Introduction
The number of people who are dying in use-of-force encountersFootnote 1 involving the police is increasing in Canada (Singh Reference Singh2020). Amid a rising tide of awareness of police violence, some activists, advocates and academics have called for fundamental changes regarding the role of policing in society (Cecco Reference Cecco2020). From calls to defund police and redirect public funds to nonpolicing social services, to enhanced accountability and transparency, policing—and the violence that is endemic to policing, the muscular embodiment of the state’s claim to a monopoly on the legitimate use of force in society—is under increased and intense scrutiny (Maynard Reference Maynard2020; Pasternak, Walby, and Stadnyk Reference Pasternak, Walby and Stadnyk2022).
However, there is a lack of knowledge on deaths related to police use of force across Canada, where inconsistent data are collected, reported and known to the police, experts and the public. This means that, in a context of increased scrutiny on policing, vital conversations about policy and budgets are taking place in which we know little about the life and death outcomes of policing.
Some media outlets and independent academics have worked to keep track of police-involved deaths and families of victims of police violence have been tracking violence in their communities.Footnote 2 However, there is no publicly available central resource that consistently documents police use-of-force-involved deaths at a national scale, or that disaggregates such information by race, socioeconomic status, location or other variables such as the police service involved and the type of force used.
In this article, we explain our process of developing a collaborative and publicly accessible database of police-involved deaths across Canada. We begin by outlining the landscape of police-involved deathsFootnote 3 in Canada—what we refer to and conceptualize further as “death data”—recognizing the complexities and affective nature of documenting police violence. We then outline the methods of our interdisciplinary team who are working in criminology, social work, data science and a range of community-based advocacy, legal and human rights organizations, providing a step-by-step guide on how we built our database. The article then begins to make sense of 745 instances of police-involved deaths in which force was used from 2000 to 2023. We conclude with a series of proposals for research and action to understand and address police violence in Canada in a context of limited transparency and oversight in which we identify the opportunities and pitfalls of public criminology, and call for the collaborative data governance of policing death data, which includes communities that are most impacted and the families of those who are killed by police.
Policing and death data landscape
Canadian institutions routinely track vital information about the criminal legal system to examine the state of crime, victimization and law-enforcement practices across the country (Statistics Canada 2022a). Statistics Canada does not report on police-involved deaths and, although there are internal systems for reporting deaths within policing agencies, these agencies do not consistently report such information to the public. When a death occurs in the context of use of force, it is required by law that jurisdictional police oversight bodies launch an investigation. Such police oversight agencies exist in varying forms in most provinces, including the Special Investigations Unit in Ontario, the Bureau des enquêtes indépendantes in Québec, the Independent Investigation Office in British Columbia, the Independent Investigation Unit in Manitoba and Serious Incident Response Teams in Alberta, Saskatchewan, Newfoundland and Nova Scotia. Depending on the province, the police oversight body may publicly report aspects of individual cases that they are investigating to the public; however, the amount of information varies across the nine jurisdictions (Puddister Reference Puddister2023). Nonetheless, an overall accounting of oversight bodies’ investigations that aggregates police-involved civilian injuries or deaths, types of force and/or incident contexts has not consistently been undertaken, nor made available.
As officials do not systematically release data on use of force, since the 1990s, policing researchers in Canada have noted that one of the most significant limitations on their work is the lack of clear and consistent data and reporting practices on police-involved deaths and use of force (Stansfield Reference Stansfield1995; Davies et al. Reference Davies, Johnston, Hinton, Zdjelar and Dawson2021; Millar and Owusu-Bempah Reference Millar and Owusu-Bempah2011; Wortley et al. Reference Wortley, Owusu-Bempah, Laming and Henry2021). Indeed, a major study on the issue of police use of force found that representatives of Canadian police oversight bodies had an inconsistent and limited understanding of the issue, with some oversight representatives thinking that the use of force had decreased, while others felt that their numbers were going up (Wortley et al. Reference Wortley, Owusu-Bempah, Laming and Henry2021). The study identified that data-collection standards vary dramatically by jurisdiction and a “lack of reliable data and changes in reporting practices limited the ability of respondents to speak with any certainty” (Wortley et al. Reference Wortley, Owusu-Bempah, Laming and Henry2021, 42). Compounding data collection and reporting issues is the ubiquitous blue wall of silence—of cops protecting cops coupled with noncooperation of police officers in oversight investigations (Puddister Reference Puddister2023). Therefore, what the public, as well as experts, know about police-involved deaths in Canada is notoriously limited. As such, collectively, we have few answers to basic questions, including: How often are firearms used in fatal use-of-force incidents? Which police forces have the highest or lowest numbers of civilian deaths? How often are people killed by police in the context of a wellness call? Such questions are unanswerable due to multiple factors, including “lack of transparency, accuracy, and reporting on use of force by police” (Wortley et al. Reference Wortley, Owusu-Bempah, Laming and Henry2021, 91). The lack of available data has resulted in calls for a public national standardized data-collection system and database on the deadly use of police force (Bennell et al. Reference Bennell, Brown, Jenkins, Khanizadeh, MacIsaac and Semple2022; Laming Reference Laming2017; Kiedrowski et al. Reference Kiedrowski, Melchers, Petrunik and Maxwell2015; Malone et al. Reference Malone, Omstead and Casey2020; Wortley et al. Reference Wortley, Owusu-Bempah, Laming and Henry2021).
