FBI UCR and NIBRS — what crime statistics actually show (and don't)

Voluntary, not mandatory

Local police agencies submit UCR data voluntarily. Around 18,000 of ~18,500 US law-enforcement agencies typically report, but coverage varies by state and year. When agencies don't report, their populations effectively don't show up in national totals — pulling the apparent crime rate down.

zipradar always shows the reporting status alongside the crime numbers. "Agency did not report to UCR in [year]" is preferable to silent extrapolation.

UCR's two systems: SRS and NIBRS

Summary Reporting System (SRS, 1930–2020): aggregated counts of major offenses. Suffered from the "hierarchy rule" — only the most serious offense in a multi-offense incident was counted.

National Incident-Based Reporting System (NIBRS, 2021–): incident-level detail with all offenses, victim and offender characteristics, and circumstances. The FBI ended SRS in January 2021 and now requires NIBRS.

The transition created a discontinuity. Comparing 2019–2020 to 2021+ requires care: the same incident might have produced one record under SRS and several under NIBRS.

Per-100,000 rates, not raw counts

Always normalize crime numbers to a rate (per 100,000 residents). 5,000 robberies in New York City and 5,000 in Toledo, Ohio are not equivalent — one is a city of 8 million, the other 270,000. zipradar publishes rates, not raw counts.

A 100,000-person threshold is the FBI default. Smaller divisions (per zip, per neighborhood) require more population care.

Compare across years carefully

Methodology changes (UCR/NIBRS transition), agency reporting gaps, statistical noise on small numbers, and policy shifts (e.g., decriminalization of marijuana possession affecting drug-arrest counts) all create false signals when read across years naively.

zipradar flags methodology changes alongside the data so you can interpret in context.