NIBRS estimation and extract file creation

Abstract

The National Incident-Based Reporting System (NIBRS) offers the federal government and researchers a much richer set of information about crimes recorded by the police than the summary format previously used by law enforcement agencies. While the summary format simply provided a count of offenses recorded in a given month, NIBRS provides detailed information at the incident level about characteristics related to the incident, offense, victim, offender, and arrestee. However, because not all law enforcement agencies have transitioned to submitting information to the Federal Bureau of Investigation (FBI) using NIBRS and not all NIBRS reporting agencies provide a full 12 months of data, a robust and complex statistical estimation process is needed to produce representative and unbiased estimates of crimes recorded by police.

Beginning with the 2021 data year, RTI International has produced representative estimates of crimes recorded by the police for the use as official statistics by BJS and FBI. Estimates are produced across several levels of geography, including the national, regional, state, metropolitan statistical area, judicial district, and FBI Field Office levels. Estimates cover topics related to the characteristics of incidents, offenses, victims, and arrestees. Additional estimates are produced specifically on the topics of assaults to law enforcement officers, gun violence, and details of drug offenses. Estimates of victims and arrestees are produced among all persons as well as by population subsets by age, sex, and race (e.g., Females, aged 18–24).

In this report, the methodology implemented to produce the annual estimates of crimes recorded by police is described in detail. The report covers the following topic areas: (1) the procedures for ingesting the FBI Criminal Justice Information Services (CJIS) NIBRS database and implement the transformations needed for the estimation process; (2) the structure of the indicators for which estimates are produced; (3) the procedures for accounting for agencies that do not participate in NIBRS through weighting; (4) the procedures for handling missing or unknown values through imputation; (5) the procedures for developing population estimates; (6) the procedures for measuring statistical uncertainty; (7) the process for constructing confidence intervals; (8) the publication rules for suppressing estimates; (9) the structure and format of the output; (10) how to interpret the estimates; (11) how to conduct statistical tests using the estimates; and (12) the procedures for creating the NIBRS Extract Files.