Four Different Things Called Crime Data
Every crime statistic counts one of four things. Offenses known to police are crimes someone reported and an officer recorded. Arrests are people taken into custody, which depends on whether police identified a suspect and chose to arrest. Victim surveys ask a national sample of households what happened to them, whether or not anyone called the police. Convictions and prison counts are the people a court found guilty and sentenced. Each layer is smaller than the one before it, and different decisions shape each one.

The FBI collects the first two from local police, since 2021 through its National Incident-Based Reporting System. The Bureau of Justice Statistics runs the National Crime Victimization Survey, which found 23.3 violent victimizations per 1,000 people age 12 or older in 2024, and 11.2 per 1,000 that reached the police: about half. Death certificates and police reports both count homicides, the one crime that is hard to hide.
When someone says a group “commits” a share of crime, the figure almost always comes from arrests. Arrests measure who was caught and charged, and many crimes are never solved. Definitions move the numbers too: in 2013 the FBI rewrote its definition of rape for the first time in 80 years, so rape counts before and after that year do not line up.
The Sex Gap
No gap in crime data is larger or more consistent than the one between men and women. In 2025, arrest data in the FBI’s Crime Data Explorer show that men and boys were 72.6 percent of the 7.16 million people arrested, 78.5 percent of those arrested for the four serious violent crimes (murder, rape, robbery, and aggravated assault), 88.9 percent of those arrested for murder and nonnegligent manslaughter, and 97.1 percent of those arrested for rape. The FBI’s 2019 tables show nearly the same split.
Every other source points the same way. Men are about 93 percent of people in state and federal prisons. In The Violence Project’s database of public mass shooters, 98 percent are male. Men are also most of the dead: 77 percent of homicide victims in 2024 were male, and the United Nations found the same pattern worldwide.
Researchers point first to age. Offending climbs in the teens and falls in the early twenties, a pattern called the age-crime curve, and the male curve sits far above the female one. Evolutionary psychologists Margo Wilson and Martin Daly argued that risk-taking and status competition among young men drive much of the gap, a view known as the young male syndrome. Sociologists stress how boys are raised and what they are rewarded for. The hormone story is weaker than its reputation: meta-analyses find only small links between testosterone and aggression.
Women’s offending has a different shape. Women were 37.5 percent of people arrested for larceny-theft in 2025 and 43.9 percent of those arrested for embezzlement, and the number of women behind bars grew more than sevenfold from 1980 to 2023. When women are killed, the killer is far more often a partner: in 2024, 37.5 percent of female homicide victims were killed by an intimate partner, compared with 6.9 percent of male victims.
Source: FBI, Crime Data Explorer, national arrest data, 2025. Arrests, not convictions or offenders.
Sexual Violence: Reported, Unreported, and False
“Women make up rape accusations all the time.”
The best studies put false reports at about 2 to 10 percent of reported sexual assaults. Meanwhile, the Bureau of Justice Statistics found that only 24 percent of rapes and sexual assaults in its 2024 survey were reported to police at all. The bigger gap in the data is the crimes nobody reports.
Sexual violence is where arrest data say the least. In the Bureau of Justice Statistics survey, only 24 percent of rapes and sexual assaults in 2024 were reported to police, down from 46 percent in 2023. Most never become an arrest, so the sex of the person arrested tells only part of the story.
Victim surveys fill in some of it. In the CDC’s National Intimate Partner and Sexual Violence Survey, 94 percent of women who had been raped reported only male perpetrators. Men’s experiences differ by type: most male rape victims reported male perpetrators, while most men who were made to penetrate someone reported female perpetrators.
False reports exist, and they are a small share. A ten-year study of every sexual assault reported at one university coded 5.9 percent as false, and its authors put the range across earlier research at 2 to 10 percent. The group most often accused in online arguments about bathrooms and locker rooms, transgender people, shows up in national survey data as victims of violent crime at four times the rate of everyone else.
Race in Arrest Data
“Black people are 13 percent of the population but commit 52 percent of murders.”
FBI data do show Black Americans are overrepresented in murder arrests: about 56 percent in 2025. Those are arrests, not proof of who killed whom, and half of killings of Black victims never lead to an arrest. Most killings happen within one race, and researchers trace the gap mainly to concentrated poverty and segregation.