Theoretical and methodological underpinnings
Our research project began as an attempt to develop an open-source database on police-involved deaths in Canada. Our interdisciplinary and community-engaged collaboration acknowledges that deaths at the hands of police are political and contentious; claiming objectivity in such a space is impossible, and indeed irresponsible, as such a claim would only erase explicit assumptions and biases from view but not from the shaping of outcomes. To ensure that we are rigorous in our approach, we first detail the collective theoretical orientations that underpin our efforts, including public criminology, critical race studies, decolonial and intersectional analysis, and collaborative data governance.
Public criminology
Public criminology works to disseminate criminological research beyond the walls of academia to broader audiences, including actors in the criminal legal system, policymakers and the public (Lumsden and Goode Reference Lumsden and Goode2018; McAleese Reference McAleese2019). As a form of public criminology, our project offers a living repository of publicly accessible open-source data on police-involved deaths in which force was used. The aims of public criminology seek to not only document social issues, but also to contribute towards advocacy and transformative efforts. As noted by McAleese: “Doing criminology in this way can also ensure that research, and the advocacy work that might stem from it, contributes to meaningful, progressive, and transformative changes to criminal justice systems—changes that actually repair harm and restore individuals and communities without expanding the carceral net” (Reference McAleese2019, 368).
Towards this aim, our open-source database is accessible via our website (trackinginjustice.ca), which is updated on a regular basis. The aim is for it to be used by experts, activists, academics, journalists, policing agencies, community members and those most impacted by police-involved deaths to analyze, discuss and challenge the current context, including supporting calls for greater transparency, and to serve as an aid to calls for accountability (as the many unknowns in the data can shine a light on how little we actually know). This approach also holds our own data up for interrogation, which can be improved upon over time.Footnote 4
Critical race, decolonial and intersectional analysis
Critical race and Indigenous scholars have highlighted the need to employ a definition of race that is always connected to an analysis of racism and an understanding of race to include the material impacts and violence that racialized people face in different ways (Omi and Winant Reference Omi and Winant1994; Ahmed Reference Ahmed2004; Lee and Lutz Reference Lee and Lutz2005). We understand that the concept of race is a social construction that must be situated historically, which means attending to ongoing colonization, the historical and contemporary effects of the enslavement of Black peoples and how these historical conditions persist in structuring current racial hierarchies (Hall Reference Hall and Morley2018, 176). The history of enslavement and colonialism in Canada has a direct impact on contemporary policing, which is characterized by racial profiling, the over-policing of racialized and Indigenous populations, and disproportionate levels of violence directed at these communities (Maynard Reference Maynard2017; Maynard and Simpson Reference Maynard and Simpson2022; Murdocca Reference Murdocca2004; Razack Reference Razack2002; Gouldhawke Reference Gouldhawke2020; Monaghan Reference Monaghan2013; Pasternak, Walby, and Stadnyk Reference Pasternak, Walby and Stadnyk2022).
There is an increasing body of Canadian literature examining the relationship between racism and police use of force, which has demonstrated that Black and Indigenous people disproportionately experience harm and death when interacting with police (Pedicelli Reference Pedicelli1998; Wortley Reference Wortley2006; Carmichael and Kent Reference Carmichael and Kent2015; Wortley, Laniyonu, and Laming Reference Wortley, Laniyonu and Laming2020). Similar trends of over-policing and under-policing also impact other marginalized populations. For example, people facing mental health crises are more likely to experience negative encounters with law enforcement (Toronto Police Service 2014; Wortley et al. Reference Wortley, Owusu-Bempah, Laming and Henry2021), with a significant intersection between race and mental health when it comes to police use of force (OHRC 2014; OHRC 2018). Studies have also recognized that those experiencing the combination of mental health issues and homelessness are more likely to have interactions with police (Kouyoumdjian et al. Reference Kouyoumdjian Fiona, Wang, Mejia-Lanchero, Owusu-Bempah, Nisenbaum, O’Campo, Stergiopoulos and Hwang2019).
Awareness of the intersectional axis that asymmetrically enables intensified oppression of certain communities informs our work in myriad ways, including through ensuring that the communities that are impacted the most are involved in the analysis and that we continually interrogate state-defined categories that are used to label people as presenting risks that can justify use of force.
Collaborative data governance
Collective trauma resulting from intensified police and carceral violence has led to distrust in policing among African, Black and Caribbean people, Indigenous people, trans and gender nonconforming people, and other groups made to be marginalized, including sex workers, as well as people who experience mental health and/or substance use issues (Alberton et al. Reference Alberton, Gorey, Angell and McCue2019; Chih Reference Chih2020; Crago et al. Reference Crago, Bruckert, Braschel and Shannon2021; Salerno and Schuller Reference Salerno and Schuller2019; Selfridge et al. Reference Selfridge, Greer, Card, Macdonald and Pauly2020; Waldron Reference Waldron2020). Due to ongoing and increasing mistrust of policing institutions, relying on police or government oversight bodies to undertake the role of governing data of police-involved deaths may not be the best path forward (Puddister and McNabb Reference Puddister and McNabb2021). Furthermore, in the Canadian context of ongoing colonization, as our project develops, we are exploring how the Indigenous models of data governance that are grounded in the First Nations Ownership Control Access and Possession (OCAP) principles (First Nations Information Governance Centre 2024) could be followed in our work. We make data publicly accessible so that communities can have access and possession of the data. We want our approaches to ensure that the collection of data on police violence helps to move forward the Truth and Reconciliation Commission (2015) Calls to Action—specifically number 39, which calls for a national plan to collect and publish data on Indigenous homicides.
