Written and maintained by CASRAI Editorial Board
Last updated
The National Crime Victimization Survey (NCVS) is the United States’ main household survey of nonfatal crime, run by the Bureau of Justice Statistics (BJS). Researchers use it to estimate how much crime people experience whether or not they report it to police. This guide covers the design features that matter when you analyse the data, the difference between victimization and prevalence rates, and where the files are held.
What the NCVS is
The NCVS collects information on nonfatal crimes, reported and not reported to the police, against persons age 12 or older (Lauritsen and Rezey, BJS NCJ 241656, 2013). Because it asks people directly, it measures crime that never enters police records, which is the main reason victimization surveys exist alongside police statistics.
Crimes covered, per the BJS NCVS page (read 2026-11-07), are nonfatal personal crimes (rape and sexual assault, robbery, aggravated and simple assault, personal larceny) and household property crimes (burglary and trespassing, motor vehicle theft, other theft). Homicide is outside the survey’s scope because the survey covers nonfatal crime only.
Who runs it and what it covers
BJS sponsors the NCVS and the U.S. Census Bureau carries out the data collection. Each year the survey reaches about 240,000 people in roughly 150,000 households. The population is persons age 12 and older in U.S. households, including group quarters such as dormitories. Military bases, institutional settings and homeless people are excluded.
Design features that shape analysis
- Rotating panel. A household stays in the sample for 3.5 years, and eligible members are interviewed every 6 months, for seven interviews. The first interview is usually in person; later ones are in person or by phone. For panel concepts generally see panel surveys.
- Six-month reference period. Each annual estimate covers crimes in the prior 6 months, excluding the interview month.
- A data year spans more than 12 months of fieldwork. According to BJS, the 2024 survey covers July 1, 2023 to November 30, 2024. Do not treat a “2024” estimate as events in calendar 2024 only.
Victimization rate versus prevalence rate
A victimization rate counts incidents relative to the population. A prevalence rate counts the share of persons or households victimized at least once. Lauritsen and Rezey (2013) show why the distinction matters: from 1993 to 2010 the violent victimization rate fell 76% while the prevalence rate fell 63%, because some people are victimized repeatedly. The share of violent victims with two or more victimizations fell from 23% to 17%, yet that 17% accounted for 54% of violent victimizations in 2010. Intimate-partner violence victims were repeat victims 21% of the time, against 9% for victims of stranger violence.
For analysts: use the victimization rate when the question is about the volume of crime events, and the prevalence rate when it is about how many people or households are affected. Reporting only one can hide the repeat-victimization pattern.
Finding and using NCVS data
The BJS NCVS page links to codebooks and datasets held by the National Archive of Criminal Justice Data (NACJD) at ICPSR, plus data analysis tools, technical documentation, and participation-rate and population-count spreadsheets. It also provides variance estimation guides with example code for SAS, SUDAAN, SPSS and Stata, a State-Level Analysis User’s Guide, and the report “Criminal Victimization in the 22 Largest U.S. States, 2020-2022”.
Analyst checklist
- Start from the BJS NCVS page, then follow its link to the NACJD/ICPSR files for the release you need.
- Read the technical documentation and the variance estimation guide for your software before computing standard errors. See survey weighting for the general ideas.
- Check how the data year maps to the interview period (see above).
- Confirm your outcome is within scope: nonfatal, age 12 and older, household-resident population.
- BJS lists a redesigned 2024 instrument, with its estimates to follow later, so check BJS documentation for comparability before building a long time series across the redesign.
Strengths and limits
- Strength: captures crimes whether or not they were reported to police.
- Strength: a large national sample with a panel structure, which allows repeat victimization to be studied.
- Limit: nonfatal crime only; no homicide.
- Limit: excludes children under 12, military bases, institutional settings and homeless people.
Alternatives and companions
| Source | What it measures |
|---|---|
| NCVS | Nonfatal victimization of persons 12+ and households, reported or not to police; household survey |
| Police-recorded crime statistics (FBI UCR and NIBRS) | Crimes known to law enforcement agencies |
The two answer different questions and are usually read together. For the wider field these data serve, see what is criminology and what is criminal justice.
FAQ
How does the NCVS collect data?
The Census Bureau interviews eligible members of sampled households every 6 months for seven interviews over 3.5 years, usually in person at first and in person or by phone afterwards.
Why are crime victimization surveys used?
They capture crimes not reported to police, which police-recorded statistics cannot include.
Where are NCVS results and data published?
BJS publishes reports and links to datasets through its NCVS page; the files are archived at NACJD/ICPSR.
Does the NCVS include homicide?
No. It covers nonfatal crime against persons age 12 or older.
Sources
- Bureau of Justice Statistics, National Crime Victimization Survey (NCVS), read 2026-11-07.
- Lauritsen, J. L. and Rezey, M. L., Measuring the Prevalence of Crime with the National Crime Victimization Survey, BJS NCJ 241656, September 2013, read 2026-11-07.
