606–627 (2014).

DePaula, N., Dincelli, E. & Harrison, T. M. Toward a typology of government social media communication: Democratic goals, symbolic acts and self-presentation. Gov. Inform. Q. 35 , 98–108 (2018).

Chermak, S. & Weiss, A. Maintaining legitimacy using external communication strategies: an analysis of police-media relations. J. Crim. Justice 33 , 501–512 (2005).

Chermak, S., Scheer, C. & Wilson, J. M. Police consolidation in the news. Police Q. 17 , 150–175 (2014).

Hillygus, D. S. & Snell, S. A. Longitudinal surveys: Issues and opportunities. Oxford Handbook on Polling and Polling Methods 28–52 (2015).

Blossfeld, H.-P., Skopek, J., Maurice, J. & Bayer, M. Methodological issues of longitudinal surveys. The Example of the National Educational Panel Study. Wiesbaden (2015).

Pew Research Center. Social media fact sheet. Pew Research Center: Washington, DC, USA (2017).

Morales, A. J., Borondo, J., Losada, J. C. & Benito, R. M. Efficiency of human activity on information spreading on Twitter. Social Netw. 39 , 1–11 (2014).

Cinelli, M. et al. The COVID-19 social media infodemic. Sci. Rep. 10 , 1–10 (2020).

Stimson, J. A. Public Opinion in America: Moods, Cycles, and Swings (Routledge, 1999).

Stimson, J. A. Tides of Consent: How Public Opinion Shapes American Politics (Cambridge University Press, 2004).

Buntain, C., McGrath, E., Golbeck, J. & LaFree, G. Comparing social media and traditional surveys around the Boston Marathon Bombing . 34–41 (2016).

Lampos, V. & Cristianini, N. Tracking the Flu Pandemic by Monitoring the Social Web 411–416 (IEEE, 2010). https://doi.org/10.1109/CIP.2010.5604088 .

Surano, F. V., Porfiri, M. & Rizzo, A. Analysis of lockdown perception in the United States during the COVID-19 pandemic. The Eur. Phys. J. Spec. Top. 231 , 1–9 (2021).

Google Scholar  

Kraaijeveld, O. & De Smedt, J. The predictive power of public twitter sentiment for forecasting cryptocurrency prices. J. Int. Financial Markets Inst. Money 65 , 101188 (2020).

Mayes, L. Law Enforcement in the Age of Social Media: Examining the Organizational Image Construction of Police on Twitter and Facebook . Ph.D. thesis, Temple University (2017).

Kawintiranon, K. & Singh, L. Knowledge Enhanced Masked Language Model for Stance Detection (2021). https://doi.org/10.18653/v1/2021.naacl-main.376 .

Nicholson-Crotty, S., Peterson, D. A. & Ramirez, M. D. Dynamic representation (s): Federal criminal justice policy and an alternative dimension of public mood. Political Behav. 31 , 629–655 (2009).

Davison, W. P. Public opinion (2022).

Ferraro, K. F. Fear of Crime: Interpreting Victimization Risk (SUNY press, 1995).

Mahoney, J. Beyond Correlational Analysis: Recent Innovations in Theory and Method 575–593 (JSTOR, 2001).

Sugihara, G. et al. Detecting causality in complex ecosystems. Science 338 , 496–500 (2012).

Article   CAS   PubMed   ADS   Google Scholar  

Runge, J., Nowack, P., Kretschmer, M., Flaxman, S. & Sejdinovic, D. Detecting and quantifying causal associations in large nonlinear time series datasets. Sci. Adv. 5 , eaau4996 (2019).

Article   PubMed   PubMed Central   ADS   Google Scholar  

Schreiber, T. Measuring information transfer. Phys. Rev. Lett. 85 , 461 (2000).

Bossomaier, T., Barnett, L., Harré, M. & Lizier, J. T. An Introduction to Transfer Entropy (Springer, 2016). https://doi.org/10.1007/978-3-319-43222-9_4 .

Smirnov, D. A. Spurious causalities with transfer entropy. Phys. Rev. E 87 , 042917 (2013).

Article   ADS   Google Scholar  

Zaremba, A. & Aste, T. Measures of causality in complex datasets with application to financial data. Entropy 16 , 2309–2349 (2014).

Hlinka, J. et al. Reliability of inference of directed climate networks using conditional mutual information. Entropy 15 , 2023–2045 (2013).

Article   MathSciNet   ADS   Google Scholar  

Runge, J. et al. Inferring causation from time series in earth system sciences. Nat. Commun. 10 , 1–13 (2019).

Article   CAS   Google Scholar  

Ramos, A. M. & Macau, E. E. Minimum sample size for reliable causal inference using transfer entropy. Entropy 19 , 150 (2017).

Olzak, S. Does protest against police violence matter? Evidence from US Cities, 1990 through 2019. Am. Sociol. Rev. 86 , 1066–1099 (2021).

Chermak, S., McGarrell, E. & Gruenewald, J. Media coverage of police misconduct and attitudes toward police. Policing: Int. J. Police Strategies Manag. https://doi.org/10.1108/13639510610667664 (2006).

Quiroga, R. Q., Kreuz, T. & Grassberger, P. Event synchronization: a simple and fast method to measure synchronicity and time delay patterns. Phys. Rev. E 66 , 041904 (2002).

Malik, N., Bookhagen, B., Marwan, N. & Kurths, J. Analysis of spatial and temporal extreme monsoonal rainfall over South Asia using complex networks. Clim. Dyn. 39 , 971–987 (2012).

Gerhardus, A. & Runge, J. High-recall causal discovery for autocorrelated time series with latent confounders. Adv. Neural Inform. Processing Syst. 33 , 12615–12625 (2020).

Pearl, J. Causality (Cambridge University Press, 2009).

Fridkin, K., Wintersieck, A., Courey, J. & Thompson, J. Race and police brutality: the importance of media framing. Int. J. Commun. 11 , 21 (2017).

Turvill, W. US newspaper circulations: America’s top 25 titles have lost 30% of print sales in two years (2018). https://pressgazette.co.uk/us-newspaper-circulations/ .

US Census Bureau. 2020 Population and Housing State Data (2021). https://www.census.gov/library/visualizations/interactive/2020-population-and-housing-state-data.html .

OfficialUSA. USA newspapers (2022). https://www.officialusa.com/stateguides/media/newspapers/ .

JustAnotherArchivist. snscrape: A social networking service scraper in Python. https://github.com/JustAnotherArchivist/snscrape .

Yang, H., Zeng, B., Xu, M. & Wang, T. Back to reality: Leveraging pattern-driven modeling to enable affordable sentiment dependency learning. Preprint at https://arxiv.org/abs/2110.08604 (2021).

Robert, C., William, C. & Irma, T. Stl: A seasonal-trend decomposition procedure based on loess. J. Off. Statistics 6 , 3–73 (1990).

Seabold, S. & Perktold, J. Statsmodels: Econometric and Statistical Modeling with Python. Proceedings of the 9th Python in Science Conference 57 , 92–96 (2010).

Staniek, M. & Lehnertz, K. Symbolic transfer entropy. Phys. Rev. Lett. 100 , 158101 (2008).

Article   PubMed   ADS   Google Scholar  

Marinazzo, D., Pellicoro, M. & Stramaglia, S. Causal information approach to partial conditioning in multivariate data sets. Comput. Math. Methods. Med. 2012 . https://doi.org/10.1155/2012/303601 (2012).

Runge, J. Causal network reconstruction from time series: from theoretical assumptions to practical estimation. Chaos: Interdisciplinary J. Nonlinear Sci. 28 , 075310 (2018).

Article   MathSciNet   CAS   Google Scholar  

Porfiri, M., Sattanapalle, R. R., Nakayama, S., Macinko, J. & Sipahi, R. Media coverage and firearm acquisition in the aftermath of a mass shooting. Nat. Hum. Behav. 3 , 913–921 (2019).

Article   PubMed   Google Scholar  

Sun, J., Taylor, D. & Bollt, E. M. Causal network inference by optimal causation entropy. SIAM J. Appl. Dyn. Syst. 14 , 73–106 (2015).

Article   MathSciNet   Google Scholar  

Vallat, R. Pingouin: statistics in python. J. Open Source Softw. 3 , 1026 (2018).

Kim, S. ppcor: an r package for a fast calculation to semi-partial correlation coefficients. Commun. Statistical Appl. Methods 22 , 665 (2015).

Kokoska, S. & Zwillinger, D. CRC Standard Probability and Statistics Tables and Formulae (CRC Press, 2000).

Verma, A. K., Garg, A., Blaber, A., Fazel-Rezai, R. & Tavakolian, K. Analysis of Causal Cardio-postural Interaction under Orthostatic Stress Using Convergent Cross Mapping 2319–2322 (IEEE, 2016). https://doi.org/10.1109/EMBC.2016.7591194 .

Clark, A. T. et al. Spatial convergent cross mapping to detect causal relationships from short time series. Ecology 96 , 1174–1181 (2015).

Roy, S., Howes, K., Müller, R., Butail, S. & Abaid, N. Extracting interactions between flying bat pairs using model-free methods. Entropy 21 , 42 (2019).

Beauchene, C., Roy, S., Moran, R., Leonessa, A. & Abaid, N. Comparing brain connectivity metrics: a didactic tutorial with a toy model and experimental data. J. Neural Eng. 15 , 056031 (2018).

Nakayama, S.-I., Takasuka, A., Ichinokawa, M. & Okamura, H. Climate change and interspecific interactions drive species alternations between anchovy and sardine in the western North Pacific: detection of causality by convergent cross mapping. Fisheries Oceanogr. 27 , 312–322 (2018).

Van Nes, E. H. et al. Causal feedbacks in climate change. Nat. Clim. Change 5 , 445–448 (2015).

Wang, Y. et al. Detecting the causal effect of soil moisture on precipitation using convergent cross mapping. Sci. Rep. 8 , 1–8 (2018).

ADS   Google Scholar  

Takens, F. In Detecting Strange Attractors in Turbulence , (ed. Manning, A.) 366–381 (Springer, 1981).

Sugihara, G. & May, R. M. Nonlinear forecasting as a way of distinguishing chaos from measurement error in time series. Nature 344 , 734–741 (1990).

Kowalski, C. J. On the effects of non-normality on the distribution of the sample product-moment correlation coefficient. J. Roy. Statistical Soc C: Appl. Statistics 21 , 1–12 (1972).

Delforge, D., Muñoz-Carpena, R., Van Camp, M. & Vanclooster, M. A parsimonious empirical approach to streamflow recession analysis and forecasting. Water Resour. Res. 56 , e2019WR025771 (2020).

Lizier, J. T. & Prokopenko, M. Differentiating information transfer and causal effect. Eur. Phys. J. B 73 , 605–615 (2010).

Article   CAS   ADS   Google Scholar  

Runge, J. Discovering Contemporaneous and Lagged Causal Relations in Autocorrelated Nonlinear Time Series Datasets 1388–1397 (PMLR, 2020).

Bird, S., Klein, E. & Loper, E. Natural Language Processing with Python: Analyzing Text with the Natural Language Toolkit (O’Reilly Media, Inc., 2009).

Hutto, C. & Gilbert, E. Vader: A Parsimonious Rule-based Model for Sentiment Analysis of Social Media Text https://doi.org/10.1609/icwsm.v8i1.14550 (2015).

Brenan, M. Record-high 56% in U.S. perceive local crime has increased. https://news.gallup.com/poll/404048/record-high-perceive-local-crime-increased.aspx (2022).

Cobbina-Dungy, J. E. & Jones-Brown, D. Too much policing: Why calls are made to defund the police. Punishment Soc 25 , 14624745211045652 (2021).

Martin, J. Breonna Taylor: Transforming a hashtag into defunding the police. J. Crim. Law Criminol. 111 , 995 (2021).

Patnaude, L., Lomakina, C. V., Patel, A. & Bizel, G. et al. Public emotional response on the Black Lives Matter movement in the summer of 2020 as analyzed through Twitter. Int. J. Marketing Stud. 13 , 1–69 (2021).

Jackson, J. & Sunshine, J. Public confidence in policing: a neo-durkheimian perspective. Br. J. Criminol. 47 , 214–233 (2007).

Jackson, J. & Bradford, B. Crime, policing and social order: On the expressive nature of public confidence in policing. Br. J. Sociol. 60 , 493–521 (2009).

Vaish, A., Grossmann, T. & Woodward, A. Not all emotions are created equal: the negativity bias in social-emotional development. Psychol. Bull. 134 , 383 (2008).

Li, Y., Ren, L. & Luo, F. Is bad stronger than good? The impact of police-citizen encounters on public satisfaction with police. Policing: Int. J. Police Strategies Management 39 https://doi.org/10.1108/PIJPSM-05-2015-0058 (2016).

Miller, J., Davis, R. C., Henderson, N. J., Markovic, J. & Ortiz, C. W. Public opinions of the police: The influence of friends, family and news media (Vera Institute of Justice New York, 2004).

Schöne, J. P., Parkinson, B. & Goldenberg, A. Negativity spreads more than positivity on Twitter after both positive and negative political situations. Affective Sci. 2 , 379–390 (2021).

Hansen, L. K., Arvidsson, A., Nielsen, F. Å., Colleoni, E. & Etter, M. in Good Friends, Bad News-affect and Virality in Twitter 34-43 (Springer, 2011). https://doi.org/10.1007/978-3-642-22309-9_5 .

Skogan, W. G. Asymmetry in the impact of encounters with police. Policing Soc. 16 , 99–126 (2006).

Mazerolle, L., Antrobus, E., Bennett, S. & Tyler, T. R. Shaping citizen perceptions of police legitimacy: A randomized field trial of procedural justice. Criminology 51 , 33–63 (2013).

Weisburd, D. et al. Reforming the police through procedural justice training: A multicity randomized trial at crime hot spots. Proc. Natl Acad. Sci. 119 , e2118780119 (2022).

Louis, A. & Grantham, H. in Cultural Intelligence and Community Policing 37–60 (Springer, 2019). https://doi.org/10.1007/978-3-030-22002-0_3 .

Hu, M. et al. Breaking News on Twitter 2751–2754 https://doi.org/10.1145/2207676.2208672 (2012).

Bonnevie, E., Gallegos-Jeffrey, A., Goldbarg, J., Byrd, B. & Smyser, J. Quantifying the rise of vaccine opposition on Twitter during the COVID-19 pandemic. J. Commun. Healthcare 14 , 12–19 (2021).

Steidley, T. & Colen, C. G. Framing the gun control debate: press releases and framing strategies of the National Rifle Association and the Brady Campaign. Social Sci. Q. 98 , 608–627 (2017).

Ye, H., Deyle, E. R., Gilarranz, L. J. & Sugihara, G. Distinguishing time-delayed causal interactions using convergent cross mapping. Sci. Rep. 5 , 1–9 (2015).

Wojcik, S. & Hughes, A. Sizing up Twitter users. PEW Research Center 24 (2019).