When we began this project, we asked ourselves what such a project would look like if it were not designed and owned by government bodies but, rather, developed and governed by civil society, including the communities most impacted by policing. Could a model of collaborative data governance work? As our project is a work in progress, we are still answering these questions. As a starting point, our team collaboratively stewards and maintains the data and we make them accessible in ways that aim to be trauma-informed and supportive of the needs of the communities most impacted by police violence, in order to produce counternarratives to dominant framings from policing agencies that regularly justify the use of deadly force.
Methods: developing a database on police-involved deaths
Policing and death data initiatives
To begin our efforts, we explored jurisdictions outside Canada in which policing data initiatives had been established—those involving criminologists, journalists, data scientists, the public and the police themselves. We examined models of crowdsourcing and reporting of police-involved death data, including the open-source Mapping Police Violence and Fatal Encounters, both from the United States, and the National Deaths in Custody Program (NDICP)—an academic and government partnership model in Australia in which police directly report information on deaths to independent academic review. We summarize each below, including the approaches taken and types of data collected.
Both the Mapping Police Violence and Fatal Encounters rely on publicly available reports from media and other sources; the projects have been statistically analyzed and, from 2015 to 2019, were found to be highly comparable (Comer and Ingram Reference Comer and Ingram2023). Mapping Police Violence contains a detailed and transparent methodology on their website, including justifications for how they classify details and cases, where they state: “While we strive to employ official data sources from local and state government agencies, we believe it is important to continue collecting data from publicly accessible media sources. This allows us to identify gaps in government data, and further triangulate and validate the data” (Mapping Police Violence 2024). Both projects provide near-comprehensive lists of deaths to the public in ways that enable comparisons, aggregation of data and descriptive statistics to address pressing questions coming from communities impacted by police violence. However, it has been noted that, over time, the sources are becoming less similar, which necessitates continued inspection of the data across the various open-source platforms over time (DeAngelis Reference DeAngelis2021). An open data approach enables this form of scrutiny and comparison.
In Australia, NDICP death data come directly from police forces, which are compared with coroner information. However, the source data are not public, nor are they searchable by the public via the police force involved, nor are they updated in near real time, such as open-source projects. The NDICP data are also not analyzed by police forces and include little information on the deceased outside of race, gender and age categories.
The NDICP includes the charges facing people who died and both Mapping Police Violence and Fatal Encounters indicate allegations of the victim having a weapon (but also cite where the allegation comes from, to enable critique). None of these three initiatives identifies the involvement of people who have faced police violence in the development of the projects. However, crowdsourcing is often driven by impacted communities.
To further understand the landscape of police-involved death data in Canada, we looked at two projects developed by activists—La Coalition contre la Répression et les Abus Policiers (la C.R.A.P.) and Killer Cops Canada. We also examined the widely publicized journalist-led initiative from the Canadian Broadcasting Corporation (CBC): Deadly Force. Each employs different approaches and methods of data collection that we summarize below.
La C.R.A.P. is a grassroots organization that is known for supporting and mobilizing families of people who have died or been killed during police interactions. Their website includes a list of hundreds of “[p]ersons who lost their lives at the hands of the police in Canada (since 1987),” including their name, age, location (by province) and date of death.
Criminologist Jeff Shantz of Kwantlen Polytechnic University has documented police-involved deaths in Canada through the website Killer Cops Canada from 2015 to the present. The website includes details on hundreds of deaths, including summaries of the circumstances and police forces involved. Deaths are posted as a blog post and each is tagged by province, as well as date. Despite the impressive and comprehensive list of deaths across the country, the project is not searchable in ways that enable aggregate or comparative analysis.
In 2017, the CBC released an investigative series that tracked a subset of police-involved deaths in which force was used since 2000. The CBC first reported their data in 2017 and then again in 2020. While, initially, the CBC did not make all their data public, they shared information with Pivot Legal Society, who, along with the CBC, developed an analysis of the data (Kim Reference Kim2019). At that time, the data included location information. The analysis and the accompanying database of police-involved deaths were updated in 2020 and the dataset was made available to the public for a brief period without location information. The CBC data include 555 cases, which are online and searchable to the public, including name, age, gender, race, whether the victim was alleged to be armed, including the type of weapon, information on mental health and, if available, an image of the individual. Policing scholars have noted that, while certain practices in the collection of data by the CBC meet journalistic standards, they may not meet scholarly research standards, such as lacking a transparent methodology (Bennell et al. Reference Bennell, Brown, Jenkins, Khanizadeh, MacIsaac and Semple2022). Disability rights communities and families of people who were killed have found that the framing of the project further stigmatizes and pathologizes the victims. Despite concerns, this database has been the most comprehensive list of police-involved deaths available in Canada and, as such, has been used as a source by a range of leading policing scholars (see Wortley et al. Reference Wortley, Owusu-Bempah, Laming and Henry2021).
