About the tool
Arrest Trends allows users to easily access and clearly visualize decades of policing data—including arrests, demographics, clearance rates, victimizations, and types of data reported—from disparate sources.
Arrest Trends draws from a wealth of publicly available—but disparately located—policing data. Vera sourced the data primarily from the Federal Bureau of Investigation’s (FBI’s) Uniform Crime Report (UCR), the Bureau of Justice Statistics’s (BJS’s) National Crime Victimization Survey (NCVS), and the U.S. Census Bureau’s (USCB’s) populations with bridged race categories. Vera researchers restructured, aggregated, and merged these datasets for integration into Arrest Trends.
Information about how police use arrest—not only how often they use it and what they use it for, but also who they arrest and how effective arrests are at preventing and solving crimes—is needed to understand and advance policy and practice. While evidence is lacking in many areas, we can begin to answer these questions by making the data that exists more accessible. The government invests considerable resources to track policing indicators, including civilian-police interactions, arrest and clearance rates, crime and victimization trends, and law enforcement operations. Yet, due to data complexities, these key indicators are often inaccessible to those who could most benefit from them: practitioners, policymakers, advocates, researchers, journalists, and philanthropists. In response to this need, Vera developed Arrest Trends—unlocking this important knowledge for stakeholders and the general public alike.
To date, several resources exist to aid in the analysis of policing data. These tools are beneficial in that they are efficient, relatively easy to use, and often interactive. They remain limited, though, in the following ways:
- Users looking for comprehensive information on arrests must consult multiple resources.
- Few tools allow for analysis at the national, regional, state, county, and agency levels.
- Outputs are numbers-heavy and are often presented without written explanations or clear visualizations, making it difficult for users to interpret the findings.
- Tools do not allow users to compare arrest trends across similar agencies or locations.
- Agencies that report partial data are excluded from most tools, affecting the data’s accuracy and completeness.
- Rate calculations (such as clearance rates, arrest rates, and arrest rates for specific demographic groups) based on resident populations are largely absent.
- Although these tools are interactive, they often cannot answer specific or complex research questions.
Users need a centralized and highly interactive tool to effectively explore and understand comprehensive arrest trends at both the national and local levels. Vera designed Arrest Trends to meet this need.
Arrest Trends was developed by Vera’s Policing Program. The team responsible for its relaunch in May 2021 include Frankie Wunschel, Daniela Gilbert, and Caroline Walcott. They would like to thank and acknowledge Megan O’Toole and Abdul Rad for their work on the first iteration and release of Arrest Trends in January 2019. We would like to acknowledge Nick Turner, Jim Parsons, Susan Shah, and Mary Crowley for their invaluable support and guidance. We are also grateful for the assistance of the following Vera colleagues: Jackson Beck, Aaron Stagoff-Belfort, Alex Boldin, Léon Digard, Elle Teshima, Chris Henrichson, Chris Choi, Oliver Hinds, Jacob Kang-Brown, Mawia Khogali, Chris Mai, Michael Mehler, Cindy Reed, Karina Schroeder, Kinsen Siu, Ram Subramanian, Hayne Yoon, Sara Duell, and Sona Rai. In addition, we are grateful for the assistance of Vera’s Policing Fellows and interns: Kristyn Jones, Matthew Stock, Vanessa D’Erasmo, Darren Agboh, Wenshu Yang, Mary Fleck, Daniel Bodah, and Anna Stenkamp. Special thanks to our external advisor, Ross Dakin, for his expertise and valuable advice and to Vanesha Patel for her UX/UI design expertise. We would also like to thank Wipfli, Gramener, Inc., and Civic Hall Labs for their contributions in developing Arrest Trends, and Beta NYC for their participation in user testing. Finally, this project would not have been possible without the support of Vera’s Capital Campaign; the Microsoft Cities Team, especially Elizabeth Grossman, Merisa Heu-Weller, Kevin Miller, and Shiqueen Brown; and Brianna Walden, Jordan Richardson, and everyone else at the Charles Koch Foundation.
Please send any inquiries to ArrestTrends@vera.org.
About the data
Vera Institute of Justice. Arrest Trends. Last modified May 2021. www.arresttrends.vera.org.
