Methodology and Data Sources
This analysis relies exclusively on state-mandated, publicly reported data collected and published by New York State pursuant to Executive Law § 837-t, which requires every police department and sheriff’s office to report qualifying use-of-force incidents to the New York State Division of Criminal Justice Services (DCJS). The statute further requires DCJS to make incident-level data by agency available to the public. No proprietary datasets, advocacy estimates, or third-party modeling tools were used.
Use-of-Force Data
Use-of-force figures are drawn from the DCJS case-level use-of-force dataset covering incidents from November 2020 through December 2024, with analysis focused on calendar years 2021–2024 due to a documented change in DCJS data-collection methodology in late 2020. Incidents prior to 2021 are excluded to ensure comparability across years.
The unit of analysis for racial impact is subjects, not incidents. This follows DCJS’s own reporting framework and reflects the fact that a single use-of-force incident may involve multiple individuals. Subject-level demographic variables (race and ethnicity) are reported by agencies as part of the mandated submission.
Importantly, the DCJS dataset captures incidents involving the display or use of firearms, electronic control weapons, chemical agents, impact weapons, and chokeholds or similar restraints. As DCJS itself acknowledges, incidents involving physical force alone (e.g., hands-on force) are only reportable when they result in serious injury or involve a chokehold. As a result, less than one percent of reported force statewide falls into this category. This limitation is disclosed here not to discount the data, but to accurately describe its scope.
Arrest Data (Exposure Proxy)
To contextualize use-of-force figures, the analysis incorporates adult arrest totals by agency and year from state-published historical arrest tables. Arrests are used as an exposure proxy—not as a measure of criminality—to address the frequently raised claim that disparities in force merely reflect differences in police interactions. Juvenile arrest data is excluded to avoid distortions arising from differing enforcement and reporting practices.
Where racial likelihood ratios are discussed, they compare Black and White subjects using consistent numerators (force subjects) and denominators (adult arrests), following the same comparative logic employed in prior investigative reporting.
Geographic Scope
The analysis examines three jurisdictions using the same statutory framework and data structure:
Nassau County (all reporting agencies)
Suffolk County (all reporting agencies)
New York City (New York City Police Department)
All figures are derived from incidents occurring within each jurisdiction, as reported to DCJS.
What This Analysis Does—and Does Not—Claim
This methodology does not assess officer intent, individual culpability, or the legality of any single encounter. It evaluates patterns and outcomes produced by institutional policy choices over time. Nor does it claim that use-of-force data captures every form of coercive police contact. It claims only what the state’s own data supports: how often force is reported, against whom, and how those patterns change—or fail to change—after reform efforts are announced.
By grounding the analysis in the state’s own compulsory reporting system, this methodology ensures that any conclusions drawn reflect not advocacy judgment, but the documented outputs of New York’s policing institutions themselves.
Findings Overview
Executive Summary
Five years after New York announced a sweeping effort to reform police use of force, the state’s own data tells a different story. Across Long Island and New York City, police continue to use force at racially disparate rates, and those disparities have persisted or widened during the very period in which reform was supposed to take hold.
Using state-mandated, incident-level reporting compiled by the Division of Criminal Justice Services, this analysis examines use-of-force outcomes between 2021 and 2024 for Nassau County, Suffolk County, and the New York City Police Department. The findings are consistent across jurisdictions and scale: force increased, racial disparities remained entrenched, and reform efforts produced no measurable structural correction.
On Long Island, police subjected significantly more people to force in 2024 than in 2021, with the sharpest increases occurring in Suffolk County. Black residents were subjected to force at rates far exceeding their representation in the population, and those disparities cannot be explained by arrest patterns alone. Even when arrests are used as a proxy for police exposure, Black individuals experienced force at multiples that arrest likelihood does not justify.
New York City, operating under the same statutory framework but at vastly greater scale, does not provide a counterexample. The NYPD’s data reflects the same core pattern: racial disparities in use of force that persist across years, despite public commitments to reform, revised policies, and expanded reporting.
These outcomes are not the product of data gaps or analytical ambiguity. They are drawn from the state’s own compulsory reporting system—designed, defined, and published by New York itself. Nor do they hinge on the intent or conduct of individual officers. The patterns documented here reflect institutional outputs, not isolated misconduct.
The central failure is structural. New York’s post-2020 reform effort relied on procedural compliance rather than enforceable standards. Agencies were required to submit plans, but not to meet benchmarks. Data was collected, but not tied to consequences. Disparities were documented, but not corrected. The result is a reform regime that measures outcomes without demanding improvement.
This analysis does not argue that police use of force is always unjustified, nor does it claim to capture every coercive encounter. It demonstrates something narrower—and more consequential: that New York’s approach to police reform has not altered the racial distribution of force, even after years of public scrutiny and legislative attention.
Reform without enforcement is not reform. It is delay, dressed as progress.
I. Reform Without Enforcement: Why New York’s System Was Built to Stall
When New York announced police reform in 2020, the promise sounded straightforward: rethink how force is used, confront racial disparities, and rebuild public trust. What followed looked busy—plans were drafted, meetings were held, policies were revised. But activity is not the same thing as change. And five years later, the state’s own data shows that the outcomes reform was supposed to address remain largely intact.
That is not because reform failed midway. It is because reform, as designed, was never structured to compel success.