Arrest data also show large racial disparities. In 2025, among arrestees whose race was recorded, 31.2 percent were Black, 64.4 percent were white, 2.2 percent were American Indian or Alaska Native, and 1.6 percent were Asian. Black Americans are about 13 percent of the population. The gap is widest for murder (56.5 percent of arrestees with a recorded race) and robbery (57.4 percent). It is smallest for drug possession (17.8 percent) and drunk driving (17.5 percent).
Those figures come with a caveat built into the forms. The FBI’s race categories do not separate out Hispanic people, who are counted mostly as white, and in the 2019 tables, the last full national arrest tables the FBI published by race and ethnicity, agencies left ethnicity blank on about 920,000 of 6.8 million arrests.
The murder figure is the one that travels. White supremacists turned it into a number code, “13/52,” that the ADL lists as a hate symbol. The code leaves out what the arrest figure measures, who the victims are, and what researchers have found about why the gap exists.
Do Arrests Match What Victims Report?
A disparity in arrests could reflect a disparity in offending, a disparity in policing, or both. The Bureau of Justice Statistics tested this directly. In a 2021 report, statistician Allen Beck compared FBI arrest data for 2018 with victims’ descriptions of their attackers in the national victimization survey. Black people were 33 percent of those arrested for nonfatal violent crime, 35 percent of the offenders victims described in incidents reported to police, and 29 percent of offenders across all incidents. For rape, robbery, and aggravated assault, he found no statistically significant difference by race between the offenders victims described and the people police arrested.
So for serious nonfatal violence, victims’ own accounts show a real gap in offending, not only in arrests. Two other facts belong next to it. Most violence happens within a racial group: in the FBI’s 2019 homicide data, 79 percent of white victims were killed by white offenders and 89 percent of Black victims by Black offenders. And the people harmed most by the gap are Black Americans themselves.
Who Is Harmed
Black Americans are killed at far higher rates than any other group. In the Bureau of Justice Statistics’ 2024 homicide report, published in October 2026, the homicide rate for Black Americans was 18.3 per 100,000, more than eight times the rate for white Americans (2.2). Black Americans were 52.5 percent of all homicide victims. American Indian and Alaska Native people (6.8) and Hispanic people (5.6) also died by homicide at higher rates than white people.
Violence against Native women stands out even in national surveys. A National Institute of Justice study found that 84.3 percent of American Indian and Alaska Native women had experienced violence in their lifetimes, and most of the perpetrators were of another race.
Police violence follows the same lines. Of the 10,430 people shot and killed by police from 2015 to 2024 in The Washington Post’s Fatal Force database, 26.9 percent of those whose race was recorded were Black, about twice the Black share of the population, and 95 percent were male.
What Explains the Racial Gap
“The numbers prove some races are just more violent.”
Researchers who followed nearly 3,000 young people in Chicago found that neighborhood conditions, parents’ marital status, and immigrant generation explained more than 60 percent of the Black-white gap in violence and all of the Latino-white gap. The causes of violence look the same across races. The conditions people live in do not.
The leading explanation in criminology is about places, not people. In 1995, sociologists Robert Sampson and William Julius Wilson argued that racial gaps in violence come largely from the concentrated disadvantage that segregation piled into Black neighborhoods, and that the root causes of crime are the same for every race. They did not deny that the gap in offending was real; they argued about where it comes from.
Later studies tested the idea. In Chicago, Sampson and colleagues found that the odds of violence were 85 percent higher for Black young people than white ones, and that family and neighborhood factors explained most of that gap. Ruth Peterson and Lauren Krivo’s study of neighborhoods in 91 large cities tied crime to segregation and spatial inequality. David Kirk found that disadvantage in families and neighborhoods explained much of the racial gap in arrests, and other researchers found that experiences of discrimination predict delinquency in Black youth.
Environment can be literal. Economist Jessica Wolpaw Reyes linked the removal of lead from gasoline to the violent crime decline of the 1990s; a 2023 review of individual-level studies found mostly supporting evidence along with some conflicting results. Lead exposure was heaviest in poor urban neighborhoods, which ties it back to the same map.
Policing, Charging, and Sentencing
The other half of the story is what the system does with the people it catches. Where offending rates are similar, enforcement is not. The ACLU found that in 2018, Black people were 3.64 times as likely as white people to be arrested for marijuana possession, though the two groups used marijuana at similar rates. The 2025 drug possession figures cut the other way: Black people were 17.8 percent of possession arrestees, which is why the comparison has to be drug by drug and place by place.

Traffic stops show the same pattern. The Stanford Open Policing Project analyzed nearly 100 million stops and found that police searched Black and Hispanic drivers on less evidence than white drivers, and that fewer Black drivers were stopped after dark, when an officer cannot easily see a driver’s race.