La C.R.A.P. and Killer Cops Canada provide a vital resource to activist communities and researchers, but the initiatives are not open-source or collaborative and they do not have clear methodologies or provide transparency on their sources of information that enable cases to be searched, verified or compared. The CBC database has numerous critiques and, while it is public, searchable and available for interrogation, it was not sustained beyond 2020. Furthermore, the sources of the data are not provided, making deeper analysis a challenge. And, as it is a journalistic project, it is also not clear whether people who were impacted by police violence were involved in its conception. For example, the database includes information on whether the victim was “armed” without indicating whether this was an allegation and, if so, where the allegation originated. Such allegations can be highly debated and it is important to understand whether they come from an eyewitness or from the police themselves, as allegations of a weapon being involved can stigmatize and act as justification for the use of deadly force.
Seeking to address the multiple gaps and challenges that are facing these important and available initiatives, we outline below our process for developing a public and collaborative database that involves people with lived experience, including families of loved ones who have died or been killed by police, with a transparent methodology, which is open-source, verified and available online.
Developing the database
We now turn to a step-by-step guide on how we undertook the development of our database—which documents police-involved deaths when force was used in Canada from 2000 onward—including examination of the sources of data available, data collection, verification, consulting to ensure that data are community- and trauma-informed and knowledge mobilization. We first determined the scope of the project in relation to which deaths to include in the database. A death is included in our database if police used intentional force with any form of weapon or restraint. We also include deaths in which police use environmental force—such as a no-knock raid or violent entry. For this particular iteration of the project, we do not include cases in which no such force was used, including deaths as a result of a vehicle crash that was triggered by a police chase, falls that may have occurred when police were present, deaths in relation to medical distress when no force or restraint was used, as well as deaths that occurred in custody after an arrest (although we are in the process of thinking through how to document police-involved deaths when force is not used as well as developing a separate database to document deaths in custody). While we focused on individuals who died in an incident in which police force was used, confirming an official cause of death is beyond the current scope of this project, as we do not always have access to coronial records. As a result, inclusion in the database does not necessarily mean that there is a causal link between the type of force used and the individual’s death.
We then set out to establish a data dictionary of the variables determined to be collected as part of this project for inclusion in the database. We used the CBC’s Deadly Force dataset as a starting point for scrutinizing and incorporating variables. The data we collect aim to provide consistent information about every recorded case. Variables include the date the death occurred, the individual’s name, age and identified gender and race, the location of the death (by province or territory), involved police service(s), highest level of force used and alleged weapon type that the individual may have been carrying. Some of the variables are straightforward (such as date of death and age), while others are more complex and require further explanation and examination, such as race and gender.
We follow categorizations from Statistics Canada and the Ontario Human Rights Commission, in which racial categories reflect a general understanding of race as a social descriptor that differs from ethnic origin, religion and geographical region. Due to our data sources, information about a victim’s race will often be drawn from media, coroner and/or oversight-body reports. We decided not to independently assign race to an individual by looking at photographs or examining their name. This approach results in many “unknowns,” as reports of a death do not often include race information. There may also be discrepancies, therefore, between how an individual would have self-identified and the race as assigned by others, whether it is the police or the broader community. This is why we allow the correction of incorrect information associated with race or any other data through a forum that can be submitted via our website.
Because the dataset relies primarily on media reports and official government sources, in many cases, a person’s gender will have been determined by the implicated police service, watchdog or legal body based on physical characteristics and reported in the media as such. We recognize the complexity and problems associated with relying so heavily on police-reported information, including the potential for misgendering. At times, the broader community will be a more accurate source of information for identifying an individual’s gender. We have a form on the website for family members, community members and loved ones to correct information if incorrect information associated with gender appears in this dataset.
Once we had our definitions, we identified our sources of data—some more robust than others. We work with the limited information that is already public to verify that information, aggregate it into one place and make it accessible for others to scrutinize, analyze and mobilize. Data collection occurs via web-scraping from publicly available sources, including government reports, press releases and mainstream Canadian media sources (Eisenstein Reference Eisenstein2022). All cases in our dataset are from government sources of data or verifiable mainstream media sources. We have drawn on the following sources of online data to compile the database:Footnote 5
Existing datasets: CBC’s Deadly Force, the only existing and public dataset that we were aware of and to which we had access (https://newsinteractives.cbc.ca/longform-custom/deadly-force/)—used as a foundation for the database.
Mainstream Canadian media: Online via targeted searches for dates, locations, police services or names of individuals.
Police oversight-body reports: Online on various oversight-body website data portals and media release web pages (e.g. https://www.siu.on.ca, https://www.alberta.ca/alberta-serious-incident-response-team, https://iiobc.ca/).
Coroner’s inquest reports: Online at various provincial coroner’s websites (e.g. https://www2.gov.bc.ca/gov/content/life-events/death/coroners-service/inquest-schedule-jury-findings-verdicts, https://www.ontario.ca/page/2024-coroners-inquests-verdicts-and-recommendations), as well as the independently-run https://inquestsca.knack.com/inquestsca#.
