Description of original award (Fiscal Year 2015, $39,973)
The National Incident Based Reporting System (NIBRS) is a national data set on crime incidents with contributions from hundreds of police agencies across the country. NIBRS data analysis has begun to contribute substantially to knowledge, particularly in explaining arrests. However, research suggests problems with reliability of arrest data, which are not always updated to capture new case information after submission to NIBRS. The proposed study would assess the reliability of NIBRS arrest data and explore factors supporting or detracting from its accuracy. Findings could help spur improvements in NIBRS that would lead to higher quality crime research.
The research team is a partnership of a victimization researcher from the University of Illinois at Urbana-Champaign; a crime data specialist from Fisher College in Boston; and a recently retired NIBRS expert from the Crime Reporting Unit (CRU) of the Massachusetts State Police. The team
will draw a stratified random sample from NIBRS of 480 incidents of aggravated assault, simple assault, sexual assault and intimidation investigated by local police in Massachusetts from 2011 to 2013. Cases from small and medium police departments will be oversampled and data weighted accordingly. Individualized data sheets will be prepared and sent to each local police 1 agency represented in the sample, listing that agency's cases by incident number. For each case on the sheet, agencies will be asked to provide data on arrests. Agency and NIBRS data will be merged in SPSS files. Analysis will examine several indices of reliability, including the percentage of cases in which local and NIBRS agreed, positive and negative agreement (focusing on reliability of arrests and non-arrests respectively), and Cohen's kappa, a measure of reliability that corrects for artificially inflated agreement due to high base rates for one category. Additional analysis will examine whether results differ by size of police department, type of crime, and type of case data software use. Interviews will be conducted with data specialists from 18 local agencies represented in the sample; these will explore factors that enable or interfere with accurate data delivery, such as NIBRS instructions, definitions and training and case data software. Through a journal article, conference presentations, webinars, research briefs and newsletters, the research team will disseminate findings on reliability of arrest data and ideas for
: improving NIBRS. Files merging NIBRS and local data will be archived for further study.
This project contains a research and/or development component, as defined in applicable law.
- Can Law Enforcement Training Address Racial and Ethnic Disproportionality? An Experimental Evaluation of Effective Youth Interactions
- A Whole Community Approach to Hate Crime Reduction: Research to Improve Prevention, Reporting, and Victim Support
- A Statewide Mixed-methods Evaluation of Pennsylvania’s 8th Edition Sentencing Guidelines and their Impacts on Racial and Ethnic Disparities in Sentencing Outcomes