In the wake of national protests and intense public pressure, then-Governor Andrew M. Cuomo issued Executive Order 203. The order required every police department and sheriff’s office in New York to review its use-of-force policies and develop a reform plan addressing racial bias and disproportionate policing. Those plans were to be submitted to local governing bodies for approval.
At first glance, this appeared meaningful. Departments were required to engage with the issue. Communities were invited to participate. The language of the order spoke of urgency and reform. But the structure of the mandate told a different story.
The order required submission, not results. It did not require departments to reduce force, narrow racial disparities, or meet any measurable benchmarks. It did not empower the state to reject a plan because it was inadequate. It did not authorize audits, penalties, funding consequences, or corrective directives if outcomes failed to improve. Once a plan was filed and approved locally, the state imposed no continuing obligation to show that anything had actually changed.
This distinction—between process and enforcement—is the fault line that runs through New York’s reform effort.
At the same time the state was encouraging departments to “rethink” their practices, it already had something far more concrete in place. For years, New York law has required police agencies to report use-of-force incidents in detail to the Division of Criminal Justice Services. Those reports include what type of force was used, where it occurred, and the race and ethnicity of the people subjected to it. The data is published annually and broken down by agency.
In other words, by the time reform was announced, New York already possessed a system capable of answering the most basic questions reform was meant to confront: how often force is used, and against whom.
What the state did not do was connect those answers to obligation.
This separation was not accidental. It reflects a broader pattern in how institutions respond to crisis. Transparency is offered in place of accountability. Data collection is treated as progress. The act of measuring a problem becomes a substitute for fixing it.
Under this model, disparities can be acknowledged year after year without triggering action. Agencies can point to revised policy language while producing the same outcomes. Leaders can cite reform plans as evidence of good faith, even as the data shows no structural change. Reform becomes something that is performed, not enforced.
This approach is appealing precisely because it diffuses responsibility. When no benchmarks exist, no one technically fails. When no consequences attach to outcomes, persistence can be explained away as complexity, culture, or time. And when reform is defined as participation rather than improvement, the system can declare itself responsive while remaining fundamentally unchanged.
By 2021, New York had clear, legally mandated data showing how force was being used across the state. By 2024, that same data showed that racial disparities persisted across counties and across the state’s largest police department. The question, then, is not whether leaders knew. It is whether the reform framework ever required them to act on what they knew.
Reform without enforcement does not malfunction. It operates exactly as designed. It absorbs pressure, generates documentation, and allows institutions to move forward without altering the underlying distribution of harm.
Section II turns to what the data itself shows once reform is measured not by paperwork or promises, but by outcomes.
II. What the Law Required the State to Measure—and What the Data Reveals
The patterns described in this section are not the product of voluntary disclosure or journalistic investigation. They exist because New York law requires them to exist.
Long before the state announced police reform in 2020, the Legislature enacted Executive Law § 837-t, which mandates that every police department and sheriff’s office in New York report qualifying use-of-force incidents to the Division of Criminal Justice Services. The statute requires reporting at the incident level, by agency, and includes demographic information for each person subjected to force. DCJS, in turn, is required to compile and publish that information annually.
This statutory framework matters because it defines what the data is—and what it is not. The use-of-force data analyzed here is not anecdotal, incomplete by choice, or selectively released. It is the formal output of a compulsory reporting system, created by law, administered by the state, and relied upon by policymakers when reform was announced.
In 2020, when then-Governor Andrew M. Cuomo issued Executive Order 203, the state already had this reporting regime in place. The executive order did not create use-of-force data collection; it assumed it. Reform was premised on the idea that the state could identify disparities and respond to them. The data was the foundation.
What the data shows, once examined over the first full reform period from 2021 through 2024, is not uncertainty but continuity.
Across Nassau County, Suffolk County, and New York City, the number of people subjected to police force did not decline during the reform era. In some jurisdictions, it increased substantially. These are not marginal fluctuations. They represent hundreds of additional individuals subjected to force during a period in which restraint and de-escalation were publicly emphasized.
Equally significant is the racial distribution of that force. Year after year, Black residents accounted for a disproportionately large share of people subjected to police force. This pattern did not meaningfully narrow over time. It appeared across different counties, across different policing structures, and across agencies of vastly different size.
This persistence is legally meaningful. A single year of disparity might be dismissed as noise. Multiple years, across jurisdictions, within a statutory reporting system designed to surface precisely these outcomes, cannot.
New York City underscores the point rather than complicating it. The New York City Police Department operates under the same statutory reporting obligation as every other agency in the state. It is subject to the same definitions, the same data architecture, and the same public disclosure requirements. Despite differences in scale, leadership, and resources, the city’s data reflects the same core pattern: racial disparities in use of force that persist across years.
The reporting system itself also exposes an important structural limitation, one acknowledged by DCJS in its own publications. Under the statute and implementing rules, incidents involving physical force alone—such as hands-on restraint—are generally reportable only when they result in serious injury or involve a chokehold or similar restraint. As a result, these forms of force comprise a very small fraction of reported incidents statewide.
This limitation does not undermine the conclusions drawn here. It places them in context. The dataset captures the most formally recognized and potentially dangerous categories of force. If racial disparities are evident even within that constrained universe, the absence of lower-level physical force from the data does not suggest equality at those levels. It suggests that the documented disparities represent a minimum, not a ceiling.
