Racial Disparities in Police Use of Deadly Force Against Unarmed Individuals Persist After Appropriately Benchmarking Shooting Data on Violent Crime Rates
Open accessThe DOI matches its Crossref record; 35 of the 35 references it cites were found in a registry or an index; 13 of the 43 sources its claims cite were checked for support.
Abstract
Cesario et al. argue that benchmarking the relative counts of killings by police on relative crime rates, rather than relative population sizes, generates a measure of racial disparity in the use of lethal force that is unbiased by differential crime rates. Their publication, however, lacked any formal derivation showing that their benchmarking methodology has the statistical properties required to establish such a claim. We use the causal model of lethal force by police conditional on relative crime rates implicit in their analyses and prove that their benchmarking methodology does not, in general, remove the bias introduced by crime rate differences. Instead, it creates strong statistical biases that mask true racial disparities, especially in the killing of unarmed noncriminals by police. Reanalysis of their data using formally derived criminality-correcting benchmarks shows that there is strong and statistically reliable evidence of anti-Black racial disparities in the killing of unarmed Americans by police in 2015–2016.
Overview
CoChat Summary
TL;DR
The paper formally evaluates Cesario et al.’s crime-rate benchmarking method for police deadly force and shows it does not remove crime-rate bias, instead biasing estimates in ways that can mask anti-Black disparities, especially for unarmed noncriminals. Reanalyzing the same 2015–2016 data with corrected benchmarks, the authors find strong evidence of anti-Black disparities in police killings of unarmed civilians, while armed-suspect disparities track violent crime rates.
Key findings
- Formal derivation shows Cesario et al.’s benchmark does not generally yield unbiased estimates; it can create stronger statistical biases that mask true racial disparities, especially in killings of unarmed noncriminals. Using NCVS 2016 violent crime data, the bias would multiply the true anti-Black disparity by approximately 0.38; if unarmed, noncriminal Black individuals were 2.6 times more likely to be killed by police than unarmed, noncriminal White individuals, the Cesario et al. methodology would suggest no racial disparities.
- Reanalysis with corrected criminality benchmarks finds substantial evidence of anti-Black racial disparities in police killings of unarmed civilians across all crime benchmarks in both 2015 and 2016.
- For armed suspects, the authors recover Cesario et al.’s principal finding: racial disparities in the killing of armed suspects are proportional to relative rates of violent criminality. NCVS data suggest almost perfect proportionality, whereas NIBRS data suggest White suspects are killed at greater rates than expected relative to their violent crime rates.
- NIBRS data show greater racial differences in crime rates than NCVS data across almost all years and crime classifications; benchmarking on NIBRS may underestimate anti-Black disparities in police use-of-force.
Methods
- The authors build a formal causal/generative model of police killings conditional on race and criminality, derive expected relative risks under population-size and crime-rate benchmarks, and prove the bias properties of Cesario et al.’s methodology.
- Data on civilian killings by police come from “The Counted” (The Guardian, 2015–2016), with “unarmed and nonaggressing” data as presented in Cesario et al. Crime benchmarks come from NIBRS and NCVS for violent crimes (major), violent crimes (minor), and weapons violations.
- Bayesian statistical models are coded in Stan and fit using R and rstan; the analysis compares posterior distributions of the relative probability of Black versus White individuals being killed by police, with central 90% credible intervals.
Limitations
- Paper-stated: the causal model assumes armed individuals killed by police come from the criminal subpopulation and unarmed individuals come from the noncriminal subpopulation; the authors note this is unlikely to exactly hold in empirical data, though they expect a strong correlation.
- Paper-stated: NIBRS compliance is variable and may be nonrandom; NIBRS data may be biased by differential policing and reporting, so benchmarking on NIBRS may underestimate anti-Black disparities.