Liu, C. et al. Improving sentiment analysis accuracy with emoji embedding. J. Safety Sci. Resilience 2 , 246–252 (2021).

Brunson, R. K. & Miller, J. Young black men and urban policing in the United States. Br. J. Criminol. 46 , 613–640 (2006).

Verga, C., Murillo, L., Toulon, E. D., Morote, E.-S. & Perry, S. M. Comparison of race-gender, urban-suburban criminal justice college students satisfaction of the police department. J. Leadership Instruction 15 , 33–38 (2016).

Pryce, D. K. & Gainey, R. Race differences in public satisfaction with and trust in the local police in the context of George Floyd protests: an analysis of residents’ experiences and attitudes. Crim. Justice Stud. 35 , 74–92 (2022).

Laniyonu, A. Phantom pains: the effect of police killings of black americans on black british attitudes. Br. J. Political Sci. 52 , 1–17 (2021).

Reiser, C. Causal discovery for time series with latent confounders. Preprint at https://arxiv.org/abs/2209.03427 (2022).

Yeung, N., Lai, J. & Luo, J. Face Off: Polarized Public Opinions on Personal Face Mask Usage during the Covid-19 Pandemic 4802–4810 (IEEE, 2020). https://doi.org/10.1109/BigData50022.2020.9378114 .

Gallup. Party Affiliation (2020). https://news.gallup.com/poll/15370/party-affiliation.aspx .

Brown, B. & Benedict, W. R. Perceptions of the police: Past findings, methodological issues, conceptual issues and policy implications. Policing: Int. J. Police Strategies Manag. 25 https://doi.org/10.1108/13639510210437032 (2002).

Block, R. L. Fear of crime and fear of the police. Social Problems 19 , 91–101 (1971).

Childs, H. L. Public Opinion Nature Formation and Role (D van Nostrand company, 1965).

Noelle-Neumann, E. Public opinion and rationality. Public Opin. Commun. Consent 33 (1995).

Banerjee, S., Van Hentenryck, P. & Cebrian, M. Competitive dynamics between criminals and law enforcement explains the super-linear scaling of crime in cities. Palgrave Commun.s 1 , 1–7 (2015).

New York Times. Unfounded fear helps fuel police violence (2020). https://www.furman.edu/news/2020/06/24/unfounded-fear-helps-fuel-police-violence/ .

NBC News. Police violence is mostly rooted in fear. Ignoring that makes reform harder. (2020). https://www.nbcnews.com/think/opinion/police-violence-mostly-rooted-fear-ignoring-makes-reform-harder-ncna1230266 .

https://github.com/dynamicalsystemslaboratory/Police-Brutality . https://doi.org/10.5281/zenodo.10460194 .

Download references

Acknowledgements

This study was supported by the National Science Foundation under award CMMI-1953135. S.R. acknowledges the support of a Spain-US Fulbright grant co-sponsored by Fundación Séneca and of Ministerio de Ciencia e Innovación of Spain under grant PID2019-107192GB-I00. R.B.V. was supported in part through a postdoctoral award from The National Collaborative on Gun Violence Research. The funders had no role in study design, data collection and analysis, and decision to prepare or publish the manuscript. The views expressed in this article are the authors’ and do not necessarily reflect the view of the funders. The authors would like to express their gratitude to the NYU CUSP Capstone team “Understanding public opinion towards the police in New York City" (Ruoqing Lin, Lingxuan Bu, and Xiangyu Ying), who helped in data collection. This work was supported in part through the NYU IT High Performance Computing resources, services, and staff expertise.

Author information

Authors and affiliations.

Center for Urban Science and Progress, New York University, Brooklyn, NY, 11201, USA

Rayan Succar, Salvador Ramallo, Rishita Das, Roni Barak Ventura & Maurizio Porfiri

Department of Mechanical and Aerospace Engineering, Tandon School of Engineering, New York University, Brooklyn, NY, 11201, USA

Rayan Succar, Rishita Das, Roni Barak Ventura & Maurizio Porfiri

Department of Quantitative Methods, University of Murcia, Murcia, 30100, Spain

Salvador Ramallo

Department of Biomedical Engineering, Tandon School of Engineering, New York University, Brooklyn, NY, 11201, USA

Maurizio Porfiri

You can also search for this author in PubMed   Google Scholar

Contributions

Conceptualization—R.S., S.R., M.P.; methodology—R.S., S.R., M.P.; software—R.S., S.R., M.P.; formal analysis—R.S., S.R., M.P.; resources—M.P.; data curation—R.S., S.R.; writing—original draft preparation—R.S., S.R., R.D., R.B.V., M.P.; writing-review and editing—R.S., S.R., R.D., R.B.V., M.P.; visualization—R.S., S.R.; supervision—M.P.; funding acquisition—S.R., R.B.V., M.P.

Corresponding author

Correspondence to Maurizio Porfiri .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Peer review

Peer review information.

Communications Psychology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editor: Marike Schiffer. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Peer review file, supplementary information, reporting summary, rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Succar, R., Ramallo, S., Das, R. et al. Understanding the role of media in the formation of public sentiment towards the police. Commun Psychol 2 , 11 (2024). https://doi.org/10.1038/s44271-024-00059-8

Download citation

Received : 25 September 2023

Accepted : 19 January 2024

Published : 07 February 2024

DOI : https://doi.org/10.1038/s44271-024-00059-8

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing: AI and Robotics newsletter — what matters in AI and robotics research, free to your inbox weekly.

How reforms could target police racism and brutality — and build trust

Berkeley social psychologist Jack Glaser is alarmed — though not surprised — by the recent storm of police violence. But he sees paths toward more humane, effective law enforcement

By Edward Lempinen

June 9, 2020

When police are armed and clad in riot gear, they may see themselves more as warriors battling an enemy than as guardians protecting a community. (Image by Becker1999 via Wikimedia Commons, Creative Commons)

In the turbulent days since the police killing of George Floyd in Minnesota, Jack Glaser has been following the storm of protests, including dozens of incidents in which police appeared to escalate conflicts, use excessive force and target journalists. Like millions of others in the United States and worldwide, he is alarmed by what he’s seen.

Glaser, however, is watching through a different lens. He is a UC Berkeley social psychologist and an expert in police practices, and where some might see a stark conflict between police and protesters, he sees a more complex dynamic: The devastating history of American racism and the biases that all people carry. Police officers’ difficult and dangerous work. And a political system that is inconsistent, at best, in its approach to police reform.

Jack Glaser

“I wouldn’t infer that there’s anything particularly unusual happening now,” he said. “Black Americans have been encountering this for centuries. It is a steady-state problem. … The main difference we’re seeing now is that the capacity to video record these incidents is widespread. But it’s no surprise to the black community — they have known it’s been happening.”

In a wide-ranging interview, Glaser said racial prejudice remains deeply rooted in many of the nation’s 18,000 police agencies — a toxic legacy that can only be resolved through strong policy applied across many years. But he sees an emerging generation of innovative police leaders and a variety of promising efforts to improve U.S. law enforcement — perhaps including a shift of some police funding to other social services.

Glaser is a professor in the Goldman School of Public Policy at Berkeley, an expert in implicit bias among police and author of Suspect Race: Causes and Consequences of Racial Profiling (Oxford University Press, 2014). He is working with the Center for Policing Equity to build a database that tracks police stops and incidents where officers use force, and he has consulted with police departments in the Bay Area and elsewhere in California to assess and improve practices.

‘There’s no turning away’

Police violence against black men has been a persistent American tragedy. What makes the death of George Floyd such a tectonic shock was the seeming indifference of Minneapolis Police Officer Derek Chauvin and colleagues who were taking Floyd into custody, all while being filmed.

“Nine minutes of a police officer with his knee on the neck of a man who’s saying he can’t breathe,” Glaser said. “This man is being held down by two other officers, so there’s not even a concern about flight. We can all see him die in slow motion. There’s no turning away from that.”

And yet, Chauvin and other police officers who kill or brutalize are the exceptions, not the norm. Glaser expressed deep sympathy for the challenges of men and women in law enforcement. They’re human, and they’re complex. And society gives them conflicting messages: Should they be guardians? Or warriors?

Putting them into a volatile protest, or into a riot, is the ultimate stress test.

“They’re going into a situation where they think they might get hurt,” Glaser explained. “And there are very few people who are not afraid under those circumstances. … Some officers have told me that, in the first situation where they came under fire, they soiled themselves. It’s totally understandable.”

But in a moment of chaos and potential violence, fear can make some officers aggressive.

“One way people might cope with fear is to evoke some inner resolve, or even anger, and to use that to energize their response to a threat,” Glaser said. “Some officers have an intuitive sense that that’s not going to work out well, and so they resist it. Some officers just really go with it.

“The video from the officers clearing the protesters in D.C. a few days ago is a good example of that. These officers were probably told, ‘It’s imperative that you clear that corridor, so the president can walk in there.’ Maybe they took that to mean, ‘Do whatever you need to do.’ So, they’re being very forceful, very aggressive, bashing journalists with their shields. It wouldn’t seem to make sense at all, unless lives are on the line. All they were doing was putting lives at risk.”

The role of ‘implicit bias’ in police action

In circumstances that are chaotic and potentially dangerous, the moment when police engage with protesters is especially fraught, Glaser said. Clad with military body protection and armed with advanced weapons, they may see themselves more as warriors battling an enemy than as guardians protecting a community. They may become more aggressive.

An officer’s implicit biases might amp up that response. “We know that humans have automatic stereotypes that get activated about young people, males, about minorities that are going to cause them to interpret their behaviors as threatening,” Glaser explained. “So, if they see a young man of color wearing a mask walking toward them, that could be more threatening to them than a young white woman, or an older white woman.”

A different sort of bias may explain the restrained police response to recent protests by armed right-wing groups in Michigan , Kentucky and Ohio .

“We have to seriously consider the possibility that the passive police response to armed white protesters was due to racial preferences,” Glaser said. “At the very least, white police officers are going to be more likely to give white anti-government protesters the benefit of the doubt. The data on race and policing is too clear that black people are treated with greater suspicion, disrespect and physical force.”

Does the overarching political climate influence those responses? Glaser noted that President Donald Trump, over the course of three-plus years, has encouraged police to be aggressive.

“He’s calling the media the ‘enemy of people’ and retweeting videos where he’s beating up the figure of a person whose head is the CNN logo,” Glaser said. “Cops are disproportionately Republican and conservative. They probably watch more Fox News than the general public does. I think that contributes.”

A new generation rises in the ranks

Since at least the 1960s, the topic of police misconduct and brutality — often against people of color, LGBTQ people or protesters — has been a constant debate in American politics and policy.

Still, across those six decades, society has made only uneven progress, if that. In the past decade alone, police across the country have been found culpable in brutality, torture and killings , but often without penalty.

President Barack Obama recognized the need for reform, Glaser said, and his Task Force on 21st Century Policing introduced a number of initiatives to discourage police violence, including restrictions on the transfer of surplus military equipment to police. However, he said, the Trump administration has reversed many of those policies.

Still, Glaser is optimistic. He sees hopeful signals in a new generation of police chiefs — Art Acevedo in Houston, William Scott in San Francisco and Chris Magnus, formerly in Richmond, California, and now in Tucson — and in the leadership of the Major Cities Chiefs Association .

‘Walk with us! Walk with us!’

A common prescription for police misconduct is training to counter implicit bias, but Glaser sees no evidence that such training works.

Instead, he suggests that both the warrior and guardian models of policing be replaced by a model in which police see themselves as protectors of public safety. He advocates practical policies that departments can embrace to deter misconduct and build community trust, while recognizing the challenges of law enforcement. Hiring more women would be one valuable strategy, he said.

A rising protest movement is urging that police departments be “defunded,” and while Glaser called the concept “difficult,” he sees potential benefits.

“As far as I can tell, many people calling for that mean only to decrease police funding and to reallocate those funds to other social services,” he said. “If that is what is meant, that sounds prudent. Police leaders will tell you that their officers spend too much time doing things like dealing with mental health and domestic violence crises that they are not well-equipped to handle. Reallocation and restructuring of public safety funds could help.”

Such structural changes would take time. But today, Glaser said, in everyday policing, “there are all kinds of instructions that police supervisors can give that will make it easier for the officers to put their emotions aside and just do what they’re told.” Among them: Keep a distance from protesters. Speak in a calm voice. Don’t wear combat boots.

Under court order, New York City has dramatically reduced the number of discretionary stops made by police, which often targeted people of color. In Oakland, police are now prohibited from engaging in high-speed chases. San Francisco and Tucson have implemented all eight of the policies promoted by Campaign Zero’s Use of Force Project .

Glaser is further encouraged by the number of instances since the killing of Floyd in which police and law enforcement agencies all over the country have expressed support for those who are protesting such brutality.

“There was a very poignant video from Flint, Michigan,” he recalled. “The county sheriff, Chris Swanson, took off his helmet and put down his baton, and he asked the protesters, ‘What do you want from me?’ And they said, ‘Walk with us! Walk with us!’ And he did.

“Of course that, by itself, won’t solve many of the problems police have in their communities,” he said. “But it’s a good model for police in a time of conflict: Take a step back. Open up. Have a dialogue. If police departments could institutionalize these practices, it would build trust and really help them to reduce aggressive policing.”

We've detected unusual activity from your computer network

To continue, please click the box below to let us know you're not a robot.

Why did this happen?

Please make sure your browser supports JavaScript and cookies and that you are not blocking them from loading. For more information you can review our Terms of Service and Cookie Policy .

For inquiries related to this message please contact our support team and provide the reference ID below.

How policing has changed 4 years after George Floyd’s murder

Geoff Bennett Geoff Bennett

Sam Lane Sam Lane

Karina Cuevas Karina Cuevas

Leave your feedback

• Copy URL https://www.pbs.org/newshour/show/how-policing-has-changed-4-years-after-george-floyds-murder

This month marked four years since the murder of George Floyd at the hands of police in Minneapolis. Floyd’s killing sparked a global uprising and sweeping promises of racial justice and police reform. But four years later, there’s been some backlash to the changes that were set into motion and in some cases, public attitudes have changed. Geoff Bennett discussed that with Phillip Atiba Solomon.

Read the Full Transcript

Notice: Transcripts are machine and human generated and lightly edited for accuracy. They may contain errors.

Geoff Bennett:

This past Saturday marked four years since the murder of George Floyd at the hands of police in Minneapolis.

Floyd's killing sparked a global uprising and sweeping promises of racial justice and police reform. But, four years later, there's been some backlash to the changes that were set into motion and in some cases public attitudes have shifted.

To help assess where things stand now with police reforms, we're joined by Phillip Atiba Solomon. He leads the Center for Policing Equity and is chair of African American studies and a professor of psychology at Yale University.

Thanks so much for being with us.

Phillip Atiba Solomon, Center for Policing Equity: Thanks, Geoff.

We know at the federal level the George Floyd Justice in Policing Act has stalled, but at the state and local levels, how much would you say policing has actually changed since May of 2020?

