A team of researchers from NYU’s Public Safety Lab will use data science techniques to study the impacts of pretrial detention in more than 1,000 U.S. counties—including many rural counties that have remained largely unstudied.
Recent studies have shown that defendants who are held in custody pretrial are more likely to commit additional crimes after release, compared to defendants who are not detained pretrial. However, this research has largely been limited to a handful of large cities.
To address this gap in criminal justice scholarship, a team of researchers from NYU’s Public Safety Lab will use data science techniques to study the impacts of pretrial detention in more than 1,000 U.S. counties—including many rural counties that have remained largely unstudied.
“Our understanding of the effects of holding defendants in jail pretrial has left unexplored the vast landscape of jails in less populated and more rural counties and the increasing numbers of defendants detained there,” explains Anna Harvey, professor of politics at New York University and director of NYU’s Public Safety Lab. “As a consequence, policymakers in more rural states and counties may remain largely uncertain about the causal effects of their pretrial detention practices, including whether these practices are actually increasing crime in their jurisdictions.”
The Public Safety Lab research team, which will be supported by a grant from the Laura and John Arnold Foundation and which includes Gregory DeAngelo, a professor of economics at Claremont Graduate University, will use innovative web crawling and information extraction techniques to:
· Collect, on a daily basis, detention records from publicly available online county jail rosters for more than 1,000 U.S. counties over a period of three years;
· Collect defendant-level criminal case data from all available online state and county court portals over the same period and match defendant-level detention and case records;
· Conduct an analysis of the causal impacts of bail, access to counsel, and pretrial detention decisions on defendant-level case and recidivism outcomes, using jail capacity constraints as a series of natural experiments reducing detention durations.
In addition to generating new insights into the impacts of pretrial practices on defendant outcomes, the research team will produce a database of detailed detention and case records for individual criminal defendants across hundreds of counties in the United States. The resulting database will be made publicly available when the project is completed.
The high rate of pretrial detention in the United States is due both to the widespread use of monetary bail and to the limited financial resources of most defendants: less than 50 percent of defendants are able to post bail even when it is set at $5,000 or less.
Notably, the growth in pretrial detention in the U.S. over the past 50 years has been steepest in the least populated and most rural counties: pretrial incarceration rates in these counties have increased 436 percent since 1970. Overall, the number of individuals detained in local county jails on any given day has risen from 157,000 in 1970 to over 700,000 today, with approximately a half million of these daily detainees being held pretrial.
While earlier studies have shed significant light on the nature of pretrial detention, they have lacked defendant-level information about the amount of monetary bail assigned, defendants’ access to counsel, and the duration of pretrial detention.
“As a result, policymakers may not have a clear understanding of the causal effects of these practices on post-release recidivism, particularly in smaller cities and more rural counties,” observes Harvey.
The Public Safety Lab uses the tools of data science and social science to support communities’ efforts to improve public safety outcomes. Coordinating the work of researchers across multiple disciplines and universities, the lab aims to achieve more productive allocations of public safety resources. For more information, please visit publicsafetylab.org.