- Evident from the text: the analysis relies on observational police killing data and crime benchmarks that cannot directly measure encounter contexts or officer-level bias, so causal conclusions depend on model assumptions; the “unarmed and nonaggressing” coding was done by Cesario et al.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures




Verification
Claims the paper makes
10 Valid · 1 Not valid · 20 Not checked · 28 Own results and interpretations
At least one cited source doesn’t support this claim or points to a different paper, and none confirmed it — see the verdicts below.
In the paper
“Racial disparities in the frequency of taser use are consistent with such an explanation (Fryer, 2016).”
What the sources say
- Disputedcites Fryer, 2016 sourceExperimental — accuracy not yet measured
Section VI (Interpretation; page unavailable) says there are no racial differences in officer-involved shootings with a Taser or lethal weapon, either in raw data or after controls. Section V reports a racial difference in choosing a pistol rather than a Taser, but that is a comparison between weapons, not a disparity in the frequency of Taser use itself. (LLM, conf=0.84)
Checked by a reasoning model — may be wrong.
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Disputed
Every source the paper cites for this claim was checked and supports it.
In the paper
“In other words, elevated levels of sublethal assault against innocent Black individuals by a subset of police would have the effect of diminishing the apparent severity of antiBlack racial disparities in lethal force conditional on encounter (Ross et al., 2018).”
What the sources say
- Supportedcites Ross et al., 2018 source
The Discussion states the claim nearly verbatim and identifies it as a key finding of the authors’ model. It explains that excessive non-lethal force against Black individuals by a subset of officers can mask anti-Black disparities in lethal force when data are pooled and conditioned on encounters. (LLM, conf=0.95)
hedgedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“it is also the case that unarmed individuals are killed by police at rates that reflect racial disparities.”
What the sources say
- Supportedcites Ross, 2015 source
The Discussion section (“Shot by Police: Across Race/Ethnicity”; page unavailable) reports that, across almost all counties, unarmed people shot by police had a higher probability of being Black or Hispanic than white. The Introduction (“Moving Forward”; page unavailable) also explicitly refers to departments and officers who kill unarmed Black individuals at disproportionate rates. (LLM, conf=0.91)
- Not checked — no full text of the sourcecites Charbonneau et al., 2017 source
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“As Cesario et al. (2019) state, this database is more complete than official federal databases, as police departments underreport to the federal government (Feldman et al., 2017; Klinger et al., 2016; Nix et al., 2017; White, 2006).”
What the sources say
- Partly supportedcites Feldman et al., 2017 sourceExperimental — accuracy not yet measured
The Discussion reports that The Counted identified deaths more effectively than the federal NVSS and had better coverage than earlier federal ARD and FBI estimates. It also notes that many law-enforcement agencies did not report deaths to federal systems, but it does not establish police-department underreporting as the reason The Counted was more complete; for NVSS, it discusses misclassification on death certificates instead. Our copy provides section labels but no page numbers for these passages. (LLM, conf=0.88)
- Not checked — no full text of the sourcecites Klinger et al., 2016 source
- Not checked — no full text of the sourcecites Nix et al., 2017 source
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“causal inference (e.g., Knox et al., 2019; Ross et al., 2018) approaches show that if police are biased in who they encounter, then encounter-conditional approaches can be severely confounded.”
What the sources say
- Not matchedcites Knox et al., 2019
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Supportedcites Ross et al., 2018 source
The Conclusions state that conditioning on encounters can mislead when encounters are themselves an outcome of racial bias. The Discussion further explains that racially disparate encounter rates can mask disparities in encounter-conditional use of lethal force in pooled data. (LLM, conf=0.93)
plainstates a possible cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Not matched
42 more claims · 4 with their cited sources checked
Create a free account to see all 59 claims and every source verdict.
- Probably not a substantial claim. 9
- It may state two separate claims as one. 3
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Self-check results appear here once an independent check has confirmed them.
Cited works that were retracted
Works this paper cites that were later retracted, from the retraction notices publishers deposit. Major when the notice came out before this paper was published.
