Author Austin Hepburn, Research and Policy Intern, Data Foundation
Crime data – which includes data on types of crime, demographics of victims and perpetrators, corrections, recidivism and reentry and court information – is crucial evidence that is used to inform policy decisions in all jurisdictions. In the United States, national crime data is aggregated at the federal level, by the Department of Justice’s Bureau of Justice Statistics (BJS) and Federal Bureau of Investigations (FBI). Reliable and up-to-date criminal justice statistics are imperative in order for policymakers to make evidence-based decisions. However, as questions around policing and criminal justice become ever more pressing, it is worth exploring the challenges and limitations of crime data so that we may identify opportunities to improve both the data, and the policy decisions informed by the data.
The National Crime Victimization Survey (NCVS) is the nation’s leading source for statistical information on victims of crime. Victim, offender and crime characteristics are reported along with reasons for reporting or not reporting the crime. However, there are serious limitations to the NCVS. The survey is self-reported by the victim and is not recorded when a crime is committed nor when there is victim to a crime. A sample of households in the United States are taken every 3.5 years and households are interviewed 7 times within that span. Only 71% of households sampled responded to the survey in 2019. Since the survey is only a sample, it does not capture variation in victimization patterns at the local or state level. Therefore, victimization patterns at the city level would require additional research. Additionally, data on the effectiveness of policing practices when addressing crime and victimization is lacking from NCVS reports.
An example of the type of effective additional research is the Data Foundation’s Policing in America Project a multi-pronged, open data effort to systematically improve evidence about how the American people view the criminal justice system and police forces. The project focuses on the value of building data capabilities to enable a more robust understanding of the relationship between perceptions of law enforcement agencies and the conditions in select cities, including disparate perceptions by sub-populations.
In addition to the NCVS, which relies on traditional surveys of victims, a good deal of crime data is reported to the Department of Justice by local law enforcement. The FBI’s Uniform Crime Report (UCR) has been used to provide crime statistics since 1930. The BJS, the primary statistical agency of the Department of Justice, uses UCR data in their publications and datasets. BJS has long been trusted to publish up-to-date and accurate information, utilized by academia and professionals for criminal justice reports and open access data. The data from BJS and the UCR includes local, state, and national level data on corrections, courts, crime, the Federal justice system, forensic science, law enforcement, recidivism and reentry, tribal crime, and victims of crime. The data is reported by local law enforcement agencies to form a national database of criminal justice statistics. In practice, this has led to incomplete and non-standard reporting to the FBI. Local jurisdictions may have different definitions of crime that can make uniform crime reporting difficult. There may be lags in reporting for local agencies, as well as incomplete data.
One challenge comes from inconsistent reporting from local law enforcement agencies (LEAs) which can make arrests difficult to calculate. Reporting data is voluntary, so LEAs may not always report the same data every year. But UCR only uses data from these voluntary reports. The procedure to calculate the aggregated national county and state arrest rates does not take into account the population covered by the UCR. Due to the variable population in UCR coverage each year, this would have a significant effect on the arrest rates. This proposes serious problems in analyzing national time series (over time) trends. Perhaps the main limitation on UCR data, however, is the difference between actual and reported crime.
In addition to inconsistent reporting, the data that is reported is not standardized. Some states may have differing definitions of crime, as well as wholly different crimes on the books. One example would be some states, such as Minnesota, have a 3rd degree murder charge, whereas other states would classify that as manslaughter. Similar challenges exist with hate crime statutes, which may be vastly different, include different demographic information, or may not be a part of the criminal penal code.
Timeliness of data is also incredibly important for informing policy, but there are significant lags in crime data. As of June 2021, aggregated arrest data was last reported in 2016, half a decade to date. Despite the availability of raw arrest data on the FBI’s Crime Data Explorer up to 2019, the Bureau of Justice Statistics has not reported the arrest figures. This means that any policy decisions based on crime data are based on data that is missing the most timely insights.
And finally, data needs to be usable by the public, academic researchers, and policymakers. This means that it needs to be published in an accessible format. Crime data has some significant challenges in this respect. But there are tools to help. The BJS Arrest Data Analysis Tool allows researchers to find national estimates and/or agency-level counts of crime. This data is sourced from the FBI’s Uniform Crime Report. The tool is significant in that viewers are easily able to generate arrest figures at the national and local level without needing the data science background required for raw data processing. While some of these challenges are unique to the crime data, many of these challenges exist in the data infrastructure throughout the country. Many initiatives are being undertaken in order to help address these problems, and optimize data for evidence-based policymaking.
The Foundations for Evidence-Based Policymaking Act of 2018 became law on January 14th, 2019. The bill requires available agency data that is accessible and reports that utilize statistical evidence to support policymaking. Annually, agencies must craft a learning agenda to address policy concerns to the Office of Management and Budget (OMB). This is an opportunity for the Department of Justice to identify and address what needs to be improved and work with stakeholders to ensure that the necessary improvements can be made. This includes an Open Data Plan that must detail how each respective agency plans to make their data open to the public.
Investment in crime and policing data has been modest, preventing meaningful updates to data collection and modernization. Additional funding for criminal justice data collection and reporting is recommended. Increasing local law enforcement training on reporting and correctly classifying all crimes to the FBI can help increase accuracy and reliability. In addition, increased efforts for interagency collaboration between local law enforcement agencies and the FBI can provide more accurate aggregate data. With these common-sense improvements, crime data can be more effective in helping craft evidence based policymaking.