Crime statistics in the United States are collected two ways: local police departments voluntarily send incident records to the FBI, and the Bureau of Justice Statistics surveys households directly about crimes they’ve experienced. Understanding how crime statistics are collected matters because each method captures a different slice of reality, and the gaps between them are where most misreadings begin.
The FBI’s Police-Reported Data
The FBI has been gathering crime numbers from local departments since 1930 through the Uniform Crime Reporting Program.1Federal Bureau of Investigation. Crime/Law Enforcement Stats (Uniform Crime Reporting Program) For most of that history, the data flowed through the Summary Reporting System, which had a built-in blind spot. A “hierarchy rule” required agencies to count only the most serious offense in any single incident. A robbery that escalated into a homicide vanished from the robbery count. Multiply that across millions of incidents a year, and whole categories were systematically undercounted.
The FBI retired the Summary Reporting System in 2021 and replaced it with the National Incident-Based Reporting System, or NIBRS. NIBRS records up to ten separate offenses per incident, so an event involving assault, weapons violations, and property destruction now generates a full record instead of one line.2Federal Bureau of Investigation. Effects of NIBRS on Crime Statistics It also captures details the older system ignored: time of day, location type, victim-offender relationship, and whether the case was eventually cleared.3Community Oriented Policing Services. The National Incident-Based Reporting System (NIBRS) Data for 2022
The transition hasn’t been clean. As of 2024, agencies covering about 82 percent of the U.S. population report through NIBRS.4Bureau of Justice Statistics. National Incident-Based Reporting System (NIBRS) The remaining 18 percent can include large departments whose absence pulls national totals around. Participation is voluntary. Agencies submit either through a state UCR program or directly to the FBI, and no federal law compels them to.1Federal Bureau of Investigation. Crime/Law Enforcement Stats (Uniform Crime Reporting Program) When a major city drops out of the reporting pool for a year, national numbers can look like crime fell without anything actually changing on the street.
How Offenses Get Classified
Standardized classification is what lets anyone compare crime across cities and states. NIBRS sorts offenses into two groups. Group A includes 22 categories covering 46 specific crimes, chosen based on seriousness, national frequency, and whether detailed data can realistically be collected on them. Agencies report full incident-level data for every Group A offense. Group B covers everything else, and agencies report only arrest information for those.5Federal Bureau of Investigation. NIBRS Offense Codes The practical difference for anyone reading the numbers: Group A reflects the total volume of reported incidents, while Group B reflects only how many people police actually arrested.
Under the older Summary Reporting System, offenses were split into Part I and Part II. Part I — sometimes called index crimes — covered serious violent crimes (homicide, rape, robbery, aggravated assault) and major property crimes (burglary, larceny-theft, motor vehicle theft, arson).6Federal Bureau of Investigation. Offense Definitions Part II captured less serious offenses through arrest data only. You’ll still see the Part I/Part II terminology in older reports.
Across both systems, the core split between violent and property crime is the same. A violent crime involves force or the threat of force against a person. A property crime targets money or belongings without direct physical confrontation. Violent crime rates draw the most public attention, though property crimes are far more common.
The Survey That Catches What Police Reports Miss
Police data captures only crimes that someone actually reports to police. The Bureau of Justice Statistics fills that gap through the National Crime Victimization Survey, which goes to households and asks people directly whether they’ve been victimized, regardless of whether they called anyone. Each year, interviewers reach roughly 240,000 people in about 150,000 households.7Bureau of Justice Statistics. National Crime Victimization Survey
The survey exists because a large share of crime never enters any police database. Victims skip reporting for many reasons: fear of retaliation, distrust of police, embarrassment, or a belief that the crime was too minor for anyone to care about. Researchers call the gap between actual crime and reported crime the “dark figure,” and the NCVS is the primary tool for measuring it.
The survey also collects things police reports often miss: the victim’s relationship to the offender, financial losses, whether the victim sought help from any support service, and why they did or didn’t report. That produces patterns in who gets victimized and who stays silent — patterns that don’t appear in FBI data alone.
