In September 2020 New York University's Public Safety Lab reported to the National Commission on COVID-19 and Criminal Justice on trends in daily individual-level jail data being collected by its Jail Data Initiative. The report, which analyzed a sample of over 14 million daily individual-level jail records collected between January 1, 2020 and July 20, 2020, is available here.

This repository makes publicly available the data used for an updated report released in December 2020, which drew upon a sample of approximately 19 million daily individual-level daily jail records collected between January 1, 2020 and October 22, 2020. The updated report is available here. The data in this repository are aggregated to the roster/day level.

Data appears in individual CSV files subsetted by feature types. All CSVs contain the following fields:

• date
• state (two-character abbreviation)
• county_identifier (more precisely, state_county_roster_identifier; multiple jail rosters that exist within a single county are reported as e.g., WA-Thurston for the Thurston County Sheriff's Office Correctional Facility and WA-Thurston_Olympia for the Olympia City Jail)
• facility_name (multiple facilities may be reported by a single jail roster; when this occurs, facility names are separated by semicolons)
• jail_id (a unique numeric identifier for each jail roster for internal use)
• est_pop_2019 (estimated county population from the 2019 American Community Survey)
• primary_fips (the county FIPS identifier associated with the primary jurisdiction of the roster)
• bjs_juris_id (the Bureau of Justice Statistics identifier for the corresponding facility or facilities)
• missing_date_ind (an indicator for roster/days for which data were not scraped; no data are reported for rosters with gaps of 7+ days in data collection)

Additional notes:

• missing values are different from zero counts (e.g., if a roster does not report a feature, that feature will report null values; if a roster reports a feature, but there are no records of that feature for a given date, the feature will report zero counts)
• "traffic" in a CSV title refers to the presence of counts grouped by bookings and releases, in addition to total daily counts per roster/day
• "recidivism" in a CSV title refers to the presence of data on rebooking by group, in addition to total daily counts by group per roster/day
• charge class refers to classification as felony, misdemeanor, or other
• charge type refers to specific subsets of charges (e.g., failure to appear, cannabis posession, etc.)

Please contact us at [email protected] with questions or comments.