Phillip Atiba Solomon:

So, when we talk about policing, we can talk about that at multiple different levels.

When we talk about it, we can say the culture of policing itself. We can talk about attitudes around policing. But I assume what you mean are the policies that regulate policing.

And, there, I can say it's kind of a mixed bag, right?

So we have had some places that looked to literally abolish their entire police department and replace it with departments of public safety and some places that were making more incremental change. Some of the incremental change, like bans on choke holds, new pursuit policies, those have moved forward, and some of the attempts to reduce police budgets have maintained as well.

But I have to say, post the murder spike in 2021 and 2022, we found ourselves unable to maintain that broad-spread momentum. What we can say, the good news around this is, in multiple municipalities, there are example projects, things that move forward that people are looking at and saying, well, this could be the way that we make good on the moral imperative that came post-George Floyd and the political realities that came in the backlash of that.

Places like Denver that's had the STAR program, that has replaced law enforcement with mental health response in certain crises, places like Austin, and whole states like Washington state and Connecticut that are set to eliminate low-level traffic enforcement by police.

So it's a mixed bag. There's some really exciting stuff happening right now, and there's definitely been some backsliding. I don't think that we're done with the analysis of the full consequences of that moment.

Well, given your point, the degree to which the politics have changed, we have seen conservative states, even some progressive areas, pass these new tough-on-crime laws.

At this point, are there key structural reforms to policing that are both practical and scalable that you think would be effective?

Yes, and I just want to reset here, because we're talking about the term reform, and reform has a particular meaning in this context.

Reform means making the systems we have got better. And there are absolutely some things that we can do that can produce reforms that are laudable and can help ensure public safety, particularly in vulnerable Black and brown communities.

But there's another thing we got to be working on, which is entirely building up new systems and replacing the ones that we have got. Both/and are necessary. So, for instance, I talked about Washington state and Connecticut looking to ban all low-level traffic enforcement by police. This is because there's no evidence that doing that makes anything any safer.

Crashes aren't reduced by it, communities don't feel any safer, and, by the way, law enforcement doesn't want to be doing it. That is a kind of reform, right, it's a change in policy, making the systems we have got better, that folks can get behind and hopefully can both improve public safety and the efficiency of collecting whatever enforcement fees we might need. And it gets law enforcement out of the places they don't need to be.

But we're not going to stop the cycle of seeing the ugliness we saw in the aftermath of the lynching of George Floyd unless we invest upstream so that communities aren't in crisis in the first place. If communities are in crisis, the things that we send in response are never going to be sufficient.

And I'm glad to hear that we're here when we have got cities like Evanston, Illinois, cities like Ithaca, New York, and Berkeley, California, that are working to increase their spending on the social safety net so that folks don't have to call out for emergency medical, fire, or law enforcement in the first place.

Well, broadly speaking, in the wake of George Floyd's murder, there were lots of promises and millions of dollars spent on racial justice efforts, on efforts aimed at rethinking policing.

What's your take on the status of those promises right now?

Yes, you saw a lot of Twitter activism and political statements that got made, a lot of commitments of money, but those commitments of money didn't always come to fruition.

There were a lot of corporations that said, we need to do better internally, and we are committed to making sure our communities get the resources they need. And it turns out not as much got spent in that way as we had hoped.

We had a lot of municipalities saying that we were going to invest in upstream resources to keep people from being in crisis, and we were going to take that from police budgets. But what we know is that police budgets have expanded in that period of time since 2020. They have not shrunk nationwide.

In fact, there are very few cities where it's shrunk at all. So what I'd say is, as a country, we weren't ready yet again for this moment. It's not that it hasn't happened many times before, but we weren't ready for something on the scale of the moral outrage as what we saw in the aftermath of the lynching of George Floyd. We just weren't ready for it.

And so we had great ability for acute empathy, and for in-the-moment decisiveness and commitments. But we had not strapped in for the long haul, which is what's necessary to manage a system and reform, alter, replace a system that's $115 billion annually. That's how much we spend on law enforcement.

It is no less complicated than education or health care, and it's not going to be any easier to give ourselves the kind of systems in public safety that we need than it is in education and health care. The good news, again, the good news is, there was a wakeup call to the country. And there are people who understand that the ways that we have been trying to keep folks safe in crisis, the aftermath of a lack of investment doesn't make sense.

It's not cost-efficient. It's not equal. It's not just. It's not right. And so a more fulsome consideration, a more fulsome moral conversation has been sticky in this country. And as we get individual municipalities that are winning on some of these issues and demonstrating that new directions can actually keep communities safer, I am optimistic that we're going to see that start to grow.

Phillip Atiba Solomon is chair of African American studies and a professor of psychology at Yale University.

Thanks so much for being with us. We appreciate it.

Listen to this Segment

Watch the Full Episode

Geoff Bennett serves as co-anchor of PBS NewsHour. He also serves as an NBC News and MSNBC political contributor.

Sam Lane is reporter/producer in PBS NewsHour's segment unit.

Support Provided By: Learn more

More Ways to Watch

• PBS iPhone App
• PBS iPad App

Educate your inbox

Subscribe to Here’s the Deal, our politics newsletter for analysis you won’t find anywhere else.

Thank you. Please check your inbox to confirm.

An official website of the United States government, Department of Justice.

Here's how you know

Official websites use .gov A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS A lock ( Lock A locked padlock ) or https:// means you’ve safely connected to the .gov website. Share sensitive information only on official, secure websites.

NCJRS Virtual Library

Community accountability, minority threat, and police brutality: an examination of civil rights criminal complaints, additional details, no download available, availability, related topics.

• Bipolar Disorder
• Therapy Center
• When To See a Therapist
• Types of Therapy
• Best Online Therapy
• Best Couples Therapy
• Best Family Therapy
• Managing Stress
• Sleep and Dreaming
• Understanding Emotions
• Self-Improvement
• Healthy Relationships
• Student Resources
• Personality Types
• Guided Meditations
• Verywell Mind Insights
• 2024 Verywell Mind 25
• Mental Health in the Classroom
• Editorial Process
• Meet Our Review Board
• Crisis Support

Understanding What Police Brutality Is and Why It Occurs

Arlin Cuncic, MA, is the author of The Anxiety Workbook and founder of the website About Social Anxiety. She has a Master's degree in clinical psychology.

Cara Lustik is a fact-checker and copywriter.

Verywell / Nusha Ashjaee

Why Police Brutality Occurs

Examples of police brutality, tyre nichols, breonna taylor, george floyd, dontre hamilton, eric garner, john crawford iii, why racism can turn to violence, how to reduce police brutality, trigger warning.

Information presented in this article may be triggering to some people as it describes various examples of police-related violence.

If you are in crisis, contact the  National Suicide Prevention Lifeline  at  988  for support and assistance from a trained counselor. If you or a loved one are in immediate danger, call 911.

For more mental health resources, see our  National Helpline Database .

Police brutality refers to the excessive use of force by a police officer against a victim or victims that is deemed to go beyond the level required to sustain life, avoid injury, or control a situation.

Most encounters with the police do not involve violence. A U.S. Department of Justice Report measured contact between police and the public in 2018.

Around 61.5 million people had an encounter with the police the year before the survey, but only 2% of people experienced threats or use of force. However, it's worth remembering that roughly half of the encounters in this survey were traffic-related incidents, and the report did not include police behavior during protests as a category.

In order to solve the problem of police brutality, it is necessary to understand the underlying factors that lead to it happening in the first place. In fact, there are a number of different factors that may play a role, not all of which have to do with the underlying personality of the officer who engages in the act.

However, each of them can be considered from a psychological standpoint or psychological lens. This helps us to understand how to fix the problem from a psychological view.

Individual-Level Factors

What are the individual-level factors that contribute to police brutality? These can be understood as those that originate from the offending officer. Some examples of individual-level factors are given below.

Mental Health Issues

The mental health of the offending officer may play a role. A 2019 study found that officers who self-reported engaging in abusive police practices tended to have higher levels of PTSD symptoms.

It is possible that officers with post-traumatic stress disorder (PTSD) from job-related stressors and trauma may have an increased startle response , a tendency toward suspicion, and problems with aggression. These traits can make it more likely that they will overreact and use deadly force when not necessary. However, it is also possible that engaging in excessive force results in a sense of profound guilt and moral injury that in turn lead to PTSD symptoms.

Some researchers theorize that traits of "psychopathy", also called antisocial personality disorder (APD) , may be more prevalent in police officers than the general population. Traits such as "fearless dominance" or "cold-heartedness" can be adaptive in dangerous or emotionally charged situations, but they can also make an individual more likely to engage in excessive use of force or to feel that they do not need to follow the rules.

That said, research on this theory is limited. It is unlikely that APD, which is very rare, could explain most police brutality cases.

Personal problems experienced by police officers may increase the likelihood of them engaging in excessive force, such as relationship problems or other stressful life events.

Organizational-Level Factors

What are the organizational-level factors that contribute to police brutality? These can include policies of the police department or the general working environment.

If the police department sets limits for the use of force that allows police officers to use their own discretion (in other words, limits that are too vague or lenient), then the likelihood that officers will use excessive force is going to increase.

In addition, if the general working environment of the police department is such that excessive use of force is not punished or reprimanded, then that sends the message to the police force that it's an acceptable part of their job description.

The Washington Post's police shootings database shows that police shoot and kill roughly 1,000 people a year in the United States. However, only 110 officers since 2005 have been charged with murder or manslaughter, and only 42 officers have been convicted.

In other words, the use of force becomes legitimized because everyone does it and nobody says anything about it.

This, despite the fact that if a civilian were to inflict the same level of force on another individual in the same situation, it would be considered to be a violation of the law. Due to qualified immunity, it can be difficult to prosecute officers for misconduct.

In order to understand the problem of police brutality, it is helpful to consider some of the more prominent examples in recent times. Below are some of the more well-known cases and issues surrounding them.

On January 7, 2023, 29-year-old Black man Tyre Nichols was pulled over in Memphis, TN, due to claims of reckless driving.

The five cops who stopped him, who were also Black, brutally beat him for about three minutes. As a result of his injuries, he died three days later.

The charges brought against the officers included second-degree murder and kidnapping.

After body-camera footage was released on January 27th, the public was outraged as many deemed it to be one of the most heinous acts of police violence ever witnessed.

Breonna Taylor was a 26-year-old Black woman who died after being shot in her apartment on March 13, 2020 in Louisville, Kentucky. Her death was the result of a search warrant that was being executed by white police officers from the Louisville Metro Police Department.

The raid began shortly after midnight. Her boyfriend, Kenneth Walker, thought the officers entering the apartment were intruders and fired a warning shot at them, which hit one officer in the leg. In return, the officers fired 32 shots, leaving Breonna Taylor dead and Walker physically unharmed.

While the City of Louisville agreed to pay $12 million to Taylor's family, the three police officers involved were not indicted on charges related to Taylor's death. The incident led to subsequent protests throughout the United States.

George Floyd was a 46-year-old Black man who died on May 25th, 2020 in Minneapolis, Minnesota after being arrested for using a counterfeit $20 bill. During the arrest, former police officer, now convicted murderer Derek Chauvin kept his knee on Floyd's neck while Floyd was handcuffed and lying on his face.

Bystanders who tried to intervene were prevented from doing so by other officers. Prior to his death, George Floyd pleaded for relief, saying that he could not breathe and that he was going to die. The entire incident became public when video footage shot by onlookers was released to the public. Autopsies revealed Floyd died as a result of the actions of the officers, and worldwide protests were sparked by the incident.

While these incidents occurred in 2020, police brutality has been a problem for decades. Below is a list of incidents from 2014, at the beginning of the Black Lives Matter movement that brought police brutality to the forefront of public discourse.

On April 30, 2014 Dontre Hamilton was killed after being shot 14 times by a police officer in a Milwaukee park. Local Starbucks employees had called the police for a wellness check after seeing Hamilton sleeping on a park bench. The officer who responded to the call, Christopher Manney, began what would later be described by the Police Chief Edward Flynn as an "inappropriate pat-down."

Hamilton woke up and began to struggle. Manney's defense team would later use Hamilton's prior diagnosis of schizophrenia to suggest that he was dangerous, but Flynn would later justify his firing of Manney by saying the officer ignored departmental policy and instigated the fight.

Eric Garner was killed on July 17, 2014 in New York after he was put in an illegal chokehold by a white police officer. Garner said "I can't breathe" 11 times while he was held down. The officer involved, Daniel Pantaleo, was not charged with a crime. His death sparked protests and "I can't breathe" as a slogan for protest.

John Crawford III was killed on August 5, 2014 after being shot by a police officer at a Walmart in Beavercreek, Ohio. He had been holding a pellet gun, which the store had advertised as being on sale, and there was no confrontation. The officers involved were not charged.

These are only some examples of how excessive use of force can lead to death.

Racism refers to bias held against a person or group of people because of their race or ethnicity. Why does racism turn into excessive use of force or violence among police officers? There are several factors to consider.

Prevalence of Deaths Due to Police Brutality

Research has demonstrated that the risk of being killed as a result of the use of excessive force by police in the United States varies by racial and ethnic group membership.

Specifically, Black men and women, American Indian/Alaska Native men and women, and Latin American men were shown to have a higher lifetime risk of dying due to police violence compared to their White counterparts.

In contrast, Latin American women and Asian/Pacific Islander men and women had a lower lifetime risk of dying due to police violence than White counterparts.

The overall lifetime odds were shown to be 1 in 2,000 for men and 1 in 33,000 for women. Overall, the highest risk was shown for Black men, who faced a 1 in 1,000 chance of being killed by a police officer over the course of their lifetime.

Racial Profiling

Why are Black men and other minorities at a higher risk for dying due to an excessive use of force by police than their White counterparts? Racial profiling may help to explain this phenomenon.

Racial profiling refers to assuming guilt based on race or ethnicity, a problem that mostly affects those individuals who have a higher lifetime risk of dying as a result of police brutality.

For example, police officers may use stereotypes when trying to determine the suspects in a crime, or they may perceive persons of certain races (such as Black men) as more aggressive or threatening when faced with a confrontation.

How can we work to reduce police brutality? There are a number of different steps that can be taken to reduce the risk of this phenomenon from an organizational and psychological standpoint.

In 2014, President Barack Obama signed an order to appoint a task force on 21st century policing. The task force developed a list of recommendations such as improving training and education, reducing bias among police officers and departments, introducing and improving crisis intervention training , and promoting cultural sensitivity as well as compassion.

Implicit Bias Training

Implicit bias training takes the approach that police officers operate with subconscious biases that they may not even be aware of. When these biases are activated, they may handle a situation differently than they would if, for example, a person was White instead of Black or driving a BMW instead of a old beat-up pickup truck.

The premise of this training is to help police officers understand that everyone grows up with subconscious biases, even if someone doesn't feel like they have any prejudice. The goal is to make police officers aware of their biases so that they can manage them in the moment. This is more effective than calling out police officers as racist, as most officers would not consider themselves to fall into that category. Rather, this approach takes the stance that all officers need training.

