Arrests by race and sex, and what the data cannot support

Law enforcement agencies reported 7,139,243 arrests to the FBI in the most recent year it has published in full. This is what those arrests look like broken down by the arrestee's race and sex, what population each rate divides by, and — at as much length — what the figures do not establish.

An arrest is not a crime, a conviction, or a person

An arrest is an event. One person arrested three times in a year is three arrests, so none of these counts is a count of individuals, and none of these rates is the share of a population that was arrested.

An arrest is also not a finding. It records that an agency took someone into custody; it says nothing about whether a charge followed, whether a court agreed, or whether the person had done anything.

By race

	Arrests	Population	Per 100,000
White	4,499,167	247,263,993	1,820
Black or African American	2,180,417	46,248,715	4,715
American Indian or Alaska Native	149,669	—	2,124
Asian	114,458	—	484
Native Hawaiian or Other Pacific Islander	37,024	—	3,880
Unknown	158,164	no denominator	—

The rate divides by the population of that group across all origins. That choice is not cosmetic and it is explained below.

By sex

Men accounted for 5,186,051 arrests and women 1,953,192 — rates of 3,092 and 1,122 per 100,000 residents.

Why the denominators are what they are

The FBI publishes no ethnicity field with these arrests. Not an empty one: the field does not exist. So there is no Hispanic or Latino arrest figure to publish, and an arrest recorded as White may be of a Hispanic person or not — the record does not say.

That determines the denominator. Each rate divides by the population of that race across all origins, not by the non-Hispanic part of it. Using the non-Hispanic population would divide a numerator that includes Hispanic arrestees by a denominator that excludes them, and would inflate every rate in the table.

Three categories have no rate at all. Unknown and Not Specified have no population to divide by. The FBI's Multiple is an arresting officer's classification of one person, where the Census Bureau's "two or more races" is what a person says about themselves; the two are not the same population, so their ratio is not a rate.

Race is recorded by the arresting agency, not reported by the person arrested.

What the data does not support

These figures cannot tell you who commits crime. They record who was arrested, by agencies that chose to report, in the places those agencies cover. An arrest rate is the product of what happened, what was reported to police, where police were, whom they stopped, and whom they took into custody. This data separates none of those.

Comparing the composition of arrestees with the composition of the population assumes two things at once: that offending is equal across groups, and that the likelihood of being arrested for the same act is equal across groups. This data establishes neither, and nothing here should be read as evidence for either.

Reporting is voluntary and uneven. A jurisdiction whose agencies did not submit is absent from every row above, and the agencies that submit are not a random sample of the country.

344 arrests carry no race category at all — not even “unknown” — so the race rows total very slightly less than the arrests recorded.

This brief reports what the FBI published and stops there. Where a reader wants to know why any of these figures is what it is, the honest answer is that this data cannot say.