Other tracking initiatives: Online at https://killercopscanada.wordpress.com/ and https://www.lacrap.org/—only used as contextual information and to identify other sources.
As a starting point, we built upon the original Deadly Force dataset from the CBC, as it was the largest compiled dataset built by an accredited media source in Canada on the issue. We examined each case in the CBC dataset, performing online searches to verify the sources, and looked at cases on other tracking initiative sites to identify missing information. In some cases, we were able to update information but, in most cases, further information was not available.
We then conducted independent Internet searches every month to identify new and past cases. New cases are identified via a series of Google alerts and regular and ongoing online searches for past and current police killings and deaths on mainstream media websites and on Killer Cops Canada and la C.R.A.P. When a new case is identified, a member of our team adds information about the case into a spreadsheet based on the variables we collect. We only add information into the spreadsheet from coroner documents, oversight-body sources or mainstream media sources.
A summary of the findings relevant to each variable is then subject to a verification process. Verification occurs by a second member of our team, who independently confirms that the information is an accurate summary of what is included from the sources. To confirm information, we again search for online credible sources. Once a case is verified, it is assigned a number, added into the database and then made available on our website. There may be a lag between when a case is first identified and when it is added into the online database; however, the database is updated every month.
Our process of developing this project by documenting police-involved deaths in Canada is not linear; it is cyclical. Over time, new information becomes available and, as we examine aspects of cases, the data become clarified. If a case that is based on a recent police-involved death is identified, initially, there may be very limited information available. We regularly conduct online searches to identify new pieces of information as they emerge, including consulting final investigative reports issued by police oversight agencies, which tend to offer some clarifying details. When new details are identified, they also go through our verification process and are then added into the online dataset.
Consulting to ensure that data are community- and trauma-informed
Our approach is an ongoing dialogical process with our team and those impacted by police violence. In practice, this works via a Family Council that comprises, at any given time, five to eight family members of people who died during a police interaction or were killed by police. Family members are recruited via Affected Families of Police Homicide Ontario. We meet on an ad hoc basis, generally every two months, hosting feedback and listening sessions. Family members are compensated $50 per hour to attend a session, typically one to two hours in duration. We do not collect “data” from the Family Council; rather, they assist in setting directions, identifying gaps and assisting with ethical framings.
Key examples of the impact of the Family Council are that we have made viewing the names of people in the public dataset an opt-in feature and we have a form on our website through which family members can request changes (including adding a missing case, correcting information or withdrawing someone’s name). Additionally, the Family Council has provided guidance on allegations of being armed. Families told us that this information needed to be carefully framed, as the police have used the presence of a weapon to justify deadly use of force and, in some instances, the presence of a weapon may be contested as untrue and a fabrication by police. To not further policing narratives of an incident, we have been working to ensure that we approach the discussion of a victim’s weapon allegations with care and, while doing so, we have not included this variable in the public dataset online, although we do explore preliminary findings that we have compiled further below.
Releasing findings to the public
Aligned with public criminology, our primary approach is to provide the data in accessible ways to the public. Thus, we released all the data in early 2023 on our website in downloadable and searchable formats. To support the public’s understanding of data, we also provide an overview analysis of general findings within the dataset by using descriptive statistics, which we also ground in critical race and decolonial framings. Due to the limits of what is known and available, we make no claims that the data we present are complete. When presenting the dataset publicly, we state that it is a minimum of what is known. We have also worked to ensure that families and impacted communities are engaged as sources in any media addressing the data. Below, we explore some preliminary findings based on the data that we have collected, compiled and analyzed thus far.
Findings and analysis
The findings and analysis presented below are preliminary, acting as an introductory inflection point to spearhead a conversation in Canada about the need for official bodies to better collect and divulge data related to police violence. One of the limitations of relying on media and official (e.g. coroner, police oversight body) reports is that there are many unknowns. For example, the race or ethnicity of a deceased person is rarely documented in these sources. While there are existing unknowns in the dataset, we have enough information to document some initial trends and compile preliminary analyses. It is our hope that other researchers will test the validity of our data as other information related to police violence comes to light, such as what researchers have been able to do with official statistics and various unofficial datasets in the United States (see e.g. Conner et al. Reference Conner, Azrael, Lyons, Barber and Miller2019; Feldman et al. Reference Feldman, Gruskin, Coull and Krieger2017; Gray and Parker Reference Gray and Parker2019; Ozkan et al. Reference Ozkan, Worrall and Zettler2018) and are beginning to do here in Canada (see Simpson and Nix Reference Simpson and Nix2024).