The idea behind implicit bias training is that those who are better able to manage their biases will be safer, more effective, and fairer in their role as police officers. However, there have been very few studies on the effectiveness of implicit bias training for police.

Only one 2020 study has looked at impacts on real-world behavior. While implicit bias training seemed to improve officer knowledge of implicit bias concepts and motivation to act without prejudice, the study found that training had little to no effect on racial and ethnic disparities in police enforcement. In other words, implicit bias training alone was not enough to change behavior.

Improved Hiring Practices

One way to reduce the risk of police brutality is to hire individuals who have a lower risk of becoming violent on the job.

Personality psychology can be helpful in making these decisions, as there are assessments that can be used to predict how individuals will respond to stressful situations as well as predict their behavior when on the job.

The use of personality assessments can also be a way to level the playing field for minorities, as it can be an unbiased way to determine who is the best fit for the job.

Improved Disciplinary & Supervision Measures

Suppose a police officer engages in excessive or deadly force, and there is no punishment. In that case, this sends the message to the rest of the department that the behavior is acceptable.

Instead, adequate supervision to identify police officers acting inappropriately before that behavior gets out of control, as well as disciplinary measures to send the message that the behavior is unacceptable, are necessary to identify and reprimand police officers who are the most likely to use excessive or deadly force.

Using such measures will also deter other officers from acting in the same manner and set the tone for the overall behavioral expectations of police officers in a department.

In other words, police departments should begin to lead by example, and that starts with enforcing the law for police officers in the same way that it would be for civilians.

Provide Mental Health Support for Police Officers

When police officers are better able to manage their emotions under stress, understand which emotions they are experiencing, and communicate well despite being in high-stress situations, they will be better able to de-escalate complex scenarios rather than react by using excessive force.

In other words, there is a tipping point at which excessive force begins to be used, and this tipping point can be dialed backward when police officers receive adequate support for their mental health needs.

Additionally, given the fact that PTSD can be a risk factor for the use of excessive or deadly force, providing swift and adequate support to officers who have experienced trauma on the job seems to be a necessary prerequisite to preventing the use of excessive force.

This begins by providing adequate funding to support the mental health of police officers, and it also means reducing stigma and encouraging police officers to come forward when they are struggling with their mental health.

As a society in general, mental health is still surrounded by stigma , so it is doubly important that police officers are made to feel that it is acceptable for them to talk about their mental health struggles. Rather than feeling isolated with their trauma, stress, or unmanageable emotions, police officers should be made to feel that they know exactly who to speak to for support and that those supports will be in place and easily accessible when they are most needed.

This also means the police departments should be trained to recognize the symptoms of PTSD so that they can intervene and offer support when an officer may not recognize their own symptoms of post-traumatic stress.

Improve Relationships Between Police & Community

To reduce the use of excessive and deadly force, it is important to improve the relationships between the police department and the community, particularly the Black community, as this sector is generally the one most affected by police brutality (and subsequent anxiety, depression, and post-traumatic stress).

This could take the form of programs and initiatives that place police officers in the community in a helping or educational role instead of a policing role. It could also mean having the police department work with the community or participate in marches and rallies to show their support and understanding. This was seen taking place when some police departments chose to attend Black Lives Matter protests and marches and kneel in support instead of taking a combative stance.

When police officers and the public begin to see each other as individuals rather than as groups to fear or cast stereotypes upon, real change will begin.

Conduct Research

In addition to the above measures, it is also necessary to continue to conduct research to understand the psychology behind police brutality. Which personality factors are most likely to correlate with excessive use of force? Which mental disorders show the highest correlation with deadly use of force? What forms of training help most to reduce implicit bias and improve the situation?

Ongoing research on these and other topics is the cornerstone of moving forward and improving the situation when it comes to the excessive use of force by police officers and the disproportionate impact that it has on racial minorities.

Defunding Police Departments

What about defunding police departments? This is a tactic that has been brought up as a solution to police brutality.

Defunding the police means taking money away from funding the police department and instead sending those funds to invest in the communities that are struggling the most and where most of the policing occurs.

It's very much similar to the concept of directing money toward prevention instead of dealing with problems after the fact. While not a simple solution, there is merit in funding programs and communities that are struggling instead of putting more people behind bars.

A Word From Verywell

Understanding the psychology behind police brutality is the first step toward fixing the problem. Unfortunately, the situation is inherently one that needs to be fixed from the top down, beginning with the government systems and how they allocate their funding. When better training and education are in place for police officers, as well as better mental health support, then better outcomes may result.

It's also worth noting that while this problem seems to be most prominent in the United States, other countries may have their own racial tensions (for example, in Canada and Australia, there is tension between the government and Indigenous people). The United States, however, struggles more than most with using deadly force in the form of gun violence.

Mental health support is available if you or someone you know has been affected by or witnessed police-related violence. Please reach out to a mental health professional . Acts of police brutality are traumatizing, and you deserve care, understanding, and support.

Amnesty International. Police violence .

U.S. Department of Justice. Contacts Between Police and the Public, 2018 .

DeVylder J, Lalane M, Fedina L. The association between abusive policing and PTSD symptoms among U.S. police officers . J Soc Soc Work Res . 2019;10(2):261-273. doi:10.1086/703356

Falkenbach D, Balash J, Tsoukalas M, Stern S, Lilienfeld SO. From theoretical to empirical: Considering reflections of psychopathy across the thin blue line . Personal Disord Theor Res Treat. 2018;9(5):420-428. doi:10.1037/per0000270

Thomson-DeVeaux A, Rakich N, Buchireddygari L. Why it's so rare for police officers to face legal consequences . FiveThirtyEight .

American Bar Association. Qualified immunity .

NPR. What we know about the killing of Tyre Nichols .

The New York Times. A timeline of Nichols's Lethal Police Encounter .

D'Amore R. Breonna Taylor: What we know about her death, the investigation and protests . Global News .

BBC News. George Floyd: What happened in the final moments of his life .

CBS News. Former Milwaukee officer not charged in fatal shooting of mentally ill man .

O'Kane C. Eric Garner's mom says seeing a black man plead "I can't breathe" is "like a reoccurring nightmare" . CBS News .

CBS News. Family sues over fatal shooting at Ohio Wal-Mart .

Edwards F, Lee H, Esposito M. Risk of being killed by police use of force in the United States by age, race-ethnicity, and sex .  PNAS. 2019;116(34):16793-16798. doi:10.1073/pnas.1821204116

Laurencin CT, Walker JM. Racial profiling is a public health and health disparities issue . J Racial Ethn Health Disparities . 2020;7:393-397. doi:10.1007/s40615-020-00738-2

President’s Task Force on 21st Century Policing. Final Report of the President's Task Force on 21st Century Policing .

Center for Police Research and Policy. The Impacts of Implicit Bias Awareness Training in the NYPD .

Williams DR. Stress and the Mental Health of Populations of Color:Advancing Our Understanding of Race-related Stressors .  J Health Soc Behav . 2018;59(4):466-485. doi:10.1177/0022146518814251

Johnson DK. Confirmation Bias and Police Brutality . Psychology Today .

Miller L. Why Cops Kill: The Psychology of Police Deadly Force Encounters . Aggression and Violent Behavior. 2015;22:97-111. doi:10.1016/j.avb.2015.04.007

Muller RT. Officers with PTSD at Greater Risk for Police Brutality . Psychology Today.

Sherman RA. The Problem of Police Brutality . Psychology Today .

Turner E. How can psychology advance police-community relations? Using psychological science and advocacy to contribute to solutions . American Psychological Association.

By Arlin Cuncic, MA Arlin Cuncic, MA, is the author of The Anxiety Workbook and founder of the website About Social Anxiety. She has a Master's degree in clinical psychology.

Follow U of T News

Race-based police violence impacts wealth of Black families, study finds

 (photo by Jane Tyska/Digital First Media/East Bay Times via Getty Images)

Published: May 21, 2024

By Diana Mehta

Financial decision-making for Black individuals can be dealt a major blow by race-based police violence, new research suggests, offering insight into the far-reaching effects of police brutality.

The study, titled “Race, Police Violence, and Financial Decision-Making,” examined detailed American data on home ownership and contributions to a pension plan – using statistics broken down by zip code – as well as information on fatal police encounters.

The analysis suggests police violence negatively influence financial decision-making for Black individuals, even when they are not directly involved in the incidents.

“We find that when a member of the Black community is killed in a police incident, the members of that racialized group in that local area experience changes in their financial decision-making that are not just statistically significant, but economically large,” says co-author Lisa Kramer , a professor of finance in the department of management at University of Toronto Mississauga and the Rotman School of Management. 

“I think what was most surprising was the magnitude of the effects.”

The study, which will appear in the May issue of  AEA Papers and Proceedings , was carried out by U of T’s Kramer, Duke University’s Vicki Bogan , University of Manitoba’s Chi Liao and University of Mannheim’s Alexandra Niessen-Ruenzi . It   explored whether two key pieces of the “wealth pie” for most individuals – home ownership and retirement savings – might be affected by race-based police violence. While many studies have already looked at the grief and community trauma associated with race-based police violence, Kramer says the ripple effects on economic decision-making have been analyzed to a far lesser extent.

The research showed Black individuals were 47.5 per cent less likely to own a home than their non-Black counterparts. After exposure to police-based violence, that gap increases to 62.2 per cent, the study suggests. It also found that Black individuals’ participation in defined contribution pension plans was reduced after exposure to police violence.

“We find already just to start with, just at the baseline, Black households are less likely to even own a home than others. And once they've observed one of these events in their local community, they become much less likely to own a home,” Kramer says.

Since researchers analyzed demographic, socio-economic and geographic data from U.S. households from 1999 to 2019, some recent key events – including the 2020 Minneapolis police killing of George Floyd that prompted anti-racism protests all over the world – were not captured.

“I think in this more recent era, where social media allows these events to get on the collective consciousness more quickly and more fully, we might find that any sort of traumatic implications that arise might even be more pervasive,” Kramer says.

And while the study is based on American data, Kramer says its findings likely apply in Canada as well.

“Certainly in Canada we have also had incidents of racialized violence with police involvement. We're not immune to that in Canada,” she says. “There's every reason to believe that the effects that we document aren't unique to the United States.”

The study doesn’t delve into why police-based violence could have such an effect on financial decision-making for Black individuals, but it does hint at possible explanations, including disengagement from financial decision-making after police violence in a local area and decisions to relocate following an incident.

Kramer, who notes that the study does not seek to deliberately cast police forces in a negative light, says she and her colleagues want to explore possible causes for their findings in future research.

“Right now, we’re identifying a striking set of results,” she says. “We find differences in financial outcomes at the local community level after these police-involved fatalities. Next, we need to identify the mechanisms that drive the results by testing different hypotheses.”

The hope is that the research will add to broader findings on racial inequalities and spark ideas about potential remedies to underlying problems.

“We're looking to explore those events through a financial lens because it's so important to make sure that households have the financial resources that they need,” Kramer says. “And if there is a connection there – as it appears there may be – we want to start the conversation, in a data-driven way.”

Share this page

Subscribe to The Bulletin Brief

More u of t news.

• Open access
• Published: 21 May 2024

Exploring police attitudes on victims’ delayed reporting and victim blame in technology-facilitated IPV

• Stavros Chatzisymeonidis   ORCID: orcid.org/0000-0002-7533-078X   nAff1 &
• Afroditi Pina   ORCID: orcid.org/0000-0001-8850-5625   nAff2  

Crime Science volume  13 , Article number:  12 ( 2024 ) Cite this article

146 Accesses

2 Altmetric

Metrics details

Background setting

Cyberstalking, now conceptualised as one of the forms of technology-facilitated intimate partner violence (TFIPV), has seen an exponential rise in recent years. TFIPV victims may hesitate and delay reporting cyberstalking to the police for various reasons (e.g., lack of recognition, thinking that it may be a waste of time, hoping it will stop etc.) and thus potentially influence how investigating police officers perceive their credibility and responsibility. This study investigates the recognition of cyberstalking as a crime among police personnel and the potential effect of reporting delays on police officers’ attitudes towards the victims.

An online survey was conducted with 108 police officers in the UK, who were presented with a vignette illustrating one of three almost identical scenarios, differing only in the time of reporting (after one month, after six months, after 12 months). Subsequently, participants completed a questionnaire that assessed their recognition of the case as cyberstalking and their attitudes towards victims. All police officers had received predetermined police training at various levels. In addition to these police training programmes, a minority of officers (27) had attended the specialised training programme on intimate partner violence, Domestic Abuse (DA) Matters, while the majority (81) had not.

Among the officers who completed the aforementioned special training, all except one recognised the case as cyberstalking; contrastingly, out of 81 officers without such special training 28 expressed uncertainty, whereas three did not recognise it at all. The victim’s delay to report cyberstalking had a significant effect on police officers’ victim blaming levels. The gender of police officers and their police training level were not identified as moderators of the relationship between victim’s delay in cyberstalking reporting and victim blaming.

Conclusions

These findings highlight the importance for enhanced recognition and understanding of cyberstalking among police officers, particularly through specialised training programs. The study underscores the importance of addressing attitudes towards victims with the goal of improving police responses to TFIPV.

Introduction

Intimate partner violence (IPV) remains a prevalent and persistent problem globally, despite efforts in health promotion to raise awareness (Power et al., 2006 ). The World Health Organization (WHO) defines IPV as “behaviour within an intimate relationship that causes physical, sexual or psychological harm, including acts of physical aggression, sexual coercion, psychological abuse and controlling behaviours” (WHO, 2010 , p.11). Simultaneously, the advancement of technology has provided perpetrators with new means to monitor, stalk, control and threaten their partners in online environments (Freed et al., 2017 ; Harris & Woodlock, 2019 ; Woodlock, 2016 ), resulting in the emergence of a phenomenon known as technology facilitated IPV (TFIPV) (Leitão, 2019 ; Pina et al., 2021a ).

Scholars argue that TFIPV is an extension of IPV and should be classified within typologies that align with offline forms of abuse, including sexual, verbal, and physical violence (Pina et al., 2021a ). TFIPV encompasses a range of behaviours, either overt or covert, occurring in public or private settings, with the primary objective of exerting control over the victim, violating their privacy, or damaging their reputation. These behaviours may manifest individually or concurrently (e.g., monitoring activity and sending harassing messages (Freed et al., 2019 ). Based on these behavioural patterns, research identifies four main types of TFIPV: cyberstalking, online harassment, image-based sexual abuse and indirect non-sexual abuse (Dardis & Gidycz, 2019 ; Huber, 2023 ; Lopez-Cepero et al., 2018 ; Martellozzo et al., 2022 ; Pina et al., 2021a ; Watkins et al., 2018 ).