Taken together, the data required by Executive Law § 837-t demonstrates something fundamental: the reform framework announced in 2020 did not alter the distribution of police force across racial lines. The same disparities visible at the outset of reform remained visible years later, despite full statutory reporting and public disclosure.
Section III turns to arrest data—not as a justification for these disparities, but as a test of the claim that they merely reflect differences in police exposure rather than differences in treatment.
III. Arrests as an Alibi: Why “Exposure” Arguments Persist—and Why the Law Has Rejected Them
When evidence shows that police use force disproportionately against Black residents, the most common response is familiar: police are simply encountering different populations at different rates. This argument appears reasonable on its face. It sounds empirical. It sounds neutral. And it has been repeated so often that it now passes as common sense.
But in law and in practice, this explanation has a long history—and a long record of failure.
To evaluate claims about exposure, this analysis introduces adult arrest data as a point of comparison. Arrests are not treated here as a measure of criminal behavior. They are used for a narrower and more defensible purpose: as a proxy for police contact. Arrests reflect discretionary enforcement decisions made by officers and agencies, and they approximate how often individuals are drawn into encounters where force might plausibly occur.
This is not a novel approach. Courts, regulators, and civil-rights investigators have relied on exposure controls for decades to test whether disparities reflect neutral conditions or discriminatory application. The point is not whether arrest data is perfect—it is whether disparities in force track disparities in enforcement or exceed them.
When arrest data is placed alongside use-of-force data, the result is revealing.
Across Nassau County, Suffolk County, and New York City, racial disparities in police use of force are greater than racial disparities in arrests. Black individuals are more likely to be subjected to force relative to White individuals than they are to be arrested relative to White individuals. If arrests fully explained force, those ratios would move together. They do not.
This divergence is not accidental. It reflects the fact that force is not a necessary consequence of arrest, nor is arrest a prerequisite for force. Force is a discretionary escalation—governed by training, supervision, culture, and institutional tolerance. When disparities widen at the point of force, they indicate that discretion is being exercised unevenly.
The persistence of the “exposure” argument despite this evidence is not about data. It is about narrative convenience.
Historically, exposure-based explanations have been used to rationalize disparities in education, housing, employment, and criminal justice. In each context, the argument follows the same structure: unequal outcomes are attributed to unequal inputs, and institutional responsibility is displaced onto communities rather than systems. Over time, courts have recognized this pattern and rejected exposure explanations that merely restate disparities rather than explain them.
Policing is no exception. Arrest data does not absolve disparities in force; it tests them. And when force disparities exceed arrest disparities, the implication is clear: something is happening during encounters that is not explained by who police encounter, but by how those encounters are handled.
This distinction is legally and practically significant. Under New York law, police departments are required to report both arrests and use-of-force incidents through formal state systems. The state therefore possesses the information necessary to identify when force outcomes diverge from enforcement patterns. The fact that such divergence persists is not a failure of measurement—it is a failure of response.
It is also important to be precise about what this analysis does not claim. It does not suggest that every use of force is unjustified. It does not assume bad faith by individual officers. And it does not treat arrest data as a moral benchmark. Instead, it asks a narrower question: once police contact occurs, how is discretion exercised?
When that question is answered with data rather than assumption, the exposure narrative collapses. Disparities that survive exposure controls point toward institutional practices governing escalation, not toward community behavior.
At that point, the conversation can no longer remain at the level of explanation. It must move to accountability.
Section IV turns to how those institutional practices are obscured by aggregation—and why averaging harm across agencies has become one of the system’s most effective forms of self-protection.
IV. Averaging Harm: Why Aggregation Is Not Neutral—and Never Was
When the public is told that racial disparities in police use of force exist “countywide” or “citywide,” the language sounds comprehensive. It suggests that the problem has been captured in full. But that sense of completeness is misleading. Aggregation does not merely summarize harm—it repackages it.
To understand why aggregation matters, it helps to step away from policing for a moment. In almost every area of public life, averaging is used to stabilize perception. Economic inequality disappears when income is reported as a mean. School segregation fades when districtwide enrollment is emphasized over classroom composition. Environmental exposure looks manageable when regional averages replace neighborhood-level data.
Policing is no different.
On Long Island, aggregation collapses a fragmented policing landscape into a single statistical surface. Nassau and Suffolk Counties contain dozens of distinct law-enforcement agencies—county police departments, sheriff’s offices, town and village departments, park police, and task forces—each with its own leadership, training priorities, and tolerance for escalation. These agencies do not operate interchangeably. They do not produce identical outcomes. Yet aggregation treats them as though they do.
This flattening has consequences.
When use-of-force data is presented at the county level, the public sees disparity but not distribution. A handful of agencies may account for the majority of force incidents, while others contribute little. Some departments may show extreme disproportionality even with low overall volume. Others may drive total numbers while appearing less skewed individually. Aggregation obscures both patterns at once.
What results is a form of statistical camouflage. Harm is visible, but responsibility is not.
This matters because accountability does not attach to abstractions. Counties do not train officers—departments do. Counties do not supervise patrols—command structures do. Counties do not decide when force is acceptable—policies and supervisors do. When data is averaged upward, the level at which reform must occur is rendered indistinct.