Our dataset covers deaths that followed from any intentional police use of force in Canada from 2000 to 2023. There are 745 recorded deaths when police force was used during this 24-year period. The data include deaths in relation to police shootings and instances in which a person died after being subjected to other types of police weapons (e.g. Tasers, batons) or physical interventions (e.g. punches, holds or restraints). In some cases, more than one type of force was used; however, we indicate the highest level of force used from the following typology: gunshot, physical force, Taser/CEW (conducted energy weapon), intermediate weapon, restraint or other.Footnote 6 As mentioned above, we are not suggesting that the cause of death in every case was due to police use of force but, in every instance, some type of police force was used surrounding the death. The data reveal that police-involved deaths in which force was used increased from 2000 to 2023. Breaking down the time frame into two 12-year segments—2000–11 and 2012–23—there were an average of 23.3 deaths per year between 2000 and 2011 and 38.8 deaths per year between 2012 and 2023. This represents an increase of 67.0 percent. Figures 1 and 2 document the counts and rates per 100,000 people per year from 2000 to 2023, for which a general upward trend is indicated. The most significant increases began in 2020, when annual deaths surpassed fifty and peaked in 2022 at sixty-two documented deaths in which police force was used. The overall increase in deaths since 2000 could be due to a range of factors, including increased access to information; however, access to information alone does not explain the dramatic increase since 2019. Our focus on a critical race and decolonial analysis assists in understanding why and how policing is organized in Canada but, to better understand this increase, more research and other theoretical tools are required, which is why we have made this information public.
Figure 1. Deaths involving police use of force—counts.
Figure 2. Deaths involving police use of force—rates per 100,000 people.
In the twenty-four years from 2000 to 2023, the national annual average was 0.088 deaths per 100,000 people. Yet, the average annual rates of deadly use-of-force encounters had increased over that time. When considering an increase in the rates relative to population growth, the average annual deadly use-of-force rate in Canada was 0.071 per 100,000 people between 2000 and 2011. By comparison, the average annual rate rose to 0.104 per 100,000 people between 2012 and 2023, representing a 46.0 percent increase. Although some may expect rates of crime and police violence to rise in tandem with increases in population, the increase in deaths occurring during police use-of-force encounters outstrips population growth. When it comes to the violent crime severity index, although rates of violent crime have steadily increased since 2014 (with the exception of a slight decrease from 2019 to 2020), the rate of 97.79 incidents per 100,000 people in 2000 was almost identical to the 97.4 incidents per 100,000 people in 2022.
Despite many unknowns, the data underscore persistent racial disparities. For example, people identified as Black and Indigenous are over-represented in police use-of-force-involved deaths in Canada. Of the 745 recorded deaths in the Tracking (In)Justice dataset (2000–23) (Figure 3), 244 were identified by media, police or other authorities as white, 120 were identified as Indigenous, fifty-seven were identified as Black and 281 were determined to be unknown. The remaining forty-three cases were identified as Asian (n = 18), South Asian (n = 12), Arab (n = 5), Other (n = 5) and Latin American (n = 3).
Figure 3. Deaths involving police use of force—percentages for racial groups (2000–23).
Based on Statistics Canada’s 2021 census data, Black people made up 4.3 percent of the population and Indigenous people comprised 6.1 percent (Statistics Canada 2022b), whereas our data show that people who identified as Black and Indigenous represent 7.7 and 16.1 percent, respectively, of police-involved deaths in Canada. Racial disparities are further reflected in the numbers specific to police-involved shooting deaths. In these figures, people identified by police or other authorities as Black represent 8.3 percent of the total number, while people identified by police or other authorities as Indigenous represent 18.2 percent. Taken together, Black and Indigenous people comprise around 10.4 percent of the population in Canada, yet account for 26.5 percent of police-involved shooting deaths, when and where the race of the victim has been identified by authorities. We must emphasize that numbers are likely higher, but we simply do not have data on the identified race of 297 individuals.
Figure 4 includes a breakdown of deaths by province and territory, each of which has experienced at least one police-involved death when force was used in the past 24 years. Ontario has the most deaths, at 233, followed by British Columbia at 148, Alberta at 134, Quebec at 120, Manitoba at forty-two and Saskatchewan at thirty-one. The remaining provinces and territories have experienced ten or fewer deaths since 2000.
Figure 4. Deaths involving police use of force—counts by province/territory (2000–23).
Overall, these numbers, when compared with overall population proportions, reveal some stark disparities. For example, British Columbia comprises 13.5 percent of the Canadian population, yet accounts for 19.9 percent of the total number of deaths. Likewise, while Alberta comprises 11.5 percent of the Canadian population, the province has experienced 18.0 percent of the total deaths. Manitoba and Saskatchewan also have higher proportions of deaths relative to population, at 5.6 to 3.6 and 4.2 to 3.1 percent, respectively, as does Nunavut, at 1.1 percent of deaths compared with 0.1 percent of the overall population. Notable among these five particular provinces and territories (British Columbia, Alberta, Saskatchewan, Manitoba and Nunavut), with higher proportions of police-involved deaths relative to population, is that ninety-five of the 120 deaths (or 79.1%) of people identified as Indigenous have occurred there.
When it comes to police use-of-force-related deaths, certain police departments are more represented than others (Figure 5). The three largest police forces—the Royal Canadian Mounted Police (RCMP), Ontario Provincial Police and Sûreté du Québec—are implicated in a large number of deaths (n = 290). Although the RCMP and the two provincial police forces are responsible for policing larger amounts of territory, they may not necessarily be responsible for policing larger population numbers. More research is required to understand the proportions of the population under specific police jurisdictions to make a proper comparison across jurisdictional lines. At the municipal level, the Toronto Police Service is implicated in the greatest number of deaths (n = 67). Of note is the Edmonton Police Service—despite ranking as Canada’s fifth-largest municipality by population according to the 2021 census, the EPS is ranked second behind Toronto (n = 41).
