Harm and protection

Published October 2026 · Data window: MPD assault reports via FBI NIBRS, 2022–2025, updated yearly

The question: Are Black women in DC assaulted at a higher rate than other women, and do their cases end in an arrest as often?

The answer: Assaulted at a higher rate: yes. MPD's reports put assaults on Black women at 4,205 per 100,000 a year, 3.2× Hispanic women's rate and 10.2× White women's rate. Cleared by arrest as often: mixed. Compared like with like, their cases end in an arrest 0.86 times as often as Hispanic women's and 0.90 times as often as White women's. The gap sits in partner and other family assaults and not in stranger or unknown ones, so no single direction is asserted.

MPD's public incident feed records no victim's race or sex; its reports to the FBI do. In 2022 to 2025, 56,092 people were recorded as victims of aggravated or simple assault. This page measures who was harmed, not who offended: nothing here describes offenders, and a higher rate of harm is not a reason to police a neighborhood more. Whether enforcement follows reported crime is tested on the crime-baseline page.

4,205 per 100,000

Assault reports with a Black woman victim, per 100,000 Black women a year, 2022 to 2025: 3.2× Hispanic women's rate (1,304), 10.2× White women's rate (412) and 10.1× Asian women's rate (415). Reports, not people: someone assaulted twice counts twice.

Who is harmed

Source: MPD incident reports as submitted to the FBI (NIBRS), aggravated (13A) and simple (13B) assault, individual victims, 2022 to 2025; women victims: 26,341 Black, 2,110 Hispanic, 2,056 White and 289 Asian. Residents: ACS 2024 5-year.

Race is recorded by officers. If every multiracial Black resident were recorded as Black (there are 9% more people who are Black alone or in combination), the ratios would be 3.0× Hispanic women's and 9.4× White women's. Hispanic ethnicity is often missing in NIBRS: if women recorded as White with unknown ethnicity were Hispanic, the ratios would be 2.2× Hispanic women's and 19.1× White women's. The Asian women comparison moved most between the window's two halves (9.2× then 11.2×), so it is reported but not headlined.

By how serious the assault was: Black women's rate as a multiple of each group's.

Assaults	vs Hispanic women	95% interval	vs White women	95% interval	Victims (Black / Hispanic / White)
Simple assault, no serious injury	3.1×	2.9 to 3.2	9.7×	9.3 to 10.2	22,727 / 1,913 / 1,858
Assault with serious injury	4.6×	3.9 to 5.5	13.8×	11.6 to 16.5	2,367 / 133 / 137
Assault with a gun	8.0×	5.7 to 11.4	23.3×	16.8 to 33.0	1,110 / 36 / 38
Homicide	5.2×	1.9 to 19.4	12.7×	5.2 to 40.3	80 / 4 / 5

Thin base. Homicide rests on 4 Hispanic women and 5 White women, so its intervals are wide.

If the gap came from Black women reporting more, it would narrow for the assaults most often reported. Against White women it widens instead, from 9.7× for simple assault without serious injury to 23.3× for assaults with a gun. Nationally, 40% of simple assaults reach police against 75% of crimes involving a gun (Criminal Victimization, 2024 (Tapp and Coen, BJS, September 2025, NCJ 310547)).

Neighborhood accounts for part of the gap. Placing MPD's public incident locations in census tracts, the tracts where Black women live predict an aggravated assault rate 2.8× White women's; it is 14.4×, leaving a factor of about 5.2 that this check does not place. 43.1% of Black women live in tracts where a quarter or more of residents are poor, against 6.2% of White women. Tract averages blur differences between blocks, so what is left over marks the limit of this check, not something beyond place.

Whether those cases end in an arrest

An arrest within 90 days of the incident, from MPD's arrest records in the same files, for women victims. The test was written down before it was run. DC law requires an arrest in domestic and family cases when there is probable cause (D.C. Code § 16-1031), and the relationship mix differs by group, so cases are compared like with like: inside each relationship and assault type.

