New York's recent bail reform law, which went into effect in January 2020 and was amended on July 2, 2020, was expected to reduce the footprint of jail incarceration by limiting the use of money bail. The new law mandated pretrial release for the vast majority of nonviolent charges and required that judges consider a person's ability to pay bail. The combined effects of bail reform legislation and the COVID-19 pandemic drastically changed the landscape of jail incarceration in New York State. The Vera Institute of Justice designed this tool to help people better understand the impacts these enormous changes had on the jail population across the state. The underlying data was collected individually from nearly every county in the state and from the New York State Division of Criminal Justice Services.
Data collection
Vera researchers collected data from multiple sources. For the overall statewide jail population, Vera researchers analyzed monthly jail population data published by the New York Division of Criminal Justice Services (DCJS) between January 2018 and December 2020. For the demographic and charge analysis of people in jail, researchers analyzed 54 county-level jail admission and release datasets obtained from (a) Freedom of Information Law (FOIL) requests (47 counties); (b) data-sharing agreements with sheriff's departments (two counties); and (c) the NYC Open Data site (five counties). Vera excluded eight counties (Chemung, Genesee, Hamilton, Jefferson, Nassau, Putnam, Rockland, and Ulster) because they did not provide complete data or because there was a large discrepancy between the monthly jail population estimated from the data and the monthly jail population reported by DCJS. To calculate incarceration rates, researchers used yearly county-level estimates of the 16- to 64-year-old population from the Center for Disease Control's Bridged-Race Population Estimates. Vera used 2019 population data in lieu of 2020 population estimates. All rates were calculated per 100,000 residents.
Researchers also obtained data for all fingerprintable arrests outside of New York City between January 2018 and June 2020 from the Division of Criminal Justice Services (DCJS). Users can analyze the arrest data using the visualization tool. Researchers also used the arrest data to calculate the percentage of arrests that resulted in a jail admission.
Data standardization
Researchers reconciled disparate county-level datasets with incompatible variables into one file with the following standardized variables.
Admission status. Researchers standardized admission status into six groups: pretrial, sentenced, parole, federal, probation, and unknown. For New York City, researchers identified pretrial admissions as entries with a 'DE' inmate status code on the first date they appeared in daily custody files published on NYC Open Data. Similarly, Vera considered entries with a 'CS' inmate status code to be sentenced admissions. Outside of New York City, Vera identified pretrial and sentenced admissions using admission status and bail variables for each county. People admitted on a parole violation with a new arrest or on a probation violation were counted as people admitted pretrial. Researchers assumed that people awaiting transfer to state prison had been admitted pretrial.
Counties with a Centralized Arraignment Part (CAP) conduct some of their off-hour arraignments in their jails and therefore count some people who are arraigned and released as being booked into the jail. Researchers decided to remove these records when possible as arraignments in counties without a CAP are not counted as jail admissions. For 18 counties with a CAP, researchers could identify and remove CAP arraignments recorded as jail admissions. However, researchers could not identify and remove CAP admissions for three counties with a CAP—Onondaga, Orleans, and Yates.
Researchers identified federal and parole admissions using admission status variables, long-form charge descriptions, and county-specific variables such as 'Parole Violator' flags. Vera researchers removed admission entries for people 'housed in' from other counties to avoid double-counting admissions. Admissions were counted in their originating county, not the county where the person was held. For example, people whose cases were under the jurisdiction of Greene County but who were boarded in Albany county during the construction of Greene County Jail were counted in the Greene County jail admissions.
Release date. The release date for each person in custody in New York City was assumed to be the day after their final appearance in the daily custody files published on NYC Open Data. For all other county datasets, release date data was available. Admission entries with no release date recorded were considered in custody as of December 31, 2020, the last date of the sample period. Admission entries with no release date recorded and an inordinately long length of stay on December 31, 2020, (90th to 100th percentile, varied by county) were removed to minimize the discrepancy between each county's monthly jail population as calculated by Vera and the monthly jail population reported by DCJS.
Race and ethnicity. Vera researchers categorized race and ethnicity into five groups: Black people (including Latinx Black people), non-Black Latinx people, non-Latinx white people, people of another race, and people of unknown race. Forty-three counties provided both race and ethnicity information, and seven provided only race information (including two that recorded Latinx as a race). For those seven counties, all 'white' admissions were assumed to be non-Latinx white admissions.
Length of stay. Monthly median length of stay was determined by calculating the length of stay of each person in custody on each day of a given month and then calculating the median of those values.
Top charge. When a pretrial admission included more than one charge, researchers selected the most serious charge as the top charge and used this in all charge-severity metrics. Charge severity was based on the Division of Criminal Justice Services Code Manual. Only penal law (PL) and vehicle and traffic law charges were considered; all other charges were categorized as unknown. Only PL charges were recorded in NYC Open Data, resulting in a higher percentage of unknown charges than outside New York City.
Bail eligibility. Vera researchers determined bail eligibility from the January 2020 and July 2020 lists of qualifying offenses in PL 510.10. Bail eligibility for a given admission was determined based on all charges associated with that admission. Researchers ignored charge subsections across the data set for the sake of consistency. As a result, all burglary in the second degree (PL 140.25), robbery in the second degree (PL 160.10), and criminal contempt in the second degree (PL 215.50) charges were considered bail eligible regardless of subsection. In addition, none of the counties provided a 'domestic violence' flag—another criterion for cases to be bail eligible. Researchers considered criminal contempt in the first degree (PL 215.51) and aggravated criminal contempt (PL 215.52) charges, which are often associated with domestic violence charges, to be bail eligible regardless of whether they were associated with a domestic violence charge.
New York State Jail Population Brief, January 2018–December 2020
Six County Bail Reform Deep Dive