Aggregation also reshapes how failure is explained. A countywide disparity can be attributed to population mix, call volume, or geography. An agency-specific disparity demands a harder question: why this department, under this leadership, with these policies, produces these outcomes. Aggregation keeps the conversation safely general.
New York City illustrates the same phenomenon at a different scale. The New York City Police Department is a single legal entity, but it is operationally decentralized. Precincts vary widely in enforcement intensity, demographic composition, leadership culture, and informal norms around escalation. Yet public discussion almost always treats the NYPD as a unitary actor.
Citywide use-of-force statistics confirm that disparities persist. But they do not reveal where force is concentrated, how supervisory practices differ, or which precincts reproduce the same outcomes year after year. The scale that gives the NYPD its visibility also gives it cover. Variation disappears into magnitude.
This is not an accident of data presentation. It aligns neatly with how institutions prefer to be evaluated.
Aggregation allows leaders to speak in averages rather than specifics. It permits acknowledgment without ownership. A disparity can be conceded while responsibility is distributed so widely that no single actor is compelled to respond. Reform becomes collective and therefore optional.
There is also a deeper legal implication. Civil-rights accountability turns on policy, custom, and deliberate indifference—concepts that require identifying where decisions are made and tolerated. Aggregation interferes with that inquiry. It turns patterns of conduct into ambient conditions rather than traceable outputs of governance.
In this way, aggregation functions less as a descriptive tool than as a buffer. It absorbs the shock of bad outcomes and releases them back into the system as generalized concern. Harm is acknowledged, but its sources are blurred. The public is told there is a problem, but not where it lives.
That is why transparency alone does not disrupt entrenched practices. Data can be accurate and still be strategically incomplete. When information is presented at a level that diffuses responsibility, it becomes compatible with inaction.
Aggregation, then, is not just a statistical choice. It is a political one. It determines whether disparities lead to targeted reform or dissolve into managed concern. It decides whether institutions must change—or merely explain.
Section V turns to the consequence of this design choice: how transparency without enforcement and specificity becomes not a catalyst for reform, but a mechanism for delay.
V. Reading the Numbers as Evidence, Not Explanation
When police departments report use of force to the state, they are not offering opinions. They are complying with a legal obligation. The numbers that emerge from that obligation are therefore not interpretations of policing; they are records of what policing does.
This section treats those numbers accordingly.
Between 2021 and 2024, police departments across New York were required to report each qualifying use-of-force incident to the state, identifying the agency involved and the individual subjected to force. These reports are standardized. They use common definitions. They apply across counties, cities, and departments of wildly different size. This uniformity is what allows the data to be read not as local storytelling, but as comparative institutional output.
The first thing the numbers make clear is that the reform period did not produce a downward shift in the use of force. When the data is read year over year, there is no sustained pattern of reduction across Long Island or New York City. In some jurisdictions, the number of people subjected to force increases. In others, it fluctuates. What does not appear is a consistent reform trajectory.
That absence is itself evidence.
Reform, if it is functioning, leaves traces. It appears unevenly at first. It produces partial corrections. It creates volatility before settling into new baselines. None of that is visible here. Instead, the numbers stabilize. They repeat. They reproduce.
More telling than volume, however, is distribution.
Across the period studied, Black residents account for a share of force subjects that far exceeds their share of the population. This is not a marginal disparity, and it is not confined to a single year or a single department. It appears repeatedly, across counties and in New York City, under different leadership structures and enforcement environments.
This persistence matters because it changes how the data must be interpreted. A one-year disparity might invite debate about conditions or anomalies. A four-year pattern, produced under standardized reporting rules, demands a different explanation. It points away from momentary factors and toward systemic practice.
At this stage, it is essential to understand what these numbers do not represent. They are not crime rates. They are not arrest counts. They are not population surveys. They are records of state-sanctioned physical coercion — moments when officers escalated encounters to the point of deploying weapons, chemical agents, electronic control devices, or other forms of reportable force.
That distinction is often lost in public discussion, where force data is casually blended with crime statistics. The law does not treat force as a byproduct of crime. It treats force as a separate decision, governed by training, policy, supervision, and discretion. The numbers reflect those decisions.
The structure of the dataset reinforces this point. Each “force subject” represents a person, not an incident count inflated by multiple officers or multiple tools. This design choice matters because it keeps the focus on impact rather than activity. The question is not how busy officers were, but how many people were subjected to force.
When the data is read through that lens, its consistency becomes harder to dismiss. The same communities appear again and again as the recipients of force. The same racial imbalance reasserts itself across jurisdictions. Whatever explanations are offered — population density, call volume, neighborhood conditions — they do not disrupt the pattern.
New York City’s inclusion in the analysis underscores this point. The city is often treated as sui generis: too large, too complex, too different to compare. Yet when the same legally required reporting framework is applied, the city’s data aligns with the broader pattern rather than breaking from it. Scale changes magnitude, not direction.
This is where the numbers cease to be descriptive and become diagnostic.
A dataset that captures only the most formally recognized uses of force — excluding most hands-on physical coercion — nonetheless reveals persistent racial disparities. That means the disparities are not artifacts of over-reporting or definitional creep. They survive even within a constrained reporting universe. The implication is not speculative: what is visible is likely less than what exists.