Although all types of TFIPV are of paramount importance, recent emphasis has been given to cyberstalking incidents involving current or ex-intimate partners. This may be primarily attributed to a surge of 300% in reported cases of cyberstalking in recent years, as documented by the Suzy Lamplugh Trust ( 2021 ). Similarly, Bracewell et al. ( 2020 ) found a significant increment in cyberstalking during the first Covid-19 lockdown in the UK. According to the National Centre for Cyberstalking Research in the UK, cyberstalking is defined as “a course of action that involves more than one incident perpetrated through or utilizing electronic means, that causes distress, fear or alarm” (Maple et al., 2011 , p.4). This definition aligns with provisions outlined in the UK Protection of Freedoms Act 2012, which places a series of acts or omissions referring to cyberstalking (e.g. monitoring the use by a person of the internet, email, or any other form of electronic communication) in the Protection from Harassment Act 1997 regulating stalking offences. These involve behaviours that threaten, control, insult or harass the victim (Dardis & Gidycz, 2019 ; Henry et al.,  2018 ; 2020 ; Reyns et al., 2012 ; Strawhun et al., 2013 ). Such behaviours have been found to have detrimental effects on the victims’ well-being (Dreßing et al., 2014 ) and are often associated with conflicts, aggression (Kellerman et al., 2013 ; Schnurr et al., 2013 ) and domestic homicides (Todd et al., 2021 ).

Notably, cyberstalking primarily affects female victims (Begotti et al., 2020 ; Dreßing et al., 2014 ; Sheridan & Grant, 2007 ), who employ various coping strategies in response to such incidents (Amar & Alexy, 2009 ). Some victims opt to modify their daily routines and engage in uncommon activities, such as reducing internet usage, to evade the perpetrators, while others adopt a passive approach and ignore cyberstalkers (Begotti et al., 2022 ). Nonetheless, most cyberstalking victims adopt proactive tactics to address these incidents, seeking informal support from trusted individuals or formal support by reporting the case to social networking site administrators or the police (Alexy et al., 2005 ; Fissel, 2021a , b ; Reno, 1999 ).

Despite seeking formal support and reporting cyberstalking cases to the police, many TFIPV victims have expressed disappointment with the support provided by police personnel (Leukfeldt et al., 2019 ; Worsley et al., 2017 ). According to Leukfeldt et al. ( 2013 ), such dissatisfaction primarily arises from the limited understanding and handling of online offences by the police. For example, the report of Storry and Poppleton ( 2022 ) brought to light that many cyberstalking victims feel their claims are trivialised or disbelieved by the police (with some accusing officers of lacking empathy during the reporting process), leading to traumatic experiences. These findings are consistent with Martellozzo et al. ( 2022 ), who suggested that the inadequate management of cyberstalking incidents by the police may result from the disparity between new modes of online abuse and the traditional tactics employed by police officers to address offline abuse.

With respect to crime reporting, prior literature has indicated a strong relationship between the perceived severity of offense and victims’ decision to report (Cass & Mallicoat, 2015 ; Clay-Warner & McMahon-Howard, 2009 ; Fisher et al., 2003 ; Saxton et al., 2020 ). This relationship is consistent with pioneering work by Gottfredson and Hindelang ( 1979 ), who analysed data from the USA National Crime Survey and found that the severity of offence constitutes a primary factor in influencing the victim’s decision to seek police support. However, many victims often hesitate and delay reporting severe offences like cyberstalking to the police for various reasons, such as the fear of harassment continuation (Rosalili et al., 2021 ), the stalkers’ anonymity (Goodno, 2007 ), the lack of trust to the authorities (Martellozzo et al., 2022 ), as well as the negative responses of the police (Powell & Henry, 2018 ). Delayed reporting may complicate the task of evidence collection for police officers, potentially affecting their attitudes toward victims. Similar to jurors’ perceptions in severe crimes such as sexual assaults (Ellison & Munro, 2009 ), officers may develop doubts regarding victims’ responsibility and credibility. However, it remains unknown whether the delays in reporting cyberstalking cases between intimate partners influence the police officers’ attitudes towards victims’ responsibility.

Prior studies have explored the subject of police officers’ victim blaming from different angles, particularly in the context of online abuse. For instance, research (Millman et al., 2017 ; Zvi & Shechory-Bitton, 2020 ) has revealed that officers tend to hold victims (at least to an extent) accountable for their ordeal, arguing that victims are failing to undertake necessary precautions to protect themselves online, or self-disclose information online that can put them in danger, and as a result, these behaviours may undermine police investigations. Similarly, Huber ( 2023 ) identified victim blaming attitudes among officers who underestimated the severity of online abuse and failed to acknowledge victims’ needs during investigations. According to Chang ( 2020 ), the underestimation of cyberstalking incidents might be attributed to gaps in police training, particularly in recognising the nature and severity of such cases.

After officers have joined the force in the UK, they receive police training at various levels in order to effectively investigate offline and online crimes (Bryant & Bryant, 2015 ). This includes 11 standard police training programmes divided into three main levels (basic, moderate, high), which can be seen in Table  1 .

Additionally, to the above predetermined police training, the charity SafeLives on behalf of The College of Policing developed a specialised training programme, named “DA Matters”, after the report of HM Inspectorate of Constabulary in 2014 stressing the need for improvements in how the police respond to domestic violence (HMIC, 2014). The primary objective of this victim-centered training programme is to have a long-lasting effect on the police culture and behaviours, enabling officers to improve their response to domestic violence incidents (including cyberstalking between intimate partners). Although the evaluation results were promising in 2019 (e.g. increased arrests due to coercive control between intimate partners; DA Matters, 2019), there is no further evidence regarding its long-term impact. Currently, 26 out of 43 police forces in England and Wales have been licensed to deliver this specialised training to their front-line police personnel. Approximately 75% of the licensee police personnel have completed this specialised training (College of Policing, 2022 ).

Existing research highlights that the propensity for victim blaming is influenced by observer-specific traits, such as gender (Grubb & Harrower, 2009 ). Specifically, several studies have found that male observers tend to display more severe victim blaming attitudes concerning crimes committed in offline environments, such as rapes and sexual assaults, in comparison to female observers (Kanekar et al., 1985 ; Grubb & Harrower, 2009 ; Grubb & Turner, 2012 ; Pinciotti & Orcutt, 2017 ). Some scholars have argued that this may also be the case for crimes in online environments such as non-consensual dissemination of intimate images, commonly known as ‘revenge pornography’ (Bothamley & Tully, 2018 ; Zvi, 2022a ; Zvi & Shechory-Bitton, 2020 ). Nonetheless, it is essential to note that some studies have reported either contradictory findings to this trend (Cameron & Stritzke, 2003 ), or no difference between male and female observers in victim blaming attributions (Davies et al., 2009 ; Cassidy & Hurrell, 1995 ; Yarmey, 1985 ).

In light of the aforementioned considerations, this study aims to investigate police officers’ levels of engagement in victim blaming in cyberstalking between intimate partners, and recognition of a case as cyberstalking, focusing on the impact of the DA Matters specialised training programme, victim delay to report, gender of officers and level of police training. The following hypotheses will guide this investigation:

Ha. It is expected that the police officers who have completed the DA Matters special training programme will demonstrate higher levels of cyberstalking recognition ratings compared to their colleagues without such specialised training. Hb. Delayed cyberstalking reporting is expected to have an effect on police officers’ levels of victim-blaming. Hc. The relationship between victim’s delay to report cyberstalking and victim blaming will be moderated by the gender of officers. Hd. The relationship between victim’s delay to report cyberstalking and victim blaming will be moderated by the level of police training officers have received.

By examining these hypotheses, this study aims to contribute to a better understanding of how police officers perceive cyberstalking incidents between intimate partners, ultimately informing the development of more effective police responses for victims of TFIPV.

Methodology

Participants.

The current survey was conducted between mid-February and early July 2023. Participants were current police officers from various police forces across the UK, regardless of their rank and years of service. Police officers for whom a misconduct case was pending against them were not included. Several power analyses were conducted using G*Power Software version 3.1.9.7 (Erdfelder et al., 1996 ; Faul et al., 2007 ), to determine the sample size of participants required to examine the study’s hypotheses. While we had multiple key statistical analyses with different tests, each a-priori power analysis is reported close to the corresponding type of statistical analysis in the Results section, following the recommendations by Giner-Sorolla et al. ( 2024 ).

Next, contact was made with several police forces via their head offices, requesting the distribution of an electronic survey link to their police personnel through their official emails. As the police are a complex population to reach with multiple demands on their time, snowball sampling was also utilised to identify additional police officers individually within the community. The participation in the survey was voluntary, and identifiable details were not collected to ensure participants’ anonymity. In total, 108 police officers participated in the survey, comprising 51 (47.2%) males and 57 (52.8%) females. For the practical purposes of this study, the 11 predetermined police training programmes were further classified into three main levels based on police rank (basic = unspecified staff & police constables, moderate = sergeants to inspectors, high = inspectors to superintendents). For a comprehensive overview of the main characteristics of male and female police officers who took part in this study, please see Table  2 .

Research design and procedure

Initially, participants accessed the research materials through a Qualtrics link, where they were presented with a brief research description, followed by a consent form. Upon providing consent, officers were requested to complete a form capturing basic characteristics including gender, age, police rank, years of service, level of police training (i.e. COP), and whether they had received specialised training in IPV (Yes vs. No).

This study employed a between-subjects design, through which each participant was randomly assigned to one of three nearly identical vignette conditions that depicted a cyberstalking case between intimate partners after completing the sociodemographic form. The sole distinction between the vignettes was the time frame (one month, six months, 12 months) after which the victim decided to report the cyberstalking case to the police against their intimate partner. The vignettes were accompanied by a concise questionnaire exploring two parts: (a) whether the case was recognised as a cyberstalking crime or not and (b) the attitudes of police officers concerning the victim’s responsibility in the case. Finally, all participants were fully debriefed.

Cyberstalking vignette

Each participant read one of three almost identical cyberstalking vignettes specifically created for this study, featuring a victim named “Emma”, who was subjected to cyberstalking by her intimate partner, named “Sam”. It is worth noting that the storyline was constructed based on true facts to mirror a realistic case handled by police officers. The scenarios differed solely in the time of reporting (after one month, after six months, after 12 months) and an example of the prototype was as follows:

Sam (33) and Emma (24) met for the first time when the former wanted to buy a flat and the latter was working as a real estate agent. After a couple of dates, they started a romantic relationship. However, Emma was still thinking about her ex-boyfriend, and she asked Sam to take a break for a while. She wanted to take some time to be with herself and clear her mind regarding her past. Although Sam accepted Emma’s position, he was devastated. He became jealous and distressed when he started watching her activity on Instagram (stories and photos) and joining the same pages. During the week that followed, Sam sent about 20 messages to Emma asking her to meet at least one more time. Although Emma left his messages on read many times, she finally accepted to meet him close to her work. She explained to him that she was not ready to move forward with something serious and wanted some time to be with herself. Sam told Emma he only wanted to help her overcome her past, but he would never bother her again. However, he was still watching her Instagram activity and was desperate while realizing that she had distanced herself from him. Sam, after that meeting, sent Emma about 20 short video messages communicating his sadness at watching her having a great time with friends. He also sent her emails saying that he could no longer live without her and was considering harming himself or taking his life if she did not give him a chance to be with her. After one month had elapsed, Emma decided to report the case to the police.

Cyberstalking recognition and attitudes towards cyberstalking victims

Upon reading the scenario, all participants were required to answer a question indicating whether they recognised the dispute case between intimate partners as cyberstalking or not. Subsequently, participants completed a victim blaming scale to evaluate the extent to which Emma was responsible for the cyberstalking perpetrated by Sam, her intimate partner. The victim blaming scale comprised six items adapted from previous research (Grubb & Harrower, 2009 ), modified to align with the purposes of the current study. Participants rated their agreement on a 5-point scale ranging from 1 (completely disagree) to 5 (completely agree), in response statements mentioned in Table  3 . A victim blaming score was computed by calculating the sum of participants’ ratings, with higher scores indicative of a greater degree of victim blame. Given the presence of unequal factor loadings, the reliability of the scale was computed using the coefficient omega (McDonald, 1999 ; Hayes & Coutts, 2020 ), indicating a satisfactory level ( ω  = 0.78).

The study obtained approval from the Institutional Psychology Research Ethics Committee (REC) under Ethics ID: 202316750142348252 and adhered to the UK Data Protection Act 2018. All participants were informed that their participation was voluntary and confidential. To ensure anonymity, police officers were requested to assign a unique 6-digit personal identification number (PIN), which could be used in the event that a participant wished to withdraw their data at a later stage.

Data preparation

Data were analysed employing R Software version 4.3.1. All statistical tests were two-tailed, and the chosen level of statistical significance was set at 0.05. The sampling distribution of the total sample ( N  = 108) was assumed to be normally distributed based on the Central Limit Theorem since the condition of sample ( N  ≥ 30) is met (Anderson, 2010 ). The percentage of missing values for the selected variables were less than 10% in total, and therefore did not constitute biases in estimates (Dong & Peng, 2013 ). The use of boxplots and Cook’s distance estimates did not reveal any univariate or multivariate outliers, respectively (Aguinis et al., 2013 ; Leys et al., 2019 ). The models’ results were not influenced by skewness, as the condition of sample ( N  ≥ 10) was met (Schmidt & Finan, 2018 ). Means and standard deviations can be found in Table  1 .

Recognition of cyberstalking in relation to the DA matters special training

A chi-square test of independence was performed to investigate whether there is a relationship between the DA Matters special training programme and the recognition of the case as cyberstalking. An a-priori power analysis for sample size estimation was conducted and revealed that a total sample of N  = 108 participants would be needed to detect by convention a medium effect Cohen’s w  = 0.30, with 80% power using a chi-square hypothesis test of this design with alpha = 0.05. The analysis revealed a significant association between these two variables, X 2 (2, N =  108) = 11.62, p =  .003, with a medium effect size Cohen’s w  = 0.33. Based on the aforementioned observed medium effect size, an a-priori power analysis was additionally conducted to estimate the sample size needed to replicate the same effect. The analysis established that 95 participants would be adequate to achieve 80% power given w  = 0.33 and alpha = 0.05. Among officers who completed the aforementioned special training, the vast majority (96.3%) recognised the case between intimate partners as cyberstalking, with only one officer responding “do not know” (3.7%). Similarly, among officers without such special training, a majority (61.7%) recognised the case as cyberstalking, followed by those who were uncertain (34.6%), while a tiny minority did not recognise the scenario as cyberstalking (3.7%). (See Table  4 ).