In federal court, the U.S. Sentencing Commission found that from 2017 to 2021, Black men received sentences 13.4 percent longer than white men. Until 2010, federal law treated one gram of crack cocaine like 100 grams of powder. And among people later proven innocent, the National Registry of Exonerations found that innocent Black Americans were seven times as likely as innocent white Americans to be convicted of serious crimes.
Crimes the Usual Charts Leave Out
Arrest tables count street crime well and other crime poorly. The FBI’s arrest categories have no line for price-fixing, wage theft, or pollution, and in 2025 the FBI’s data held only 85,758 fraud arrests and 7,235 embezzlement arrests among 7.16 million. The people who commit crimes that rarely end in handcuffs rarely appear in the racial and sex breakdowns either.
Some crimes carry a different profile. In 2019, 52.5 percent of known hate crime offenders whose race was reported were white. The ADL counted at least 13 people killed by domestic extremists in 2024, every one by right-wing extremists, and the Center for Strategic and International Studies found that right-wing attacks and plots made up the majority of U.S. terrorist incidents from 1994 to 2020. Mass shooters, as the sex data showed, are almost all men.
Immigrants and crime is its own subject, with its own data problems, and the sibling page on immigration arguments takes it up in full.
The Long Decline, the Spike, and the Drop
“Crime is out of control and getting worse every year.”
The FBI’s 2025 data show murder fell 18.1 percent in one year, to a rate of 4.1 per 100,000, tied for the lowest since the 1950s. Violent crime fell 9.3 percent. The Council on Criminal Justice found big-city homicides down another 18 percent in the first half of 2026.
Crime in the United States peaked in the early 1990s. From 1991 to 2014, the national murder rate fell by more than half, from 9.8 to 4.4 per 100,000. In 2020, murder rose by nearly 30 percent, the largest one-year increase on record. Then it fell. In August 2026, the FBI reported that murder dropped 18.1 percent in 2025 to 4.1 per 100,000 people, tied with 1955 and 1956 for the lowest rate since 1900, and the Council on Criminal Justice found big-city homicides down another 18 percent in the first half of 2026.
Public belief lagged behind. In Gallup’s polling, 77 percent of Americans said crime was rising nationally in 2023; by 2026, 46 percent still said so, as murder fell to its lowest rate in about 70 years.
Politics shapes which numbers people hear. The center-left think tank Third Way reported that murder rates were higher in states Trump won; the Heritage Foundation answered that the gap reverses when the same data are cut by county. Both can be true, which is the point: the unit of analysis decides the headline. The sharpest warning about crime forecasts came in the 1990s, when the “superpredator” prediction helped drive harsher juvenile laws just as juvenile crime began to fall.
How Crime Numbers Became an Argument About Race
Using crime statistics to argue about race is older than the FBI. In 1896, the American Economic Association published Frederick Hoffman’s Race Traits and Tendencies of the American Negro, a statistical case that Black Americans had “a greater tendency to crime” than whites. Hoffman was the statistician for the Prudential Insurance Company.
Historian Khalil Gibran Muhammad traced what followed in The Condemnation of Blackness: reformers and academics built a “statistical discourse” that treated Black crime figures as evidence about Black people, while European immigrant groups accused of the same things were not judged the same way. The pattern runs through the coded “law and order” politics covered in It Worked Last Time and into today’s number memes.
The tell is what gets left out. A statistic used as propaganda drops the denominator, drops the victims, drops the clearance rate, and drops every researcher who studied the cause. It keeps one number and lets the reader supply the rest.
Checking a Crime Statistic in Five Minutes
Ask which of the four layers it counts. “Commit” almost always means “were arrested for.” If the claim is about murder, remember that about half of homicides with Black victims and a third with white victims were not cleared in 2024.
Check the race categories. In FBI tables, most Hispanic arrestees are counted as white. In BJS and census tables, “white” usually means non-Hispanic white. Comparing one to the other inflates or hides gaps.
Find the year and the coverage. A chart that crosses 2013 is mixing two definitions of rape. Numbers from 2021 come from the first year of the FBI’s new reporting system. Pull the current figure yourself from the FBI’s Crime Data Explorer or the BJS website; both are free.
Look for the denominator and the victims. A share of arrests without the share of victims, or a count without a rate, is half a statistic. And when a number arrives as a code, a meme, or a screenshot with no source, treat it the way Hoffman’s readers should have treated his.