At this point in the analysis, the numbers have done their first job. They establish that the problem is not isolated, not declining, and not confined to one type of agency. They show a stable pattern produced under stable rules.
The next question is the one institutions most often raise in response: whether these disparities merely reflect differences in police contact rather than differences in how force is applied. That question cannot be answered by assertion. It requires comparison.
Section VI takes that step by placing use-of-force data alongside arrest data — not to excuse the disparities, but to test whether they are proportionate to enforcement exposure or exceed it.
VI. Arrest Data and the Myth of Neutral Exposure
To understand why arrest data is so frequently invoked to explain racial disparities in police use of force, it is necessary to first understand what arrest data represents—and what it does not.
An arrest is often spoken of as though it were a natural event, something that simply “happens” when a law is broken. In reality, an arrest is the end point of a long sequence of discretionary decisions. An officer must decide to initiate contact, decide that further investigation is warranted, decide that the encounter cannot be resolved informally, decide that a custodial response is appropriate, and decide that arrest—rather than warning, summons, or disengagement—is the correct outcome. Each of these decisions is shaped by training, supervision, departmental culture, and institutional expectations.
Arrest data, then, is not a record of crime in the abstract. It is a record of how police choose to enforce the law.
This distinction matters because arrest statistics are routinely treated as neutral baselines in public debate. When disparities in force are raised, arrest data is offered as an explanation: police use force more often against Black residents, the argument goes, because they encounter Black residents more often in enforcement contexts. Arrests are presented as proof of exposure.
But exposure is not a natural condition. It is itself produced.
This is why arrest data has long occupied an uneasy position in civil-rights analysis. Courts and regulators have recognized that enforcement statistics cannot be taken at face value as objective measures of underlying behavior. They reflect the choices institutions make about where to police, whom to stop, and how aggressively to intervene. For that reason, arrest data is not used to excuse disparities; it is used to test them.
The logic of that test is straightforward but often misunderstood.
If disparities in police use of force were simply the mechanical result of differing levels of police contact—if force followed exposure in a proportional way—then the racial distribution of force would closely track the racial distribution of arrests. Where arrest likelihood rises, force likelihood would rise in roughly the same proportion. Where arrest likelihood falls, force likelihood would fall accordingly.
This expectation does not require perfect alignment. Policing is not a mathematical system. But it does require directional consistency. Exposure-based explanations only hold if disparities remain stable as enforcement intensifies.
What happens when that expectation is applied to the data is where the exposure narrative begins to unravel.
Across jurisdictions examined in this analysis, racial disparities do not remain constant as encounters escalate. They widen. The likelihood that Black individuals are subjected to police force increases at a greater rate than their likelihood of being arrested, relative to White individuals. In other words, the disparity is not introduced at the point of arrest alone. It intensifies at the point of force.
This is the critical analytic moment, and it is worth lingering on it.
Force is not synonymous with arrest. Many arrests occur without force. Many uses of force occur without arrest. Force is an escalation within an encounter, not a necessary outcome. When disparities expand at this stage, they cannot be explained by contact frequency alone. They reflect how discretion is exercised during encounters.
Institutions often resist this conclusion by reframing arrest data as a proxy for danger rather than exposure—suggesting that arrests signal situations where force is more likely to be necessary. But this move collapses two distinct concepts. Arrest does not measure threat; it measures enforcement choice. The law itself recognizes this distinction. Police authority to use force is governed by standards of necessity and proportionality, not by arrest status.
This is why arrest comparisons are used in civil-rights litigation and oversight. They are not meant to absolve institutions. They are meant to reveal whether escalation decisions are distributed evenly or unevenly across groups once enforcement has begun.
When disparities grow larger at later decision points, the inference is not that exposure explains the disparity, but that institutional practices governing escalation are operating differently.
The persistence of the exposure argument despite this logic is revealing. It functions less as an analytic claim than as a narrative shield. By locating disparity upstream—in neighborhoods, demographics, or “crime patterns”—it shifts attention away from the discretionary moment where force is actually applied. It turns institutional decisions into environmental inevitabilities.
But arrest data does not support that move. Properly understood, it narrows the field of explanation rather than expanding it. Once exposure is accounted for, what remains is not mystery but responsibility.
This does not require an assumption of malicious intent. Institutions do not need to be consciously biased to produce biased outcomes. Discretion can be uneven without being explicit. Culture can guide behavior without formal instruction. Supervision can tolerate patterns without endorsing them. The law recognizes this. That is why it looks to outcomes, not motives.
By the time arrest data is placed alongside use-of-force data, the question changes. It is no longer whether police encounter different communities at different rates. It is whether those encounters are governed by the same rules of escalation.
The data suggests they are not.
Section VII turns to how that disparity is maintained—not through explicit policy mandates, but through supervision, training, and institutional tolerance that allow unequal escalation to persist without formal acknowledgment.
VII. The Numbers: Long Island and New York City (DCJS Use-of-Force, 2021–2024)
This section establishes the numerical baseline for reported police use of force between 2021 and 2024. The figures presented here are drawn from incident-level reporting submitted by law enforcement agencies to the Division of Criminal Justice Services (DCJS) pursuant to state law. The data counts force subjects—individuals against whom reportable force was displayed or used during an incident—rather than incidents themselves.