Arrest Trends is populated with data from the following data series:
- Uniform Crime Report (UCR): Arrests by Age, Sex, and Race;
- UCR: County-Level Detailed Arrest and Offense Data;
- Arrests Data Analysis Tool National Estimates;
- U.S. Census Populations with Bridged Race Categories;
- American Community Survey 5-Year Estimates (ACS);
- UCR: Offenses Known and Clearances by Arrest;
- National Crime Victimization Survey (NCVS) Victimization Analysis Tool;
- Federal Bureau of Investigations (FBI) Offenses Known to Law Enforcement;
- Law Enforcement Agency Identifiers Crosswalk; and
- Law Enforcement Management and Administration Statistics (LEMAS).
To learn more about the data available within Arrest Trends, see Data Sources and Methodology. To learn more about where you can access these data series, see Resources.
Police agencies report data to the UCR on a voluntary basis. Because Vera did not collect the data and is instead publishing and aggregating existing publicly available data, any errors in primary data sources will be repeated here. If you see an error and would like to have it corrected, please contact ArrestTrends@vera.org.
Arrest Trends does not currently support dataset downloads. Vera anticipates releasing this feature in future updates to the tool.
Years of available data differ by indicator. Data for the arrests, demographics, and data-reported indicators generally covers the period from 1980 to 2018. Data for the clearance-rate indicator includes years from 1964 to 2016. Data for the victimization indicator spans from 1993 to 2018. The Data Sources and Methodology page includes further information on available data and corresponding time ranges within Arrest Trends.
Most of the data within this tool drills down to the police-agency level. County, state, regional, and national data is also available.
For a variety of reasons, not all agencies report their data completely—or even partially—which can cause issues such as undercounting. (To learn more, see Data Sources and Methodology.) To account for this, the FBI provides two sets of UCR data: estimates and reported data. For a more complete picture that adjusts for underreporting, users can opt to view estimated data. To examine trends in the data exactly as reported to the FBI and to avoid any potential challenges with data accuracy that may be caused by nonrandomly estimating missing data, users can opt to view reported data.
This tool features Part I (serious violent and property, excluding arson) offenses and Part II offenses, as classified by the FBI, as well as the FBI’s 29 individual crime types. For a complete list of the crime types featured in this tool, along with their definitions, see Data Sources and Methodology Appendices A, B, and C.
Clearance rates are the percentage of offenses known to law enforcement that are cleared by arrests. According to the FBI, crimes are considered cleared by arrest (i.e., “solved”) when: “at least one person is: a) arrested; b) charged with the commission of the offense; and c) turned over to the court for prosecution (whether following arrest, court summons, or police notice). Although no physical arrest is made, a clearance by arrest may also be reported when the individual is a person under 18 years of age and is cited to appear in juvenile court or before other juvenile authorities.” Federal Bureau of Investigations, “2019, Crime in the United States: Clearances,” https://perma.cc/29VZ-QGPU.
Clearance rates are calculated as the volume of crimes solved by arrests compared to the volume of offenses known to the police. Clearance rates are important to explore as they can indicate police-community collaboration and police effectiveness. However, clearance rates of 0 percent or 100 percent should be interpreted with caution: these may indicate data-entry errors, missing data, or the inclusion of arrests that pertain to offenses committed in previous years.
All Arrest Trends arrest rates are calculated as the number of arrests per 100,000 residents. The tool presents two sets of arrest rates: reported and estimated. Vera calculates reported arrest rates for agencies that reported at least partial data, using the population sizes indicated in the UCR data series. Vera calculates estimated arrest rates using population sizes indicated in the UCR data series for all agencies in a given geographic unit (county, state, region, or nation), regardless of whether they reported arrest data. Estimated arrest rates for specific demographic groups are calculated using the population sizes indicated in the U.S. Census Populations with Bridged Race Categories data series. Arrest rates are helpful in comparing relative trends across geographic units and over time in instances where population sizes have changed. However, arrest rates should be interpreted with caution because not all people are residents of the place where they were arrested. Additionally, virtually no arrests are made of people under age 13, which may skew rates in places with particularly young populations.
Based on UCR data availability, users can explore reported and estimated arrest trends at the national level from 1980 to 2016 (slight variations from this time range exist at the county, state, and regional levels). Data is labeled as “not applicable” when police departments did not collect the relevant information in a given year. Data is labeled as “missing” when police departments collected the relevant information but did not report it to the UCR, as specified by indicators within the datasets. Lastly, the data in this tool may inadvertently include uncommon errors from the UCR datasets (for example, an agency mistakenly noting “1,000 arrests” in their UCR report when 100 occurred, or instances of missing data being misclas-sified as “0 arrests”).