Victim blaming in relation to delayed cyberstalking reporting

A one-way ANOVA/GLM model was employed to examine our second hypothesis. Our a-priori power analysis for sample size estimation returned that a total sample of N  = 156 participants would be required to detect by convention a medium effect partial eta-squared  = 0.06 with 80% power using a one-way ANOVA hypothesis test of this design with alpha = 0.05. Considering that only N  = 108 participants took part in the study the results below should be interpreted with caution. Our model revealed a significant effect of delayed cyberstalking reporting on police officers’ victim blaming levels, F (2, 105) = 3.13, p  = .048, eta-squared  = 0.06. In particular, police officers indicated significantly greater victim blaming levels towards the scenario where the victim decided to report the case after 12 months ( M  = 16.08, SD  = 3.05) than when the victim reported the case after one month ( M  = 14.55, SD  = 2.92) ( p  = .027). However, there was no significant difference on police officers’ victim blaming levels between the victim who decided to report the case after one month and after six months ( p  = .935). (See Table  5 for contrasts p -values).

Moderation analyses

Two linear regression models with interaction terms (moderations) was conducted to examine our next two hypotheses. An a-priori power analysis indicated that a total sample of N  = 82 participants would be necessary to detect by convention a medium effect partial R 2  = 0.09 with 80% power using a moderation hypothesis test of this design with alpha = 0.05.

Main effects and interaction of police officers’ gender and victim’s delay to report cyberstalking on police officers’ victim blaming

A moderation model was conducted to examine whether the gender of officers moderates the relationship between the victim’s delay in reporting cyberstalking and victim blaming. The main effects from a separate linear regression model included the victim’s delay to report the cyberstalking case, B  = 0.72, 95% CI (0.03, 1.41), p  = .042 and officers’ gender B  = 0.18, 95% CI (-0.95, 1.31), p  = .748. However, only victim’s delay to report cyberstalking emerged as a significant predictor of police officers’ victim blaming levels. The interaction of these two variables was not found to be significant, B  = 0.13, 95% CI (-1.26, 1.53), p  = .852. The final model did not reach statistical significance, indicating a small to medium effect size, R 2  = 0.04, F (3, 104) = 1.61, p  = .094 (See Table  6 ). Thus, the results did not identify the gender of officers as a moderator of the relationship between victim’s delay to report cyberstalking and victim blaming. Nonetheless, considering the observed small-to-medium effect size of the final model, an a-priori power analysis was additionally utilised to estimate the sample size required to replicate this effect. The analysis returned that a sample of N  = 191 participants would be sufficient to achieve 80% power given R 2  = 0.04 with alpha = 0.05.

Main effects and interaction of police training and victim’s delay to report cyberstalking on police officers’ victim blaming

A second moderation model was conducted to examine whether the predetermined police training moderates the relationship between victim’s delay to report cyberstalking and victim blaming. The main effects from a separate linear regression model involved the victim’s delay to report the cyberstalking case B  = 0.64, 95% CI (0.00, 1.29), p  = .050 and police training B = -1.45, 95% CI (-2.27, -0.62), p  = .001. Victim’s delay to report cyberstalking positively, and police training negatively, predicted officers’ victim blaming. However, the interaction of these two variables was not significant, B  = 0.34, 95% CI (-0.68, 1.36), p  = .513. The final model was statistically significant, displaying a medium to large effect size, R 2  = 0.15, F (3, 104) = 5.94, p  = .001 (See Table  7 ). Consequently, the results did not identify the predetermined police training as a moderator of the relationship between victim’s delay to report cyberstalking and victim blaming. Similarly to the rationale of our first moderation hypothesis test, an a-priori power analysis was also conducted to estimate the sample size needed to replicate this effect. The analysis revealed that a sample of N  = 47 participant would be required to achieve 80% power given R 2  = 0.15 with alpha = 0.05.

The phenomenon of TFIPV has been primarily examined through the accounts of victims and support workers (Clevenger & Navarro, 2021 ; Flynn et al., 2023 ; Pina et al., 2021a ; Short et al., 2015 ). Although the police have a key role in combating cybercrimes (Wall & Williams, 2013 ), and are expected to act with fairness and impartiality towards victims to ensure their support and the provision of compelling evidence during investigations (Bryant & Bryant, 2015 ), there is a paucity of evidence regarding police responses to TFIPV. Generally, victims’ satisfaction with police responses to disputes within intimate partner relationships ranges from 60 to 80% (Buzawa & Buzawa, 2003 ; Lewis et al., 2000 ; Peirone et al., 2021 ; Statistics Canada, 2016 ; Stephens & Sinden, 2000 ). Nevertheless, there are some IPV victims dissatisfied with the way they are treated by the officers (Gillis et al., 2006 ; Stewart et al., 2013 ), including victims of online abuse incidents such as cyberstalking (Sheridan & Grant, 2007 ; Short et al., 2014 ; Stevens et al., 2021 ; Storry & Poppleton, 2022 ). Therefore, this study attempted to explore the perceptions of police officers towards victims who are cyberstalked by their intimate partners and to examine specific factors may influence these perceptions (and ultimately their responses to these incidents).

As hypothesised, the proportion of police officers who recognised the case as cyberstalking was considerably higher for those who had completed the DA Matters special training programme than those without such training. A notable number of officers without such special training either expressed uncertainty about the classification of the case as cyberstalking or did not recognise it at all, whereas only one officer with the special training was uncertain. These findings lend support to the literature, which emphasises the critical need for special police training that focuses on the recognition and effective management of crimes related to online abuse (Chang, 2020 ; Powell & Henry, 2018 ; Harris & Woodlock, 2019 ; Flynn et al., 2021 , 2023 ). Given the results above, it becomes evident that providing special training to officers on IPV, including TFIPV forms, may strengthen their skills and knowledge around the new modes of abuse between intimate partners in online environments.

The present research aimed to enrich the literature by approaching the issue of victim-blaming from a practical perspective related to the critical task of police investigations and collecting evidence, in light of victims’ delayed reporting. The results supported the second hypothesis, indicating that the delayed cyberstalking reporting had a significant effect on officers’ victim blaming. Police officers demonstrated significantly higher tendency for victim blaming, when the victim delayed reporting for more than 12 months, compared to the victim delayed reporting after one month. Surprisingly, the same tendency for victim blaming was not significant when the victim delayed reporting for more than six months compared to the victim delayed reporting after one month. This inconsistency may have arisen due to lack of statistical power (Lorah, 2020 ), highlighting the need for further research with a larger sample size to establish reliable conclusions. Possibly, the increment in reporting delay after a period of time may considerably increase suspicions and lead to negative perceptions among police personnel regarding the victim’s credibility and responsibility. The passage of time may create an impression that the victim’s experience might not have been severe enough to report promptly, potentially influencing officers’ perceptions of the validity of the victim’s allegations and leading to blame attribution towards the victim for not acting sooner. This finding is consistent with previous studies that examined jurors’ perceptions, showing that victims who delay reporting crimes in offline environments, such as sexual assaults, are viewed less favourably compared to those who report immediately (Franiuk et al., 2019 ; Thompson et al., 2021 ).

Given the observed link between victim’s delay in reporting cyberstalking and heightened officers’ victim-blaming, coupled with the established role of gender in judgements towards victims, this study investigated whether officers’ gender moderates their victim-blaming attitudes towards victims who delay reporting cyberstalking. Although this study found that female police officers exhibited slightly higher levels of victim blaming compared to male police officers, this difference was not significant. A separate regression model showed that victim’s delay to report cyberstalking was a positive predictor of officers’ victim-blaming attitudes, whereas the gender of officers - as observers- was not. However, the final multivariate regression model with interaction terms revealed that the gender of officers did not moderate the relationship between the victim’s delay to report cyberstalking and officers’ victim-blaming attitudes. An explanation might be the role of perpetrator’s gender, which was clearly a male perpetrator in the vignettes. This could have guided the participants’ similar evaluations of victim blaming and explain the present findings (Lee, 2019 ). Furthermore, traditional gender-based stereotypes that suggest male police officers are more likely to exhibit victim-blaming attitudes towards rape victims (Zvi, 2022b ) or ‘revenge pornography’ victims (Zvi & Shechory Bitton, 2020), as mentioned above, may not hold true for all types of victims within policing contexts. The present study’s findings contrast those of previous research, which has found that the male observers display higher victim-blaming attitudes regarding crimes committed offline, like rape and sexual assault (Davies & McCartney, 2003 ; Davies et al., 2011 , 2012 ; Davies & Hudson, 2011 ; Donovan, 2007 ; Howard, 1984; Schneider et al. 2009 ; White & Kurpius, 2002 ). This discrepancy may support the notion that victim blaming attitudes between male and female officers could be subject to the online/offline nature and/or environment of a crime. Our findings are in line with Davies et al. ( 2009 ) and Schneider et al. ( 1994 ), who did not find evidence of a difference between observers’ gender in victim blaming of offline crimes, as well as with those of Pina et al. ( 2021b ), who found no differences in victim blaming towards victims of image based sexual abuse. Nevertheless, we must interpret our results with caution as our sample was underpowered.

An additional goal of this study was to shed light on the role of predetermined police training when interacting with victims’ delay in reporting a cyberstalking case, on the officers’ victim-blaming levels. A separate multivariate regression model initially indicated that victim’s delay to report cyberstalking and police training were positive and negative predictors, respectively, of officers’ victim-blaming attitudes. This finding aligns with the scholars who have concluded that police training may play a key role in addressing critical barriers such as the display of masculinity and victim-blaming attitudes by police officers during investigations (Blaney, 2010 ; Holt & Bossler, 2012 ; Lopez-Neria et al., 2019 ). Contrary to what was expected, the final multivariate regression model with interaction terms did not detect a moderation effect of police training between victim’s delay to report cyberstalking and victim-blaming. It may be possible that the predetermined police training programmes do not sufficiently address specific aspects related to victim-blaming in TFIPV cases that include cyberstalking or delays in reporting. Existing training may have concentrated on more general issues of IPV and cybercrime rather than addressing officers’ potential biases that contribute to victim-blaming.

Despite providing a novel insight into police officers’ attitudes towards cyberstalking victims who delay reporting, this study had its limitations that may have influenced the findings. The participants were requested to provide answers regarding a sensitive topic, utilizing a victim-blaming scale. Therefore, the findings could appertain to socially desirable responses as the officers may have been consciously aware of the subject matter or sought to present themselves in a favorable light. It is suggested that future studies adopt short social desirability measures to control for this confound (Greenwald & Satow, 1970 ). The present study assigned a gender to the perpetrator of the scenario in order to be as similar as possible to real cases handled by police officers; nonetheless, this approach does not allow for examination of other variables that may influence officers’ responses (i.e. homophobia) pertaining to the role of gender in victim blame (Lee, 2019 ). Another limitation concerns the representativeness of the information, as the officers’ responses were not collected from all police forces in the UK. The findings may not fully capture the perceptions of officers from other regions or jurisdictions not taking part, thereby limiting generalisability. Due to time constraints, data collection was based on self-selection and snowball (non-probability) sampling, which could lead to an imbalanced representation of police officers. As such, the results may not fully capture the perspectives of each UK police force took part in the study. For the same reason, the study had a smaller than expected sample size with regards to the examination of our second hypothesis based on the power analysis outlined in the sub-section of Victim Blaming in Relation to Delayed Cyberstalking Reporting. Consequently, insufficient statistical power may have increased the risk of false negatives. Future research is encouraged to replicate these findings, with a larger sample size to shed more light in the relationship between reporting delay and victim blaming with more representative and powerful results.

Overall, understanding the influence of digital technologies on the emergence of new forms of IPV in online environments, like cyberstalking, and how victims behave in such circumstances are critical issues directly related to the effectiveness of police responses. The findings of this study clearly demonstrate the necessity of specialised training sessions for police personnel on IPV, with a specific focus on methods that facilitate the recognition of cyberstalking. Insufficient knowledge in this area is likely to precipitate the strong dissatisfaction of victims when reporting TFIPV cases to the police and consequently lead to higher rates of unreported online abuse within relationships. Additionally, it is recommended that police forces incorporate training sessions addressing the attitudes of officers towards the responsibility and credibility of victims, considering the present findings that revealed a high tendency of victim blaming in relation to reporting delays.

Future research could conceptually replicate the methodology of this study and investigate further sociodemographic and psychosocial factors that might moderate the relationship between victims’ delay in reporting a cyberstalking case and officers’ victim-blaming levels. As the matter of victim-blaming and risk minimisation attitudes within police culture is convoluted when it comes to online abuse, alternative variables might be also employed to examine this topic in relation to victim’s vulnerability and inconsistency during investigations. For example, investigating the relationship between recurring complaints for the same TFIPV case and officers’ victim-blaming levels may be promising for further nuanced understanding of case handling. Understanding how repeated reporting and withdrawal of charges by victims influence officers’ attitudes could provide insights into the challenges faced by victims and the dynamics of police responses.

In conclusion, the attitudes and beliefs of police officers towards TFIPV victims remain an understudied area, despite significant calls for examination from both support service stakeholders and victims in previous studies. Nevertheless, we have some supportive data showing that specialist training improves recognition and decreases victim blaming that solidify the necessity for specialist training across different police ranks. Further investigation is necessary to expand our current knowledge and develop more targeted police training in this field. This in-depth exploration will provide empirical evidence that could enable police force educators to develop well-informed and victim-centered training programmes, aiming to substantially improve the police response in TFIPV crimes.

Data availability

Data sharing is not applicable to this article due to confidentiality agreements with the participant police forces.

Availability of analytic code

The R code is available in the Open Science Framework repository, https://osf.io/q86pg/?view_only=e49ac41a27f24f3d882c59c1a635103b .

Aguinis, H., Gottfredson, R. K., & Joo, H. (2013). Best-practice recommendations for defining, identifying, and handling outliers. Organizational Research Methods , 16 (2), 270–301. https://doi.org/10.1177/1094428112470848 .

Article   Google Scholar  

Alexy, E. M., Burgess, A. W., Baker, T., & Smoyak, S. A. (2005). Perceptions of Cyberstalking among College Students. Brief Treatment and Crisis Intervention , 5 (3), 279–289. https://doi.org/10.1093/brief-treatment/mhi020 .

Amar, A. F., & Alexy, E. M. (2009). Coping with stalking. Issues in Mental Health Nursing , 31 (1), 8–14. https://doi.org/10.3109/01612840903225602 .

Anderson, C. J. (2010). Central Limit Theorem. The Corsini Encyclopedia of Psychology. https://doi.org/10.1002/9780470479216.corpsy0160 .

Begotti, T., Bollo, M., & Acquadro Maran, D. (2020). Coping strategies and anxiety and depressive symptoms in young adult victims of Cyberstalking: A Questionnaire Survey in an Italian sample. Future Internet , 12 (8), 136. https://doi.org/10.3390/fi12080136 .

Begotti, T., Ghigo, M. A., & Acquadro Maran, D. (2022). Victims of known and unknown cyberstalkers: A Questionnaire Survey in an Italian sample. International Journal of Environmental Research and Public Health , 19 (8), 4883. https://doi.org/10.3390/ijerph19084883 .

Blaney, E. (2010). Police officers’ views of specialized intimate partner violence training. Policing: An International Journal of Police Strategies & Management , 33 (2), 354–375. https://doi.org/10.1108/13639511011044939 .