Nassau County Police Department
Between 2021 and 2024, the Nassau County Police Department reported 972 force subjects. Over this four-year period, the annual number of force subjects increased from 166 in 2021 to 229 in 2024, representing an increase of approximately 38 percent.
Across the full period, racial identification of force subjects shows that 480 individuals were identified as Black and 433 as White, with the remaining subjects classified as Asian, other racial categories, or unknown. Within Nassau County’s reported universe of serious force, Black individuals constitute the largest single racial group of force subjects during this period.
Suffolk County Police Department
The Suffolk County Police Department reported 918 force subjects between 2021 and 2024. Annual totals rose from 166 force subjects in 2021 to 266 in 2024, an increase of approximately 60 percent over the period.
Racial counts across the four years indicate 388 Black force subjects and 473 White force subjects, with additional subjects recorded as unknown or within smaller racial categories. While Suffolk reported slightly fewer total force subjects than Nassau over the period, the rate of increase in force usage was steeper.
New York City Police Department
The New York City Police Department reported 7,015 force subjects between 2021 and 2024. Annual force-subject counts increased from 1,468 in 2021 to 2,062 in 2024, an increase of approximately 40 percent.
Across the four-year period, 4,652 force subjects were identified as Black and 1,749 as White, with the remainder recorded as unknown or within smaller racial categories. In absolute terms, Black individuals account for a substantial majority of all reported force subjects in New York City during this period.
Scope and Function of the Data
These figures reflect counts generated by a statewide, statutorily mandated reporting system. They do not purport to capture every form of physical interaction between police officers and civilians, nor do they include uses of physical force that fall outside DCJS reporting thresholds. The limitations of the reporting framework are acknowledged and apply uniformly across jurisdictions.
The purpose of this section is descriptive rather than interpretive. It establishes the volume and distribution of reported force by agency over the 2021–2024 period. Comparative analysis—particularly the relationship between force usage, arrests, and population—is addressed in the sections that follow.
Population-based likelihood measures presented later in this analysis rely on American Community Survey (ACS) 5-year pooled estimates, using adult (18+) population by race as denominators. This approach ensures consistency across jurisdictions and aligns with standard demographic practices used in policy analysis and investigative reporting.
Standing alone, raw counts do not answer questions of proportionality or explanation. To assess whether these force outcomes reflect differences in enforcement exposure or differences in escalation, the figures above must be evaluated against arrest patterns and population benchmarks. That analysis follows.
VIII. Force Likelihood Versus Arrest Likelihood: Testing the Exposure Claim
The figures set out in Section VII establish how often reportable force was used and against whom. They do not, by themselves, resolve the most common explanatory claim offered in response to racial disparities in policing: that differences in use of force merely reflect differences in police “exposure” to certain populations. This section addresses that claim directly by placing use-of-force outcomes alongside arrest patterns and population benchmarks.
The logic of the comparison is straightforward. Arrests represent a downstream point of police contact—one that already incorporates discretion, enforcement priorities, and decision-making. If disparities in use of force were simply the product of differential exposure, then disparities in force would track disparities in arrests. The relationship would be proportional. Where arrest likelihood increases, force likelihood would increase at roughly the same rate.
That is not what the data shows.
Method of Comparison
For each jurisdiction, force likelihood and arrest likelihood are calculated using the same structure. The numerator is the number of individuals subjected to reportable force or arrested during the relevant period. The denominator is the adult population (age 18 and over) by race, drawn from American Community Survey (ACS) 5-year pooled estimates, which provide the most stable demographic baselines for multi-year analysis.
This produces two ratios for each jurisdiction:
Force likelihood (Black/White) — the rate at which Black adults are subjected to police force relative to White adults.
Arrest likelihood (Black/White) — the rate at which Black adults are arrested relative to White adults.
The comparison does not assume that arrests are neutral or bias-free. It uses arrests precisely because they are a conservative proxy for police contact. If force disparities exceed arrest disparities, then exposure alone cannot account for the difference.
Long Island: Nassau and Suffolk Counties
On Long Island, the divergence between force likelihood and arrest likelihood is pronounced.
In both Nassau and Suffolk Counties, Black adults are arrested at higher rates than White adults, relative to population. That disparity is real and measurable. However, when force outcomes are examined, the disparity widens substantially.
In Nassau County, the Black/White force likelihood ratio approaches eleven to one, while the corresponding arrest likelihood ratio is markedly lower. The escalation from arrest disparity to force disparity is not incremental; it is structural. The point at which force is introduced amplifies the disparity rather than merely reflecting prior enforcement patterns.
Suffolk County exhibits the same pattern. Although arrest disparities exist, the likelihood of being subjected to force diverges more sharply than the likelihood of being arrested. Here again, the data shows that the racial gap grows at the point of force.
This widening gap is the critical analytical finding. It indicates that use of force is not a mechanical byproduct of arrest frequency. It is a separate decision point—one governed by training, policy, supervision, and institutional tolerance for escalation.
New York City
New York City provides a larger dataset but follows the same trajectory.
Across the 2021–2024 period, Black New Yorkers were arrested at higher rates than White New Yorkers relative to adult population. Yet the Black/White force likelihood ratio exceeds the arrest ratio by a substantial margin, mirroring the pattern observed in Nassau and Suffolk.