A note, not an issue: for example, a cited work that was retracted only after this paper was published, which its authors could not have known.
In the paper
“For this reason, anti-Black encounter bias is a confounding factor in recent encounter-conditional studies finding anti-White racial disparities (e.g., Fryer, 2016; Johnson et al., 2019; Worrall et al., 2018).”
Cited work 10.1073/pnas.1903856116 · retracted 2020-07-10 · notice 10.1073/pnas.2014148117 · this paper published 2020-06-18
Citations
Cites
1 retracted
RETRACTED: Officer characteristics and racial disparities in fatal officer-involved shootings
Johnson, D. J., Tress, T., Burkel, N., Taylor, C., & Cesario, J. (2019) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1903856116 · Cited by 83 · cited 4×
“For this reason, anti-Black encounter bias is a confounding factor in recent encounter-conditional studies finding anti-White racial disparities (e.g., Fryer, 2016; Johnson et al., 2019; Worrall et al., 2018).”
BackgroundInfluentialRetractedR: A Language and Environment for Statistical Computing
R Core Team (2025) · doi:10.32614/r.manuals · Cited by 352,181
Stan : A Probabilistic Programming Language
Carpenter, B., Gelman, A., Hoffman, M. D., Lee, D., Goodrich, B., Betancourt, M., Brubaker, M., Guo, J., Li, P., & Riddell, A. (2017) · Journal of Statistical Software · doi:10.18637/jss.v076.i01 · Cited by 7,754 · cited 10×
“Empirical findings aside, Cesario et al. (2019) argue convincingly that to understand racial disparities in killings by police, we have to engage in proper benchmarking, such that we compare the relative rates of police use-of-force against criminals, fB fW , and noncriminals, yB yW , independently. The push to take b…”
BackgroundMethodInfluentialGelman, A., Fagan, J., & Kiss, A. (2007) · Journal of the American Statistical Association · doi:10.1198/016214506000001040 · Cited by 873 · cited 2×
“Both recent and decade-old data show that Black individuals are more likely to be stopped by police than White individuals (Fryer, 2016; Gelman et al., 2007; Miller et al., 2017; U.S. Department of Justice, 2016), even after a variety of statistical controls have been applied.”
BackgroundStatistical Rethinking: A Bayesian Course with Examples in R and Stan
McElreath, R. (2018) · doi:10.1201/9781315372495 · Cited by 726
Self-citationLanguage from police body camera footage shows racial disparities in officer respect
Voigt, R., Camp, N. P., Prabhakaran, V., Hamilton, W. L., Hetey, R. C., Griffiths, C. M., Jurgens, D., Jurafsky, D., & Eberhardt, J. L. (2017) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1702413114 · Cited by 407 · cited 1×
“Such data raise hopes for accountability and detection of discriminatory violations of individual rights (Scheindlin, 2010) and even study of racial bias in respectfulness of language use and escalation or de-escalation of encounters (Voigt et al., 2017).”
BackgroundRoss, C. T. (2015) · PLOS ONE · doi:10.1371/journal.pone.0141854 · Cited by 359 · cited 6×
“Ross (2015) and Charbonneau et al. (2017), for example, show that conditional on being shot by police, a White suspect is more likely to be armed than is a Black suspect.”
BackgroundInfluentialSelf-citationA Bird's Eye View of Civilians Killed by Police in 2015: Further Evidence of Implicit Bias
Nix, J., Campbell, B. A., Byers, E. H., & Alpert, G. P. (2017) · Criminology & Public Policy · doi:10.1111/1745-9133.12269 · Cited by 336 · cited 2×
“As Cesario et al. (2019) state, this database is more complete than official federal databases, as police departments underreport to the federal government (Feldman et al., 2017; Klinger et al., 2016; Nix et al., 2017; White, 2006).”