By the rule set before the test, the answer is mixed against Hispanic women and mixed against White women: a direction is asserted only when the like-with-like interval excludes 1 and partner and stranger assaults point the same way. The gap sits in partner and other family assaults: in partner assaults, 44.1% of Black women's cases end in an arrest, against 59.2% of White women's and 57.3% of Hispanic women's (ratio 0.74, 95% interval 0.69 to 0.81, against White women). There is no clear difference for stranger or unknown relationships. It holds in both halves of the window. It sits in same-day arrests (0.88), not later ones (1.06).

Relationship	Black women: N, arrested	Hispanic women: N, arrested	White women: N, arrested	Black vs Hispanic	95% interval	Black vs White	95% interval
Partner	8,623, 44.1%	639, 57.3%	453, 59.2%	0.77	0.72 to 0.83	0.74	0.69 to 0.81
Other family	3,953, 57.4%	198, 66.7%	56, 71.4%	0.86	0.78 to 0.95	0.80	0.68 to 0.95
Known, not family	7,587, 36.4%	429, 41.3%	319, 35.4%	0.87	0.77 to 0.97	1.01	0.87 to 1.18
Stranger	1,304, 41.1%	251, 36.7%	367, 38.4%	1.09	0.92 to 1.30	1.05	0.91 to 1.21
Unknown	4,874, 15.2%	593, 15.7%	861, 14.4%	0.97	0.79 to 1.18	1.04	0.87 to 1.24
All, like with like	26,341, 38.4%	2,110, 40.8%	2,056, 33.4%	0.86	0.81 to 0.90	0.90	0.85 to 0.96

Source: MPD incident and arrest records as submitted to the FBI (NIBRS), 2022 to 2025; women victims of aggravated and simple assault; 1,098 with unknown race left out and counted. Ratios are Mantel-Haenszel risk ratios across relationship by assault type, with the Greenland-Robins 95% interval.

Groups and relationships are in a fixed order; this is not a ranking.

Raw rates point the other way, which is why the comparison is made like with like: before matching, 38.4% of Black women's cases end in an arrest against 33.4% of White women's, because more of them are partner assaults (32.7% against 22.0%), where the law requires an arrest, and fewer have an unknown relationship (18.5% against 41.9%), where arrests are rare.

Cases with Black women victims are closed as “victim refused to cooperate” 2.10 times as often as White women's, like with like (95% interval 1.54 to 2.87; 6.4% against 1.9% before matching). That is the reason police recorded; it cannot separate a victim's choice, or her safety, from how the case was handled.

How these numbers are built (and where they're soft)

Source. MPD's own incident reports as submitted to the FBI's National Incident-Based Reporting System, 2022 to 2025: the only public record of MPD cases with the victim's race and sex. Aggravated and simple assault, individual victims of any age. Left out and counted: 5,678 Metro Transit Police victims, 3,067 assaults on officers, 18,691 intimidation victims (threats with no attack) and 2,844 victims whose race or sex is unknown or outside the compared groups.
Annual, not live. Unlike most pages on this site, this one is not recomputed on page load: the FBI publishes each year's file once. The numbers are computed by published code in the DC assault analysis (commit d4232ed) and stored here; the page is refreshed when the FBI releases the next year.
Rates. Reports per 100,000 residents a year, against ACS 2024 5-year population by race and sex. Reports, not people. Residents are the denominator, so visitors and commuters count in the numerator only; leaving out victims recorded as non-residents gives 3.6× Hispanic women's rate and 11.6× White women's rate.
Race. Recorded by officers. A victim is Hispanic when the ethnicity field says so, whatever the race, otherwise the recorded race, which matches the census tables. The race-coding and missing-ethnicity bounds above are the worst cases, stated beside the numbers.
The arrest test. Written down before it was run, with its decision rule. Arrest within 90 days, the same follow-up for every year; compared inside relationship by assault type because D.C. Code § 16-1031 makes arrest mandatory in intrafamily cases with probable cause.
What an arrest rate cannot say. It is not a measure of police effort. It depends on whether the person who did it was still there, whether there was probable cause, the victim's cooperation and the evidence; exceptional clearances reflect victims and prosecutors as well as police.
Who was harmed, not who offended. Victim records say nothing about offenders, and police data holds only what was reported; unreported assaults stay invisible here.
Full report. Every check behind these numbers (severity, neighborhood, missing race, the two halves of the window) with its caveats: the full report.