The scale of NYPD operations eliminates the possibility that these results are statistical noise or artifacts of small numbers. When a disparity persists across thousands of force subjects and multiple years, it reflects systemic practice rather than episodic deviation.
The significance of New York City’s data is not that it differs from Long Island’s, but that it aligns with it. Different leadership structures, different geographies, and different political environments yield the same outcome: racial disparities intensify at the point where force is exercised.
What the Comparison Establishes
This comparison does not rely on assumptions about intent, officer motivation, or individual encounters. It tests a single proposition: whether differential exposure explains differential force.
The answer, based on the data, is no.
If exposure were the primary driver, force likelihood would mirror arrest likelihood. Instead, force likelihood consistently exceeds arrest likelihood across jurisdictions. The disparity is not inherited from arrest patterns; it is produced during encounters.
That distinction matters. It shifts the locus of explanation away from background conditions and toward institutional decision-making at the moment of escalation. It is at that moment—when officers choose whether and how to apply force—that disparities widen most sharply.
Limits and Implications
The analysis here is deliberately bounded. It does not claim that arrests are unbiased. It does not purport to explain every instance of force. It demonstrates something narrower and more defensible: that racial disparities in police use of force cannot be explained away by arrest rates alone.
Once that claim fails, explanations rooted solely in exposure lose analytical force. What remains are questions about policy design, training doctrine, supervisory oversight, and accountability mechanisms—questions that are institutional rather than individual.
Those questions are taken up next.
IX. Concentration, Fragmentation, and Institutional Responsibility
Sections VII and VIII establish two propositions that are no longer in dispute. First, the use of reportable police force increased between 2021 and 2024 across Nassau County, Suffolk County, and New York City. Second, racial disparities in force cannot be explained by arrest patterns alone; they intensify at the point of escalation. What remains is an institutional question: where responsibility for these outcomes resides.
This section addresses that question by examining how force is distributed across agencies and what that distribution reveals about governance rather than individual behavior.
Concentration of Force Within Agencies
Use-of-force outcomes are not evenly dispersed across the policing landscape. They are concentrated within a small number of large agencies that account for the overwhelming majority of reported force subjects in their respective regions.
On Long Island, Nassau County Police Department and Suffolk County Police Department dominate the force record by design. Each operates as a centralized, countywide law enforcement body with thousands of sworn officers, broad patrol authority, and limited external oversight. The numerical concentration of force within these departments is therefore not surprising. What is significant is that the same departments repeatedly generate racially disparate outcomes over multiple years, despite public commitments to reform.
New York City presents an even clearer case of concentration. The New York City Police Department accounts for the vast majority of reported force subjects statewide, simply by virtue of its size. Yet scale alone does not explain the racial composition of those outcomes. The persistence of disparity across years and precincts indicates that force is not merely a function of volume but of institutional norms governing escalation.
Fragmentation as Insulation
New York’s policing structure is fragmented. Hundreds of municipal departments report force data to the state, but responsibility for outcomes is diffuse. This fragmentation functions as insulation rather than accountability.
When disparities appear at the statewide level, responsibility is diluted across agencies. When they appear at the county or city level, they are attributed to local conditions. When they persist over time, they are reframed as intractable social facts. At no point does the structure compel sustained institutional correction.
On Long Island, fragmentation takes a different form. County departments operate with broad authority, yet external oversight is limited. Reform plans mandated by executive order were approved locally, without substantive state review. No agency lost funding for noncompliance. No standardized benchmarks were enforced. The result was uniformity in form and divergence in outcome.
In New York City, internal fragmentation—across precincts, bureaus, and commands—serves a similar function. Responsibility for training, supervision, and discipline is dispersed internally, even as outcomes remain consistent. This allows disparities to be treated as localized anomalies rather than systemic outputs.
Governance, Not Individual Error
The concentration of force within large agencies, combined with the fragmentation of oversight, clarifies the nature of the problem. These outcomes do not arise from a handful of aberrant officers. They arise from organizational design.
Use of force is governed by policy manuals, training curricula, supervisory practices, and disciplinary thresholds. When disparities persist across years and leadership changes, they reflect choices embedded in those systems. When disparities widen at the point of force rather than arrest, they reflect how discretion is structured and rewarded.
Institutional responsibility follows from institutional control. Agencies that design training, authorize tactics, and evaluate performance are responsible for the predictable outputs of those systems. The data presented in this analysis shows that those outputs have remained stable despite reform rhetoric.
The Consequence of Unassigned Responsibility
When responsibility is not clearly assigned, reform becomes performative. Plans are written, listening sessions are held, data is published, and outcomes continue unchanged. The absence of enforcement mechanisms ensures that disparities can be acknowledged without being corrected.
This is not a failure of measurement. New York has required use-of-force reporting for years. It is a failure of governance. Data without consequence does not constrain institutional behavior; it legitimizes it.
The final question, then, is not whether disparities exist or how they are measured. It is whether the legal and policy framework governing policing in New York assigns responsibility commensurate with authority.
That question is addressed in the concluding section, which examines the legal implications of persistent disparity in the face of statutory reporting, constitutional constraints, and declared reform obligations.
X. Legal Consequence and Institutional Exposure
The preceding sections establish a record that is no longer descriptive alone. When a government entity mandates reporting, publishes the results, and repeatedly reproduces the same disparities over time, the question shifts from whether reform has occurred to what legal consequences attach to its absence.
