Daily Crime Data from 50+ Large American Cities

The lack of public and researcher access to real-time crime data represents a critical gap in the U.S. data infrastructure. When data are limited, policymakers and the public often base decisions on emotion and ideology. In the case of crime policy, this substitution results in solutions to crime that are less efficient, less effective, and less responsive.

Live Crime Tracker meets this data need by providing up-to-date statistics on criminal offenses in the United States. This site acts as an early warning system to detect trends in violence and victimization. Currently, more than 50 of the 200 largest U.S. cities regularly report crime data. We will continue to incorporate data for additional cities as they become available. For more information on our data sources, visit our Methodology page.

Contact Us

For Media Inquiries: Contact Eric Young at young-eric@norc.org (301) 634-9536
All Other Inquiries: Contact John Roman at roman-john@norc.org (202) 695-0518

Data Sources

We obtained Live Crime Tracker data from city police departments or city governments that provided daily incident-level data starting in January 2023. A city’s data are included if individual crime incidents are reported, the type of crime is clearly identified, and a date for the offense is reported. Currently, we do not include data from cities that only report aggregated monthly or yearly counts. We also do not include crime incident data from cities without a publicly accessible source. For more information about each city, please visit its website (available below in City Data Details). Population data used for analysis were obtained from the U.S. Census Bureau.

All data are collected from open data portals. Open data, specifically incident-level crime data, are critical to enhancing public and stakeholder understanding of risks and harms from crime. Unlike crime maps or summary data, open, incident-level data allows the public to evaluate how crime varies over time and by crime type and to assess when in a month or a year risk increases or decreases. This leads to a more nuanced understanding of the causes and consequences of crime and more effective solutions.

To define the cities included in the current Live Crime Tracker daily monitor, our team began with the 200 largest cities in the U.S. We searched each city's public web portals for the availability of crime incident data, including police and other official websites. The initial list of cities was created using the most recent year of U.S. census data available. After compiling a list of cities with publicly available crime data, we categorized cities by how frequently their data were updated, the type of data reported (i.e., raw incident data, aggregated monthly counts, or quarterly reports), and how and which crimes were reported (NIBRS offense codes or other categories). Sampling criteria required that cities report incident-level data and update the data frequently, with no regular severe lags. We found over 50 cities that met our criteria.

Data files are downloaded and processed each day with an automated Python program. The program collects most data from application programming interfaces (APIs) that cities make available through their websites or open data portals. Files for each city are first downloaded and processed individually before being combined into a single dataset. Several processing steps ensure that correctly formatted data are loaded into the final dataset. Perhaps the most consequential data processing pertains to harmonizing how cities label crimes to a consistent definition. Our tool uses NIBRS codes when available and then a set of coding rules when NIBRS codes are unavailable. Once data are coded at the incident level, we then aggregate counts of each crime type for each city each day. Crimes recorded as occurring before January 1, 2023, and after the current date are excluded, as are crime incidents that are missing information about incident type.

The statistics constructed come from a combination of administrative data and are subject to non-sampling error. For example, administrative records may contain measurement errors because of issues such as coverage problems (e.g., the data source may not include the entire geography); conceptual and timing misalignments; reporting errors; definition and classification difficulties; errors in recording or coding the data obtained; and other errors of coverage, processing, and estimation for missing or misreported data.

The data collected often contain missing observations. For example, some cities may report no data for a given day on any type of crime, while others might report incidents only for certain crime types but not for others. Having no reported incidents for a city and crime type on a given day is not necessarily a problem because certain crimes might not occur every day, especially for rarer crime types (e.g., homicides), particularly in smaller cities. In other cases, having no crime reported may indicate missing data. To account for this, we implement different strategies. If a city reports at least one type of crime on a given day, all other crimes with no incidents reported on that day are coded as zero. If a city does not report any incident data for more than five consecutive days, the city’s data is recorded as NA until data are recorded again. Once new incident data are recorded for that city, the NA values for the extended period of consecutive missingness remain. The current release imputed values when less than five days of data were missing by filling based upon the last value. Future releases will implement imputation methods for multivariate time series, including reporting on associated uncertainty surrounding those imputed observations.

City Data Details