Bothamley, S., & Tully, R. J. (2018). Understanding revenge pornography: Public perceptions of revenge pornography and victim blaming. Journal of Aggression Conflict and Peace Research , 10 (1), 1–10. https://doi.org/10.1108/jacpr-09-2016-0253 .

Bracewell, K., Hargreaves, P., & Stanley, N. (2020). The consequences of the COVID-19 lockdown on stalking victimisation. Journal of Family Violence , 37 . https://doi.org/10.1007/s10896-020-00201-0 .

Bryant, R., & Bryant, S. (2015). Blackstone’s handbook for policing students 2015 . Oxford University Press.

Buzawa, E. S., & Buzawa, C. G. (2003). Domestic violence: The criminal justice response (3rd ed.). SAGE.

Cameron, C. A., & Stritzke, W. G. K. (2003). Alcohol and acquaintance rape in Australia: Testing the Presupposition Model of attributions about responsibility and Blame1. Journal of Applied Social Psychology , 33 (5), 983–1008. https://doi.org/10.1111/j.1559-1816.2003.tb01935.x .

Cass, A. I., & Mallicoat, S. L. (2015). College student perceptions of victim action: Will targets of stalking report to police? American Journal of Criminal Justice , 40 , 250–269. https://doi.org/10.1007/s12103-014-9252-8 .

Cassidy, L., & Hurrell, R. M. (1995). The influence of victim’s attire on adolescents’ judgments of date rape. Adolescence , 30 (118), 319–323.

Google Scholar  

Chang, W. J. (2020). Cyberstalking and Law Enforcement. Procedia Computer Science , 176 , 1188–1194. https://doi.org/10.1016/j.procs.2020.09.115 .

Clay-Warner, J., & McMahon-Howard, J. (2009). Rape reporting: Classic rape and the Behavior of Law. Violence and Victims , 24 (6), 723–743. https://doi.org/10.1891/0886-6708.24.6.723 .

Clevenger, S., & Navarro, J. (2021). The Third-Victimization: The cybervictimization of sexual assault survivors and their families. Journal of Contemporary Criminal Justice , 104398622110016. https://doi.org/10.1177/10439862211001616 .

College of Policing (2022). Domestic abuse (DA) matters change programme . College of Policing . https://www.college.police.uk/career-learning/licensed-products/domestic-abuse-da-matters-change-programme .

Dardis, C. M., & Gidycz, C. A. (2019). Reconciliation or retaliation? An integrative model of post relationship in-person and cyber unwanted pursuit perpetration among undergraduate men and women. Psychology of Violence , 9 (3), 328–339. https://doi.org/10.1037/vio0000102 .

Davies, M., & Hudson, J. (2011). Judgments toward male and transgendered victims in a depicted stranger rape. Journal of Homosexuality , 58 (2), 237–247. https://doi.org/10.1080/00918369.2011.540179 .

Davies, M., & McCartney, S. (2003). Effects of gender and sexuality on judgements of victim blame and rape myth acceptance in a depicted male rape. Journal of Community & Applied Social Psychology , 13 (5), 391–398. https://doi.org/10.1002/casp.741 .

Davies, M., Smith, R., & Rogers, P. (2009). Police perceptions of rape as function of victim gender and sexuality. The Police Journal , 82 (1), 4–12. https://doi.org/10.1350/pojo.2009.82.1.466 .

Davies, M., Austen, K., & Rogers, P. (2011). Sexual preference, gender, and blame attributions in adolescent sexual assault. The Journal of Social Psychology , 151 (5), 592–607. https://doi.org/10.1080/00224545.2010.522617 .

Davies, M., Gilston, J., & Rogers, P. (2012). Examining the relationship between male rape myth Acceptance, Female rape myth Acceptance, victim blame, Homophobia, gender roles, and ambivalent sexism. Journal of Interpersonal Violence , 27 (14), 2807–2823. https://doi.org/10.1177/0886260512438281 .

Domestic Abuse Matters (2019). (pp. 1–10). https://safelives.org.uk/sites/default/files/resources/DA%20Matters_4-force%20report_20180423%20v0.4.pdf .

Dong, Y., & Peng, C. Y. J. (2013). Principled missing data methods for researchers. SpringerPlus , 2 (1). https://doi.org/10.1186/2193-1801-2-222 .

Donovan, R. A. (2007). To blame or not to blame. Journal of Interpersonal Violence , 22 (6), 722–736. https://doi.org/10.1177/0886260507300754 .

Dreßing, H., Bailer, J., Anders, A., Wagner, H., & Gallas, C. (2014). Cyberstalking in a large sample of social network users: Prevalence, characteristics, and impact upon victims. Cyberpsychology Behavior and Social Networking , 17 (2), 61–67. https://doi.org/10.1089/cyber.2012.0231 .

Ellison, L., & Munro, V. (2009). Reacting to rape: Exploring mock jurors’ assessments of complainant credibility. The British Journal of Criminology , 49 , 202–219. https://doi.org/10.1093/bjc/azn077 .

Erdfelder, E., Faul, F., & Buchner, A. (1996). GPOWER: A general power analysis program. Behavior Research Methods Instruments & Computers , 28 (1), 1–11. https://doi.org/10.3758/bf03203630 .

Faul, F., Erdfelder, E., Lang, A. G., & Buchner, A. (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods , 39 (2), 175–191. https://doi.org/10.3758/bf03193146 .

Fisher, B. S., Daigle, L. E., Cullen, F. T., & Turner, M. G. (2003). Reporting sexual victimization to the police and others. Criminal Justice and Behavior , 30 , 6–38. https://doi.org/10.1177/0093854802239161 .

Fissel, E. R. (2021a). The reporting and help-seeking behaviors of cyberstalking victims. Journal of Interpersonal Violence , 36 (11–12), 088626051880194. https://doi.org/10.1177/0886260518801942 .

Fissel, E. R. (2021b). Victims’ perceptions of Cyberstalking: An examination of Perceived Offender Motivation. American Journal of Criminal Justice . https://doi.org/10.1007/s12103-021-09608-x .

Flynn, A., Powell, A., & Hindes, S. (2021). Technology-facilitated abuse: a survey of support services stakeholders (Research report, 02/2021). Sydney: ANROWS. https://www.anrows.org.au/publication/technology-facilitated-abuse-a-survey-of-support-services-stakeholders/ .

Flynn, A., Powell, A., & Hindes, S. (2023). Policing technology-facilitated abuse. Policing and Society , 1–18. https://doi.org/10.1080/10439463.2022.2159400 .

Franiuk, R., Luca, A., & Robinson, S. (2019). The effects of victim and perpetrator characteristics on ratings of guilt in a sexual assault case. Violence against Women , 107780121984043. https://doi.org/10.1177/1077801219840439 .

Freed, D., Palmer, J., E Minchala, D., Levy, K., Ristenpart, T., & Dell, N. (2017). Digital Technologies and intimate Partner violence. Proceedings of the ACM on Human-Computer Interaction , 1 (CSCW), 1–22. https://doi.org/10.1145/3134681 .

Freed, D., Havron, S., Tseng, E., Gallardo, A., Chatterjee, R., Ristenpart, T., & Dell, N. (2019). Is my phone hacked? Analyzing Clinical Computer Security interventions with survivors of intimate Partner violence. Proceedings of the ACM on Human-Computer Interaction , 3 (CSCW). https://doi.org/10.1145/3359304 .

Gillis, J. R., Diamond, S. L., Jebely, P., Orekhovsky, V., Ostovich, E. M., MacIsaac, K., Sagrati, S., & Mandell, D. (2006). Systemic obstacles to battered women’s participation in the Judicial System. Violence against Women , 12 (12), 1150–1168. https://doi.org/10.1177/1077801206293500 .

Giner-Sorolla, R., Montoya, A. K., Reifman, A., Carpenter, T., Lewis, N. A., Aberson, C. L., Bostyn, D. H., Conrique, B. G., Ng, B. W., Schoemann, A. M., & Soderberg, C. (2024). Power to detect what? Considerations for planning and evaluating sample size. Personality and Social Psychology Review . https://doi.org/10.1177/10888683241228328 .

Goodno, N. H. (2007). About | HeinOnline . HeinOnline. https://heinonline.org/HOL/LandingPage?handle=hein.journals/molr72÷=9&id=&page =.

Gottfredson, M. R., & Hindelang, M. J. (1979). A study of the Behavior of Law. American Sociological Review , 44 (1), 3. https://doi.org/10.2307/2094813 .

Greenwald, H. J., & Satow, Y. (1970). A short Social Desirability Scale. Psychological Reports , 27 (1), 131–135. https://doi.org/10.2466/pr0.1970.27.1.131 .

Grubb, A. R., & Harrower, J. (2009). Understanding attribution of blame in cases of rape: An analysis of participant gender, type of rape and perceived similarity to the victim. Journal of Sexual Aggression , 15 (1), 63–81. https://doi.org/10.1080/13552600802641649 .

Grubb, A., & Turner, E. (2012). Attribution of blame in rape cases: A review of the impact of rape myth acceptance, gender role conformity and substance use on victim blaming. Aggression and Violent Behavior , 17 (5), 443–452. https://doi.org/10.1016/j.avb.2012.06.002 .

Harris, B. A., & Woodlock, D. (2019). Digital Coercive Control: Insights from two Landmark Domestic Violence studies. The British Journal of Criminology , 59 (3), 530–550. https://doi.org/10.1093/bjc/azy052 .

Hayes, A. F., & Coutts, J. J. (2020). Use Omega rather than Cronbach’s alpha for estimating reliability. But… Communication Methods and Measures , 14 (1), 1–24. https://doi.org/10.1080/19312458.2020.1718629 .

Henry, N., Flynn, A., & Powell, A. (2018). Policing image-based sexual abuse: Stakeholder perspectives. Police Practice and Research , 19 (6), 565–581. https://doi.org/10.1080/15614263.2018.1507892 .

Henry, N., Flynn, A., & Powell, A. (2020). Technology-facilitated domestic and sexual violence: A review. Violence against Women , 26 (15–16), 1828–1854. https://doi.org/10.1177/1077801219875821 .

HMIC. (2014). Everyone’s business: Improving the police response to domestic abuse. In His Majesty’s Inspectorate of Constabulary and Fire & Rescue Services (pp. 1–157). https://hmicfrs.justiceinspectorates.gov.uk/publications/improving-the-police-response-to-domestic-abuse/ .

Holt, T. J., & Bossler, A. M. (2012). Predictors of Patrol Officer Interest in Cybercrime Training and Investigation in selected United States police Departments. Cyberpsychology Behavior and Social Networking , 15 (9), 464–472. https://doi.org/10.1089/cyber.2011.0625 .

Huber, A. R. (2023). Image-based sexual abuse: Legislative and policing responses. Criminology & Criminal Justice , 174889582211461. https://doi.org/10.1177/17488958221146141 .

Kanekar, S., Pinto, N. J. P., & Mazumdar, D. (1985). Causal and Moral responsibility of victims of rape and Robbery. Journal of Applied Social Psychology , 15 (4), 622–637. https://doi.org/10.1111/j.1559-1816.1985.tb00905.x .

Kellerman, I., Margolin, G., Borofsky, L. A., Baucom, B. R., & Iturralde, E. (2013). Electronic aggression among emerging adults: Motivations and contextual factors. Emerging Adulthood (Vol. 1, pp. 293–304). SAGE Publications Inc. 4 https://doi.org/10.1177/2167696813490159 .

Lee, J. (2019). Examining the Intersection of Gender and Age in Victim Blaming. https://doi.org/10.33915/etd.7439 .

Leitão, R. (2019). Technology-facilitated intimate Partner abuse: A qualitative analysis of data from online domestic abuse forums. Human–Computer Interaction , 1–40. https://doi.org/10.1080/07370024.2019.1685883 .

Leukfeldt, R., Veenstra, S., & Stol, W. (2013). High volume Cyber Crime and the Organization of the police: The results of two empirical studies in the Netherlands. International Journal of Cyber Criminology All Rights Reserved under a Creative Commons Attribution-Noncommercial-Share Alike 2 5 India License , 7 (1), 1–17. https://www.cybercrimejournal.com/pdf/Leukfeldtetal2013janijcc.pdf .

Leukfeldt, E. R., Notté, R. J., Raoul, Malsch, M., & Marijke (2019). Exploring the needs of victims of cyber-dependent and cyber-enabled crimes. Victims & Offenders , 15 (1), 1–18. https://doi.org/10.1080/15564886.2019.1672229 .

Lewis, R., Dobash, R. P., Dobash, R. E., & Cavanagh, K. (2000). Protection, Prevention, Rehabilitation or Justice? Women’s use of the Law to Challenge Domestic Violence. International Review of Victimology , 7 (1–3), 179–205. https://doi.org/10.1177/026975800000700310 .

Leys, C., Delacre, M., Mora, Y. L., Lakens, D., & Ley, C. (2019). How to classify, detect, and manage Univariate and Multivariate outliers, with emphasis on Pre-registration. International Review of Social Psychology , 32 (1). https://doi.org/10.5334/irsp.289 .

López-Cepero, J., Vallejos-Saldarriaga, J., & Merino-García, M. (2018). Digital intimate Partner Violence among Peruvian youths: Validation of an instrument and a theoretical proposal. Journal of Interpersonal Violence . https://doi.org/10.1177/0886260518803610 .

Lopez-Neira, I., Patel, T., Parkin, S., Danezis, G., & Tanczer, L. (2019). Internet of Things: How Abuse is Getting Smarter . Papers.ssrn.com. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3350615 .

Lorah, J. A. (2020). Interpretation of main effects in the presence of non-significant interaction effects. The Quantitative Methods for Psychology , 16 (1), 33–45. https://doi.org/10.20982/tqmp.16.1.p033 .

Maple, C., Short, E., & Brown, A. (2011). Cyberstalking in the United Kingdom: an analysis of the ECHO Pilot Survey . Uobrep.openrepository.com. https://uobrep.openrepository.com/handle/10547/270578 .

Martellozzo, E., Bleakley, P., Bradbury, P., Frost, S., & Short, E. (2022). Police responses to cyberstalking during the Covid-19 pandemic in the UK. The Police Journal: Theory Practice and Principles , 0032258X2211134. https://doi.org/10.1177/0032258x221113452 .

McDonald, R. P. (1999). Test theory: A unified treatment . Lawrence Erlbaum.

Millman, C. M., Winder, B., & Griffiths, M. D. (2017). UK-Based police officers’ perceptions of, and role in Investigating, Cyber-harassment as a crime. International Journal of Technoethics , 8 (1), 87–102. https://doi.org/10.4018/ijt.2017010107 .

Peirone, A., Barrett, B. J., Cheung, C. H., & Habibov, N. (2021). What makes police officers act? Police response to instances of intimate Partner violence. Journal of Interpersonal Violence , 088626051773422. https://doi.org/10.1177/0886260517734222 .