BackgroundAdministrative Records Mask Racially Biased Policing
Knox, D., Lowe, W., & Mummolo, J. (2019) · doi:10.2139/ssrn.3336338 · Cited by 314
Administrative Records Mask Racially Biased Policing
KNOX, D., LOWE, W., & MUMMOLO, J. (2020) · American Political Science Review · doi:10.1017/s0003055420000039 · Cited by 314
Research works printed but not linked: 2
- Ringrose K., 2019, Virginia Law Review, 105, 57
- Selby N., 2016, In context: Understanding police killings of unarmed civilians
Cited by
A Causal Framework for Cross-Cultural Generalizability
Deffner, D., Rohrer, J. M., & McElreath, R. L. (2022) · Advances in Methods and Practices in Psychological Science · doi:10.1177/25152459221106366 · Cited by 173
“, Kitagawa decomposition) to estimate effects of interventions for counterfactual populations (Acharya et al., 2016; Ciocca Eller & DiPrete, 2018; Kitagawa, 1955; Mize, 2016; Preston et al., 2000; Ross et al., 2021; Storer et al., 2020).”
BackgroundSelf-citationHaines, N., Kvam, P. D., Irving, L. H., Smith, C. T., Beauchaine, T. P., Pitt, M. A., Ahn, W., & Turner, B. M. (2020) · doi:10.31234/osf.io/xr7y3 · Cited by 112
Police Violence and Public Health
DeVylder, J., Anglin, D. M., Bowleg, L., Fedina, L., & Link, B. G. (2021) · Annual Review of Clinical Psychology · doi:10.1146/annurev-clinpsy-072720-020644 · Cited by 133
Faber, S. C., Roy, A. K., Michaels, T. I., & Williams, M. T. (2023) · Frontiers in Psychiatry · doi:10.3389/fpsyt.2023.1098292 · Cited by 121
What can experimental studies of bias tell us about real-world group disparities?
Cesario, J. (2021) · Behavioral and Brain Sciences · doi:10.1017/s0140525x21000017 · Cited by 70
Metadata
- Authors
- Ross, C. T., Winterhalder, B., & McElreath, R.
- Affiliations
- Max Planck Institute for Evolutionary Anthropology (DE)
- University of California, Davis (US)
- Corresponding author
- Ross, Cody T.
- Venue
- Social Psychological and Personality Science
- Publisher
- SAGE Publications
- Published
- 2020-06-18
- Volume
- 12
- Issue
- 3
- Pages
- 323-332
- ISSN
- 1948-5506 (ISSN-L)
- 1948-5514
- OpenAlex
- W3036948988
- Cited by
- 61
- References
- 29
- Citations by year
- 2026: 3 · 2025: 5 · 2024: 12 · 2023: 14 · 2022: 13 · 2021: 8 · 2020: 6
- FWCI
- 35.39
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Locations
- Social Psychological and Personality Science (publishedVersion, cc-by)
- MPG.PuRe (Max Planck Society) (submittedVersion, other-oa)
- Topics
- Social Sciences › Political Science and International Relations › Policing Practices and Perceptions
- Social Sciences › Sociology and Political Science › Criminal Justice and Corrections Analysis
- Social Sciences › Sociology and Political Science › Racial and Ethnic Identity Research
- Keywords
- police use of deadly force, racial disparities, benchmarking, violent crime rates, causal models, anti-Black racism, police killings
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-09-29, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-09-29 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Racial Disparities in Police Use of Deadly Force Against Unarmed Individuals Persist After Appropriately Benchmarking Shooting Data on Violent Crime Rates. https://app.cochat.ai/papers/10.1177/1948550620916071
- The paper
- Ross, C. T., Winterhalder, B., & McElreath, R. (2021). Racial Disparities in Police Use of Deadly Force Against Unarmed Individuals Persist After Appropriately Benchmarking Shooting Data on Violent Crime Rates. Social Psychological and Personality Science. https://doi.org/10.1177/1948550620916071
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.