Analyzing the NIBRS Data: the Impact of the Number of Records Used per Segment

Abstract

In an effort to upgrade and improve criminal justice statistics, the Uniform Crime Reporting (UCR) program is currently in the process of transitioning from the Summary Reporting System (SRS) to the National Incident-Based Reporting System (NIBRS). While this transition will increase the capacity for law enforcement agencies and analysts to make informed decisions regarding crime and policing policy, the detail of NIBRS increases analytic complexity. More specifically, NIBRS includes variables in six data segments, five of which can include multiple records per incident. As a result, analysts must decide how many records to use. However, there is currently no guidance for best practices in making this decision. This research addresses this gap by examining the impact of this decision on descriptive analyses and regression estimates. Results indicate some estimates are measured accurately using only one record, using three records reduces inaccuracy, and with some exceptions, using more than three records is methodologically unnecessary. As the NIBRS data become increasingly representative and useful in the coming years, it will be important that they are used both efficiently and effectively. Taken together, this research suggests that for most analyses there is substantial consistency when using at least three records per data segment but that there are some cases for which the number of records is consequential and researchers should consider the methodological and theoretical implications of each strategy when choosing between them.