This section addresses that question.
Statutory Obligation and Notice
New York law requires law enforcement agencies to report use-of-force incidents to the state. That requirement is not symbolic. It creates a formal record, places agencies on notice, and supplies the state with ongoing information about how force is used and against whom.
Once disparities are documented and persist across multiple reporting cycles, they cease to be unforeseen. They become known conditions. At that point, continued inaction is no longer plausibly characterized as ignorance or transition. It is acquiescence.
Notice matters because it changes the legal posture of the institution. A municipality that is aware of a pattern and fails to intervene does not occupy the same position as one confronting a newly discovered problem. The record compiled through DCJS reporting establishes knowledge—year after year—at both the agency and state levels.
Constitutional Exposure
Persistent racial disparities in the use of police force implicate constitutional protections even in the absence of explicit intent. The Equal Protection Clause does not require proof that every officer acted with discriminatory animus. It requires examination of whether state action, viewed systemically, produces racially disparate outcomes that are tolerated, reinforced, or left uncorrected by policymakers.
When disparities are stable, measurable, and resistant to reform efforts, courts look beyond individual incidents to patterns and practices. The record presented here—spanning multiple jurisdictions and years—meets the threshold for such inquiry.
Similarly, the Fourth Amendment’s reasonableness standard does not operate in a vacuum. What is deemed “reasonable” force is shaped by training, supervision, and institutional norms. When those norms repeatedly yield disparate outcomes, they invite scrutiny not only of individual encounters but of the frameworks that authorize them.
Municipal Liability and Policy Failure
Under established principles of municipal liability, a government entity may be held responsible for constitutional violations arising from official policy, custom, or deliberate indifference. Deliberate indifference does not require hostility. It requires awareness coupled with failure to act.
Here, the conditions are present. The state required reporting. Agencies complied. Disparities were revealed. Reform plans were adopted procedurally but left unenforced substantively. No corrective benchmarks were imposed. No funding consequences followed. Outcomes remained unchanged.
In that context, disparity is not an accident. It is an output of governance.
The relevance of this conclusion is not confined to litigation. It bears on oversight authority, legislative responsibility, and executive accountability. When reform is mandated without enforcement, the mandate itself becomes a shield rather than a remedy.
The Role of Reform Without Enforcement
Executive orders and reform plans are often described as evidence of good faith. They may be. But good faith does not negate responsibility when outcomes are known and persistent.
Reform without enforcement alters the legal landscape in a specific way. It demonstrates that the state recognized a problem, articulated a response, and declined to ensure its effectiveness. That sequence matters. It weakens claims that disparities are inevitable or beyond institutional control.
In this sense, the failure of reform efforts is not neutral. It compounds exposure by confirming that alternative courses were available and not pursued.
What the Record Ultimately Shows
The data presented in this analysis does not accuse individual officers. It does not reduce complex encounters to single causes. It does something more consequential: it documents institutional continuity in the face of acknowledged disparity.
Once that continuity is established, the legal question is no longer speculative. It becomes whether the frameworks governing policing in New York are adequate to their stated purpose, and whether entities entrusted with coercive power are being held to account for how that power is exercised.
Those questions do not belong solely to courts. They belong to legislatures, executives, oversight bodies, and the public itself. The record compiled here supplies the factual predicate for that inquiry. What follows is not a matter of interpretation, but of responsibility.
Conclusion: What Accountability Requires
The record assembled here does not ask the reader to choose between competing narratives. It presents a sequence of facts generated by the state’s own reporting regime, measured over time, across jurisdictions, and against stable demographic benchmarks. That sequence leads to a single, unavoidable conclusion: the continued production of racial disparity in police use of force in New York is no longer a failure of knowledge, but a failure of accountability.
Reform was announced. Reporting was mandated. Data was collected. Disparities persisted. At each stage, the state and its subdivisions retained the authority to intervene—and declined to do so in any enforceable way. The result is not ambiguity. It is continuity.
Accountability, in this context, does not require speculation about intent or motive. It requires recognition of responsibility for outcomes that are foreseeable, measurable, and repeatedly reproduced. When institutions exercise coercive power, knowledge of harm carries an obligation to correct it. When that obligation is unmet, the law does not treat the result as accidental.
What the data shows is not simply that force is used disproportionately, but that the mechanisms designed to prevent that outcome have failed to function. Whether that failure is addressed through litigation, legislation, oversight, or public demand is a question of institutional will, not evidentiary sufficiency.
The record is now established. What follows will determine whether reform remains rhetorical—or becomes real.
Reader Supplement
To support this analysis, I have added two companion resources below.
First, a Slide Deck that distills the core legal framework, case law, and institutional patterns discussed in this piece. It is designed for readers who prefer a structured, visual walkthrough of the argument and for those who wish to reference or share the material in presentations or discussion.
Second, a Deep-Dive Podcast that expands on the analysis in conversational form. The podcast explores the historical context, legal doctrine, and real-world consequences in greater depth, including areas that benefit from narrative explanation rather than footnotes.
These materials are intended to supplement—not replace—the written analysis. Each offers a different way to engage with the same underlying record, depending on how you prefer to read, listen, or review complex legal issues.