Figure 5. Deaths involving police use of force—counts by police service (2000–23). In some instances, more than one police service is involved or implicated in a fatality. For example, while the RCMP is the sole force in 206 cases, it was also involved in seven other cases. The same goes for the Sûreté du Québec and Ontario Provincial Police, who were each implicated in two other cases in addition to those listed.
Regarding the racial identity of victims, three police services were implicated in 61.4 percent of all Black-identified deaths (n = 35). These include the Toronto Police Service (n = 19), Peel Regional Police (n = 8) and the Service de Police de la Ville de Montreal (n = 8). When it comes to Indigenous deaths, the RCMP is implicated in more than half the deaths, at sixty-two out of 120. The Winnipeg Police Service is implicated in fourteen Indigenous deaths, followed by the Kativik Regional Police Force (n = 6), EPS (n = 6), Ontario Provincial Police (n = 4), Prince Albert Police Service (n = 4) and Calgary Police Service (n = 3). Together, these seven police services account for 89.1 percent of all Indigenous deaths in Canada when police force was used.
Shooting deaths account for the most fatalities in which police force is used, representing 555 out of 745 cases, or 74.5 percent. One of the main contributors to the rise in police-involved deaths has been a notable increase in the number of people shot by a firearm during interactions with police. The data clearly indicate an upward trend in gunshot fatalities since 2000, as shown in Figure 6, which displays a timeline of the highest level of force used by police.
Figure 6. Police-involved shooting deaths—counts.
According to our findings, of the people who have died in encounters with Canadian police, 27.5 percent (or 205 cases) were unarmed (Figure 7). In almost one-third of cases (n = 241, or 32.3%), the victim was allegedly holding a knife or cutting instrument. In 22.3 percent (n = 166) of cases, the person who died was allegedly in possession of a firearm. In twenty-one other cases, it is unknown or undetermined whether the victim was in possession of a weapon (or the type of weapon). In at least two cases, the allegation that the victim had a weapon has been contested by family members.
Figure 7. Alleged weapon types.
When examining the types of weapons possessed by those who died in situations in which police force was used and, more specifically, with the highest level of force, we see that police officers disproportionately use their firearms in comparison with the type of weapon that a victim may have. Of the 745 police-involved deaths, 555 were recorded as a result of gunshot wounds. This means that almost three-quarters (74.5%) of individuals in the dataset were killed by a firearm during a use-of-force encounter with police. Out of the 555 individuals, 391 (or 70.5%) were not in possession of a firearm. In another eighteen instances, it is unknown whether the individual was armed. The largest proportion of cases—39.5 percent (or 219 cases)—involved individuals who were allegedly in possession of a knife, axe or other cutting instrument when shot by police. Fifty individuals (or 9.0 percent) were determined to be unarmed.
Discussion and conclusions
We conclude with reflections on the project, including the challenges that we have faced and potential directions for the future, including the ethical complexities of datafication, knowledge development of what we call death data and the challenges of enumerating deaths, pitfalls of official sources, the data needs of communities and the living nature of the project.
Datafication
Making communities visible through data is not a neutral process (Benjamin Reference Benjamin2019). Race-based data could be used to place the burden on people who have been made vulnerable by systems of oppression. At the same time, data on how racialized communities are impacted by police-involved deaths may help to inform transformative responses to address ongoing violence and injustice. We conceive of our intervention in the context of calls for data justice, which recognizes the impacts of data-driven processes to inform policy and politics (Kidd Reference Kidd2019; Dencik, Hintz, and Cable Reference Dencik, Hintz and Cable2016; Dencik et al. Reference Dencik, Hintz, Redden and Treré2022; Green Reference Green2021). In this project, we aim to resist and counter forms of policing datafication, such as those aimed at risk classification and pre-emption, which pathologize communities, intensifying surveillance and potential violence (Brayne Reference Brayne2021). Instead, we turn the lens of data onto the institution itself, highlighting the violence of policing through data.
Death data
The material daily work of our team involves sifting through official documentations of a person’s death—oftentimes deaths that are violent and horrific. We then extract pieces of information related to the death and insert those details into a spreadsheet. Here, we rely on official accounts of a death, which are generally created by and for those who are responsible for the death and to justify the legality of it. Officials hold the power to frame people outside of how they knew themselves (e.g. via ID, self-identified gender vs. assigned sex, mental health diagnoses). Despite what is true about a person, in death, such documents stand as an official truth.
Working with death data has been emotionally taxing on our team and the outcomes can be more taxing on those impacted when faced with enumerated data about their communities or family members. The ways in which death data are handled can result in harm through retraumatizing impacted communities and the families of the people whose deaths end up as data. Certain details on deaths could be stigmatizing on both an individual and a community level. And, as our work is public, the ways in which death data are framed are of vital importance, as this could be how someone is partially remembered. Many of the people in the database only come to be known publicly because of how they died. Acknowledging the affective nature of working with death data is important for our team: these are not just data—they are death data. Such an understanding is vital to ensure that we care for ourselves, but also that we have an ethics of care for the data themselves. This ethics of care means that we reject an approach that merely documents or objectifies for academic study—we must ensure that data are generated for a social purpose.