Specialized Reporting Mandates
Two areas of crime get their own federal collection rules because ordinary reporting tends to miss them.
Hate Crimes
The Hate Crime Statistics Act requires the Attorney General to collect annual data on crimes motivated by prejudice based on race, gender, gender identity, religion, disability, sexual orientation, or ethnicity. Tracked offenses include murder, assault, intimidation, arson, and property destruction when motivated by bias.8Office of the Law Revision Counsel. 34 USC 41305 – Hate Crime Statistics Because flagging bias motivation requires local officers to recognize and code it, these numbers are widely regarded as an undercount.
Campus Crime
Every college or university that participates in federal financial aid must publish an annual security report with three years of campus crime statistics under the Clery Act. Required data covers murder, sexual assault, robbery, aggravated assault, burglary, motor vehicle theft, arson, and domestic violence, plus hate crimes and arrests for drug, alcohol, and weapons violations.9Office of the Law Revision Counsel. 20 USC 1092 – Institutional and Financial Assistance Information for Students Schools must distribute the report to current students and employees and make it available on request. Institutions that substantially misrepresent their numbers face civil penalties of up to roughly $70,000 per violation, a figure the Department of Education adjusts periodically for inflation.
What the Numbers Leave Out
Several forces pull crime data away from crime reality. Voluntary participation is the biggest. If a large department stops submitting, national totals dip without anything changing on the ground. Legislative changes add noise too: when a state reclassifies an offense or broadens a definition, the count moves even though the behavior hasn’t.
Reporting effects run in both directions. When a community rolls out an easy mobile reporting app or runs a campaign encouraging victims to come forward, the statistics may show a spike that reflects better documentation rather than more crime. In areas where residents distrust police, low reporting rates can make a neighborhood look safer than it is. NCVS data consistently shows a significant share of crimes go unreported, so police-based statistics will always understate the true picture.
What “Cleared” Actually Means
Clearance rates — the percentage of crimes police consider solved — are one of the most commonly misread statistics. In FBI reporting, an offense is cleared by arrest when at least one person is arrested, charged, and turned over to the court. An offense can also be cleared by “exceptional means” when police have identified an offender and gathered enough evidence to arrest but something outside their control prevents it: the suspect has died, the victim refuses to cooperate, or another jurisdiction won’t extradite.10Federal Bureau of Investigation. Clearances A cleared case does not mean anyone was convicted. It doesn’t even mean the right person was charged. And recovering stolen property alone doesn’t clear a property crime; someone has to be arrested for it.
Violent crime clearance rates historically run far higher than property crime rates. Police devote more investigative resources to assaults and homicides than to bike thefts, so most property crime cases never lead to an arrest — worth knowing if you’re deciding whether to file a report mainly in hopes of getting your belongings back.
Reading Crime Data Without Getting Misled
A few habits keep you from being fooled. Check the participation rate for the dataset you’re reading. If the FBI reports that crime fell nationally but only 75 percent of agencies submitted data that year, the trend line is weaker than it looks. Compare across several years rather than reacting to a single year’s jump. One year of change can be a reporting anomaly; five years of movement in the same direction is a trend.
When comparing places, look at crime rates per 100,000 residents rather than raw totals. A city of two million will have more total crimes than a town of 50,000, and that doesn’t make it more dangerous per person. The FBI calculates rates by dividing offenses by population and multiplying by 100,000, and the Crime Data Explorer displays both raw counts and per capita rates.11Federal Bureau of Investigation. Crime Data Explorer Ignoring the difference is the single most common way people misread crime statistics, and it drives most of the misleading “most dangerous cities” lists that circulate online.
Finally, treat police data and victimization survey data as complementary rather than competing. The NCVS will almost always show more crime than the FBI’s police-reported data because it captures unreported incidents. Neither number is wrong. They measure different things. Police data tells you what the criminal justice system is handling. Survey data tells you what people are actually experiencing. When the two diverge sharply in a particular category, that’s usually a signal that reporting rates for that type of crime are especially low, which is itself useful information.