Arrest rates and clearance rates should be interpreted with caution. Arrest rates should be interpreted with caution because not all people are residents of the place where they were arrested. Additionally, virtually no arrests are made of people under age 13, yet people of all ages are included in the calculation of rates. This may skew rates in places with particularly young populations. Clearance rates of 0 percent or 100 percent should be interpreted with caution because these may indicate data-entry errors, missing data, or the inclusion of arrests that pertain to offenses committed in previous years.
When agencies report arrest data to the UCR, they are required to specify arrestees’ age range and gender. Reporting arrestees’ race and ethnicity is optional. Likewise, UCR’s data-collection practices only capture certain cross-sectional demographic groups. Furthermore, UCR only collected and featured ethnicity data from 1980 to 1991, and while agencies have recently had the option to report ethnicity data, only a very small percentage have done so. Therefore, we do not include this information.
By exploring the “Data Reported” indicator, you can learn more about the gaps in policing data and how these vary by time, place, and data type. Because the UCR program instructs agencies to report their data based on which month the arrests occurred, missing and incomplete data can be detected based on which months are not included within the series. In other words, agencies with 12 months of data reported complete data, agencies with one to 11 months of data reported partial data, and agencies with zero months of data reported no data. For this reason, throughout Arrest Trends, you will see indicators of how many months of data were reported—useful markers of data completeness.
The FBI changed its definition of rape in 2013. Beginning in 2013, police agencies could report data according to either the new or old definition. Because of the discrepancies in reporting, data on offenses related to rape is not available in estimated datasets after 2012.
In 2006 and 2016,the National Crime Victimization Survey (NCVS) underwent redesigns in its sampling methodology. You should be cautious when comparing data observed in these years to other years.
Understanding & rectifying the data
UCR reporting is entirely voluntary; agencies can choose to report all their data, some of it, or none of it. Agencies that do not report any of their data to the UCR are not featured in this tool.
The UCR data series sometimes uses acronyms or abbreviations to identify agencies. Before determining that your agency is missing, try using the tool’s “browse by location” feature to filter by geographic location, in order to see if it exists under another name. You can also search by entering the jurisdiction or agency name. Note also that cities and towns are not explicitly featured within Arrest Trends at present because police agencies’ jurisdictions do not always reliably correspond to these geographic units. Instead of searching by city or town, try searching by police agency or county name.
If you determine that the agency or locale you’re looking for is indeed missing, there are still ways to understand more about arrests in the area. To circumvent undercounting issues caused by voluntary reporting practices, the FBI also releases UCR series that contain estimated data at the county, state, region, and national levels. For this reason, you are encouraged to explore your county, state, or region’s estimated data, especially if your individual agency is not present within the tool. Furthermore, departmental or jurisdictional websites often make available similar data or aggregated reports. Through a separate initiative, Vera is documenting where some of these alternative open datasets are available.
If you believe data is missing in error or would like to include data that has not been previously reported, please contact ArrestTrends@vera.org.
Finally, local advocacy for open data is a powerful means of obtaining information and engaging with local system actors to work towards community empowerment, police transparency, and accountability. Contact your local law enforcement agency by cutting and pasting the template below, populating it with your information and emailing/mailing it to your local department to advocate for open arrest data in you area.
Template:
Dear <law enforcement agency>,
<My name is ____, and> I am a concerned constituent in your jurisdiction. It has come to my attention after using the Vera Institute of Justice’s Arrest Trends tool that your agency does not report <any, complete, race, ethnicity> arrest data to the FBI’s Uniform Crime Reporting program. This program, as with many other open data initiatives, is an important measure of transparency and accountability.
< Personal narrative: Personally, this data would allow me to... >.
I urge you to report this data to the public through the FBI’s Uniform Crime Reporting protocol or through other means. I would love to talk more with you about why open data is important for our community to build trust with our police force. You can reach me at ____.
Sincerely,
A concerned citizen/your name
Policing data—especially clearances—is dynamic. For example, an arrest made in 2016 for a crime that occurred in 2014 will be credited as a 2014 rather than a 2016 clearance in the dataset. Therefore, changes in previous years may occur as Vera updates data. Likewise, agency data is constantly changing. UCR data is typically static outside of annual updates, meaning that agency records may differ if something changed after reports were provided to the FBI. If you represent an agency and would like to fix incorrect data, please contact ArrestTrends@vera.org.