Furthermore, there is also a limit to what can be understood via numbers. We have heard from family members about the lack of support in place when a death occurs. Numbers cannot help in understanding the depths of grief and system failure that families meet. For example, when someone is killed or dies during a police use-of-force incident, it is often not immediately understood by the state as wrongdoing or a crime. As a result, the family is often not eligible for Victim’s Services programmes and thus is left with little support. In many provinces, families must pay for legal representation at an inquest. Police oversight-body investigations can be lengthy, opaque and intimidating, and can lack a trauma-informed approach. Autopsy reports and investigation outcomes can be withheld. Outcomes of investigations can seek to justify the death, pathologizing the person who died. Much of what we have heard from the families with whom we have engaged mirrors studies such as the work by Outland et al. (Reference Outland, Noel, Rounsville, Boatwright, Waleed and Abraham2022), who outlined how families who face police killings in the United States must carry the trauma and grief of their experience, with limited support and intense stigma, while working to hold police accountable. Research into the repercussions of homicides has noted that there are many gaps in services available to survivors in the immediate aftermath of traumatic events when media attention is often at its highest (Cherry Reference Cherry2021). Furthermore, families of homicide victims can often feel stigma and shame related to the death of a relative and being a survivor, in which the homicide victim can come to be blamed for their own death (Armour Reference Armour2002; Sharpe Reference Sharpe2015). Research has also examined how racialized communities face intensified mental health needs that result from the impact of police violence (Avery and Ruggs Reference Avery and Ruggs2020; Smith and Robinson Reference Smith Lee and Robinson2019) and how sociocultural factors influence the coping strategies of Black family members and friends of homicide victims (Sharpe Reference Sharpe2015). With this, there are still limited qualitative data on these experiences from a Canadian context and perspective. Acknowledging the limitation of enumerating deaths is one step and the adage of “more research is needed” is another, although what is really needed is support and action.
Pitfalls for official sources
There are many gaps in understanding the issue of police-involved deaths when relying solely on media and government accounts. We recognize that what we identify as credible sources may not provide the entire context and may be biased. Sometimes, what the media, coroner or an oversight body presents as “facts” are contested or incorrect. We accept this as a limitation. Furthermore, issues can include the underreporting of incidents, incorrect information recorded and a lack of information on demographics—specifically gender, race, socioeconomic status and whether the victim was labelled with a mental health diagnosis. Thus, a reliance solely on institutional accounts of the incident may only tell one side of the story.
Data needs
Enumerating the high percentages and increasing numbers of individuals who are shot and killed by police has contributed to rising calls for police accountability, transparency, reform, reallocation of funding and abolition. However, the sheer number of unknown cases makes it difficult to piece together a full and completely accurate picture of racial representation regarding police-involved deaths. Millar and Owusu-Bempah (Reference Millar and Owusu-Bempah2011) have documented the suppression of racial data from police organizations—a practice that they refer to as whitewashing, which can serve to conceal inequalities. Police and oversight bodies rarely disclose (or collect) the identified race of a victim in official reporting. Yet, victims’ families are increasingly calling for the disclosure of racial information so that comparisons can be made with the broader population and further analyses conducted alongside other variables, including the location of deaths. For example, despite the difficulties in accessing self-reported racial data for police-involved deaths, the perceptions of police and broader societal perceptions of an individual’s race are directly relevant to understanding the links between racism and police-involved deaths in which force is used. Furthermore, such data could assist in identifying and tracking race-based disparities within the criminal justice system, including the impact of systemic discrimination and unconscious bias.
And, while people identified by police or other authorities as Black and Indigenous are over-represented in the number of documented police-involved deaths relative to their respective proportion of the population, it is not our intent to suggest that equal representation of police-involved deaths along population percentages would render a more equitable or just system of policing. Rather, our aim is to document (to the extent possible) the identified races of those individuals who are dying during encounters with police to help inform broader calls for accountability and transparency in policing.
We also need to better understand how race intersects with other indicators including age, ability, gender, mental health and socioeconomic status when evaluating levels of police use of force. A clear picture of the extent and nature of systemic discrimination in policing and the Canadian criminal legal system will not be possible unless Canada adopts systematic sociodemographic data-collection and disclosure practices.
Living database
While the database comprises death data, we refer to our work as a living database. Using an iterative approach, the work is ongoing, can be improved upon, is open to contestation and is open to improvement. As a living dataset, we have been encouraged to increase the number of variables that we collect to provide greater insight and context on incidents. This expansion could include tracking the origin of the incident (i.e. wellness check, disturbance, break and enter, traffic stop, etc.), the origin of the armed allegation (i.e. police or eyewitness account), whether the armed allegation is contested (i.e. eyewitness who was present), whether the victim was understood to be in crisis, whether a body camera or dash cam was being used and the outcomes of any oversight-body investigations. Furthermore, comparisons and analyses could be made with police use-of-force policies.
As Canada is currently reckoning with unprecedented racial justice and decolonial movements, as well as intensified scrutiny of police conduct, it is our hope that the Tracking (In)Justice project—as a form of public criminology—will shed further light on police use of force in Canada while also contributing to wider calls for greater transparency and accountability in policing.
Open access
