1 Criminology and the Measurement of Crime
Learn what criminology studies, how crime is classified and measured in the United States, and what crime statistics can—and cannot—show.
What studies
is the interdisciplinary, empirical study of crime: how it is defined, how and where it occurs, its causes and consequences, and how people and institutions respond to it. It draws on fields such as sociology, psychology, law, economics, and statistics. focuses more directly on institutions such as police, courts, and corrections, although the fields overlap.
Crime is not simply a fixed list of harmful acts. In legal terms, it generally means conduct prohibited by criminal law, and laws and definitions can differ across jurisdictions and change over time. Researchers therefore need to specify what they as crime and how they classify it. A change in recorded crime may reflect a change in behavior, but it may also reflect a change in law, reporting, recording, or enforcement.
Ways of measuring crime
No single measure captures all crime. U.S. researchers commonly compare official police data with victimization surveys; self-report studies provide another perspective.
Police-reported data come from law-enforcement agencies through the FBI’s program. The records aggregate counts for selected offenses. The records details about individual incidents, including offenses, victims, known offenders, and incident circumstances. NIBRS can multiple offenses in an incident, while the traditional SRS hierarchy rule generally counts only the most serious offense.
The Bureau of Justice Statistics’ interviews a representative sample of people age 12 or older in U.S. households about selected crimes, whether or not they were reported to police. It measures nonfatal personal crimes and household property crimes, and asks about incident circumstances and reporting.
In self-report studies, researchers ask people to report their own offending. These studies can reveal behavior that never comes to police attention, but answers depend on what respondents remember and are willing to disclose.
Counts, rates, and survey estimates
A is the number of recorded crimes. A relates that number to a population, often per residents, making places or years with different population sizes easier to compare. A survey can also estimate , the share of people victimized, and , the number of victimization events.
These measures answer different questions and should not be treated as interchangeable. Before comparing statistics, check the measure, population, time period, and denominator.
How crimes are classified
A common broad distinction is between , involving force or the threat of force, and , involving the taking or damaging of property. U.S. FBI categories include specific offenses such as homicide, robbery, assault, burglary, and theft. Statistical systems may group or define offenses differently, and NIBRS provides more offense detail than the traditional summary system.
Classification depends on the purpose and rules of a data system, not only on what happened. For example, if an incident includes a robbery and an assault, a summary system applying the hierarchy rule may only the more serious offense, while NIBRS can record both. Figures from different systems may therefore not be directly interchangeable.
Interpreting crime statistics
Crime data are indicators, not a complete or self-explanatory picture of crime. Police data describe offenses that come to police attention and are recorded by participating agencies. NCVS estimates include selected victimizations that were not reported to police, but cover a different set of offenses and populations. The two sources are complementary; a difference between their estimates does not automatically mean either is wrong.
Suppose police-recorded assaults rise in a city. The increase might reflect more assaults, but it could also reflect greater willingness to report, changes in police recording practices, or improved agency data submission. To interpret a trend, compare relevant sources and definitions, examine reporting and coverage, and consider contextual information. A correlation between a social condition and a crime , by itself, does not establish that the condition caused the crime.
Limits and cautions
Police data have several limits. Many crimes are never reported to police, and agencies differ in reporting completeness and recording practices. UCR participation by state and local agencies is not federally mandated, and some agencies may not submit complete data. Changes in participation or data quality can affect comparisons over time or between places.
Victimization surveys have different limits. The NCVS relies on respondents’ accounts and is subject to sampling error, nonresponse, recall, question wording, and interviewer or survey-mode effects. It does not cover every offense or every person: its household survey population excludes people under age 12 and people who are homeless or institutionalized, and it cannot survey a homicide victim.
Recorded crime reflects both behavior and social response. Laws determine which conduct is criminal; victims’ choices affect what is reported; and institutional practices affect what is recorded and acted upon. Comparisons should use consistent definitions and time periods, account for population size and uncertainty, and avoid treating a crime statistic as a direct measure of underlying behavior or police effectiveness. The FBI cautions against ranking agencies using crime data alone.