Pina, A., Storey, J. E., Duggan, M., & Franqueira, V. N. L. (2021a). Technology-Facilitated Intimate Partner Violence: A multidisciplinary examination of prevalence, methods used by perpetrators and the impact of COVID-19 . Kar.kent.ac.uk . https://kar.kent.ac.uk/id/eprint/95001.

Pina, A., Bell, A., Griffin, K., & Vasquez, E. (2021b). Image based sexual abuse proclivity and victim blaming: The role of dark personality traits and moral disengagement. Oñati Socio-Legal Series , 11 (5), 1179–1197. https://doi.org/10.35295/osls.iisl/0000-0000-0000-1213 .

Pinciotti, C. M., & Orcutt, H. K. (2017). Understanding gender differences in rape victim blaming: The power of social influence and just world beliefs. Journal of Interpersonal Violence , 36 (1–2), 088626051772573. https://doi.org/10.1177/0886260517725736 .

Powell, A., & Henry, N. (2018). Policing technology-facilitated sexual violence against adult victims: Police and service sector perspectives. Policing and Society , 28 (3), 291–307. https://doi.org/10.1080/10439463.2016.1154964 .

Power, C., Koch, T., Kralik, D., & Jackson, D. (2006). Lovestruck: Women, romantic love and intimate partner violence. Contemporary Nurse , 21 (2), 174–185. https://doi.org/10.5172/conu.2006.21.2.174 .

Reno, J., & U.S. Department of Justice. (1999). Cyberstalking: A new challenge for law enforcement and industry. Washington, DC:. Retrieved from https://www.ojp.gov/ncjrs/virtual-library/abstracts/cyberstalking-new-challenge-law-enforcement-and-industry-report .

Reyns, B. W., Henson, B., & Fisher, B. S. (2012). Stalking in the Twilight Zone: Extent of Cyberstalking victimization and offending among College Students. Deviant Behavior , 33 (1), 1–25. https://doi.org/10.1080/01639625.2010.538364 .

Rosalili, W., Ridhwan, A., Rani, A., & Husain, M. (2021). Rosli 1 , W., Kamaruddin 2 , S., Nabila, N., Saufi 3 , M., Ridhwan, A., Rani, A., & Husain, M. My Stalk alert: Swift Response for Cyberstalking victims. In UiTM Cawangan N. Sembilan, Kampus Seremban . https://ir.uitm.edu.my/id/eprint/55718/1/55718.pdf .

Saxton, M. D., Jaffe, P. G., Dawson, M., Straatman, A. L., & Olszowy, L. (2020). Complexities of the police response to intimate Partner violence: Police officers’ perspectives on the challenges of keeping families safe. Journal of Interpersonal Violence , 37 (5–6), 088626052093442. https://doi.org/10.1177/0886260520934428 .

Schmidt, A. F., & Finan, C. (2018). Linear regression and the normality assumption. Journal of Clinical Epidemiology , 98 , 146–151. https://doi.org/10.1016/j.jclinepi.2017.12.006 .

Schneider, L. J., Soh-Chiew Ee, J., & Aronson, H. (1994). Effects of victim gender and physical vs. psychological trauma/injury on observers’ perceptions of sexual assault and its aftereffects. Sex Roles , 30 (11–12), 793–808. https://doi.org/10.1007/bf01544232 .

Schneider, L. J., Mori, L. T., Lambert, P. L., & Wong, A. O. (2009). The role of gender and ethnicity in perceptions of rape and its aftereffects. Sex Roles , 60 (5–6), 410–421. https://doi.org/10.1007/s11199-008-9545-9 .

Schnurr, M. P., Mahatmya, D., & Basche, R. A. (2013). The role of dominance, cyber aggression perpetration, and gender on emerging adults’ perpetration of intimate partner violence. Psychology of Violence , 3 (1), 70–83. https://doi.org/10.1037/a0030601 .

Sheridan, L. P., & Grant, T. (2007). Is cyberstalking different? Psychology Crime & Law , 13 (6), 627–640. https://doi.org/10.1080/10683160701340528 .

Short, E., Linford, S., Wheatcroft, J. M., & Maple, C. (2014). The impact of Cyberstalking: The lived experience - A thematic analysis. Annual Review of Cybertherapy and Telemedicine , 2014 , 133–137. https://doi.org/10.3233/978-1-61499-401-5-133 .

Short, E., Guppy, A., Hart, J. A., & Barnes, J. (2015). The impact of Cyberstalking. Studies in Media and Communication , 3 (2). https://doi.org/10.11114/smc.v3i2.970 .

Statistics Canada (2016). Family violence in Canada: A statistical profile 2014 (Catalogue No. 85-002-X). Canadian Centre for Justice Statistics. Retrieved from http://www.statcan.gc.ca/pub/85-002-x/2016001/article/14303-eng.htm .

Stephens, B. J., & Sinden, P. G. (2000). Victims’ voices. Journal of Interpersonal Violence , 15 (5), 534–547. https://doi.org/10.1177/088626000015005006 .

Stevens, F., Nurse, J. R. C., & Arief, B. (2021). Cyber Stalking, Cyber Harassment, and Adult Mental Health: A systematic review. Cyberpsychology Behavior and Social Networking , 24 (6). https://doi.org/10.1089/cyber.2020.0253 .

Stewart, C. C., Langan, D., & Hannem, S. (2013). Victim experiences and perspectives on police responses to Verbal Violence in domestic settings. Feminist Criminology , 8 (4), 269–294. https://doi.org/10.1177/1557085113490782 .

Storry, M., & Poppleton, S. (2022). The Impact of Online Abuse: Hearing the Victim’s Voice . Victims’ Commissioner for England and Wales. https://cloud-platform-e218f50a4812967ba1215eaecede923f.s3.amazonaws.com/uploads/sites/6/2022/05/Hearing-the-Victims-Voice.pdf .

Strawhun, J., Adams, N., & Huss, M. T. (2013). The Assessment of Cyberstalking: An expanded examination including Social networking, attachment, Jealousy, and anger in relation to violence and abuse. Violence and Victims , 28 (4), 715–730. https://doi.org/10.1891/0886-6708.11-00145 .

Suzy Lamplugh (2021). Trust . https://www.suzylamplugh.org/Handlers/Download.ashx?IDMF=fcfb781a-f614-48c8-adcf-4cfa830c16a7 .

Thompson, L. E., Pica, E., & Pozzulo, J. (2021). Jurors’ decision making in a sexual assault trial: The influence of victim age, delayed reporting, and multiple allegations. The American Journal of Forensic Psychology , 39 (2), 19–46.

Todd, C., Bryce, J., & Franqueira, V. N. L. (2021). Technology, cyberstalking and domestic homicide: Informing prevention and response strategies. Policing and Society , 31 (1), 1–18. https://doi.org/10.1080/10439463.2020.1758698 .

UK Government (2018). Data Protection Act . Gov.uk. https://www.gov.uk/data-protection .

Wall, D. S., & Williams, M. L. (2013). Policing cybercrime: Networked and social media technologies and the challenges for policing. Policing and Society , 23 (4), 409–412. https://doi.org/10.1080/10439463.2013.780222 .

Watkins, L. E., Maldonado, R. C., & DiLillo, D. (2018). The Cyber Aggression in relationships Scale: A New Multidimensional measure of technology-based intimate Partner Aggression. Assessment , 25 (5), 608–626. https://doi.org/10.1177/10731911166656 .

White, B. H., & Kurpius, S. E. R. (2002). Effects of victim sex and sexual orientation on perceptions of rape. Sex Roles: A Journal of Research , 46 (5–6), 191–200. https://doi.org/10.1023/A:1019617920155 .

Woodlock, D. (2016). The abuse of technology in domestic violence and stalking. Violence against Women , 23 (5), 584–602. https://doi.org/10.1177/1077801216646277 .

World Health Organization [WHO]. (2010). London School of Hygiene and Tropical Medicine. Preventing intimate partner and sexual abuse against women: Taking action and generating evidence . World Health Organization.

Worsley, J. D., Wheatcroft, J. M., Short, E., & Corcoran, R. (2017). Victims’ voices: Understanding the emotional impact of Cyberstalking and individuals’ coping responses. SAGE Open , 7 (2), 215824401771029. https://doi.org/10.1177/2158244017710292 .

Yarmey, A. D. (1985). Older and younger adults’ attributions of responsibility toward rape victims and rapists. Canadian Journal of Behavioural Science / Revue Canadienne Des Sciences Du Comportement , 17 (4), 327–338. https://doi.org/10.1037/h0080045 .

Zvi, L. (2022a). The double standard toward female and male victims of non-consensual dissemination of intimate images. Journal of Interpersonal Violence , 37 (21–22), 088626052110501. https://doi.org/10.1177/08862605211050109 .

Zvi, L. (2022b). Police Perceptions of Sex-worker Rape Victims and Their Offenders: A Vignette Study. Journal of interpersonal violence, 37(15–16), NP14189–NP14214. https://doi.org/10.1177/08862605211005140 .

Zvi, L., & Shechory-Bitton, M. (2020). Police Officer Perceptions of Non-consensual Dissemination of Intimate Images. Frontiers in Psychology , 11 . https://doi.org/10.3389/fpsyg.2020.02148 .

Download references

Acknowledgements

The authors wish to express their gratitude to the following police forces: Avon & Somerset Police, Hampshire & Isle of Wight Constabulary, and Kent Police. Their valuable contributions were instrumental in the successful completion of this study.

The authors did not receive support from any organization for the submitted work. No funding was received to assist with the preparation of this manuscript.

Author information

Stavros Chatzisymeonidis

Present address: Institute of Population Health, Department of Psychology, University of Liverpool, Liverpool, UK

Afroditi Pina

Present address: School of Psychology, Centre of Research and Education in Forensic Psychology (CORE-FP), University of Kent, Canterbury, UK

Authors and Affiliations

You can also search for this author in PubMed   Google Scholar

Contributions

The authors read and approved the final manuscript.

Corresponding author

Correspondence to Stavros Chatzisymeonidis .

Ethics declarations

Competing interests.

The authors declare that they have no competing interests.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Chatzisymeonidis, S., Pina, A. Exploring police attitudes on victims’ delayed reporting and victim blame in technology-facilitated IPV. Crime Sci 13 , 12 (2024). https://doi.org/10.1186/s40163-024-00213-x

Download citation

Received : 22 February 2024

Accepted : 13 May 2024

Published : 21 May 2024

DOI : https://doi.org/10.1186/s40163-024-00213-x

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Reporting delay
• Cyberstalking recognition
• Victim blaming attitudes

Crime Science

ISSN: 2193-7680

• General enquiries: [email protected]

Police use-of-force is not as common as the public thinks. The news media too often fails to say so

Too often, use-of-force cases are tried not in courts of law but in the court of public opinion, despite research showing such cases are hardly the norm..

Protesters march around Kenosha on Aug. 24, 2020, on the second night of unrest after police shot Jacob Blake.

Ashlee Rezin/Sun-Times

The family of Dexter Reed, who was fatally shot by Chicago police in late March, has filed a federal civil rights lawsuit alleging officers used excessive force and violated his civil rights.

They are right in one respect: The incident does reveal significant misconduct, but by the media — not the police.

What actually happened on West Ferdinand Street in Humboldt Park is irrelevant or worse, dismissed or omitted in too many media accounts because it is inconvenient to the larger narrative that America supposedly faces an epidemic of racist, brutal, arbitrary and unjustified “police violence.”

The fact is, use of force by law enforcement is rare — and in only 1.8% of cases results in moderate or severe injury, one analysis found. Fatal police shootings are significantly rarer still. Those facts seem incongruous with the media narrative and the resulting public perception.

When Chicago’s Civilian Office of Police Accountability, its city-run police oversight body, released its initial report two weeks after Reed’s death, the headlines centered on a single detail: that the police fired 96 shots during the 41-second fatal encounter.

Nearly every news headline omitted the fact that Reed, who was reportedly wearing a full-face ski mask , fired first at officers, COPA said — as many as 11 times, wounding an officer. Those details barely appeared in the initial coverage and, if they did, they appear alongside the word “preliminary.” Reed, who was 26, is often shown in family photos obtained by the media in his high school graduation gown and basketball uniform.

In the days after bodycam footage was released, the controversial head of COPA , Andrea Kersten, gave over a dozen media interviews in which she implied the officers were lying about the reason for the traffic stop. Kersten, whose investigation is not concluded, also released statements to the media suggesting the number of shots fired by officers was disproportionate and therefore unlawful.

When the Chicago police chief urged against a rush to judgment against the officers — whom COPA had not yet interviewed — the officers’ prior complaints records mysteriously made it to the news media. Then Reed’s troubled and violent past — attacking his uncle with a knife in 2021 and charged with illegal gun possession in 2023 — emerged.

Reed’s case is not unique. All too often, police use-of-force cases are tried not in courts of law but in the court of public opinion, with the media as judge and jury, based on incomplete or false evidence.

After Michael Brown’s 2014 death in Ferguson, Missouri, media outlets reported rumors and half-truths suggesting Brown posed no threat, raised his hands and told the police officer not to shoot him. Protests and unrest roiled the country, and much of the public was indignant at the “unjustified” shooting. Congressmen and cable TV anchors performed the “Hands Up, Don’t Shoot” protest gesture. Then-President Barack Obama suggested the police shooting was both unjustified and racially motivated.

Both the grand jury and federal investigations later cleared the officer. But the perception of wrongdoing lingered, so much so that the incoming district attorney pledged during his campaign to try the officer for Brown’s death. In 2020, the prosecutor concluded that “an independent and in-depth review of the evidence” did not meet the standard for charges of murder or manslaughter.

In 2020, the Kenosha, Wisconsin, police shooting of Jacob Blake, too, raised an outcry based on incomplete — and often wrong — information. Media reports portrayed Blake, who was shot by police after his girlfriend called 911 on him, as unarmed. The city erupted in flames, and Wisconsin’s governor suggested the shooting was unjustified and racially motivated. In fact, Blake was armed with a knife, refused to comply with officer commands, and was tasered to little effect before he picked up the knife a second time, something Blake himself admitted. Two independent investigations — first by the Wisconsin Department of Justice and then the U.S. Department of Justice — cleared the officers. Yet the Blake incident is still cited as an instance of undue police violence .

The facts and the feelings about those cases and many law enforcement shootings remain divergent — thanks to media narratives.

Yet, much of the public thinks police violence is an “extremely” or “very serious” problem. Studies and surveys show the perceived prevalence of police use of force is many times higher than it is in reality. These beliefs are built on myths perpetuated by salacious and often misleading coverage. But, as psychologists have noted for decades, it is harder to dissuade someone of an earnestly believed falsehood than it is to persuade them of it in the first place.

Jason Johnson is the president of the Law Enforcement Legal Defense Fund and the former deputy commissioner of the Baltimore Police Department.

The Sun-Times welcomes letters to the editor and op-eds. See our guidelines .

The views and opinions expressed by contributors are their own and do not necessarily reflect those of the Chicago Sun-Times or any of its affiliates.