Features : : A Matter of Time: The Causes and Consequences of Rising Time Served in America’s Prisons
Executive Summary

To download the executive summary for this feature, click here.

Printable PDF

To download a printable version of this feature (minus audio files and interactive figure data), click here.


To view and download the data, click here.

We took a new approach to measuring how much time people spend in US prisons. How did we get these numbers?

The most traditional measure of time served is the average length of prison terms for all people released in a given year. A serious problem with this measure is that it only includes people released in that year and thus overrepresents people serving shorter sentences. Those serving longer prison terms, who have yet to be released, are not counted. This is a significant obstacle when studying the impact of long terms on prison population dynamics.

We developed a different approach that more effectively accounts for people serving long prison terms who have yet to be released. We look at a snapshot of everyone who was still in prison at the end of a year and calculate how many years each person had served. This allows us to capture trends in time served among people who have not yet been released. For example, someone who entered prison in 1995 and will eventually serve 25 years would not be included in a release cohort until 2020. However, our measure would capture, for each year, the amount of time he or she had been incarcerated to that point (e.g., 10 years in 2005, 15 years in 2010, and so on). This approach allows us to measure, in real time, trends in how much time everyone currently in prison has served thus far, regardless of admission or release date. Our method does not, however, measure the total length of time someone will spend in prison. It only reflects how long each person has served up to that date.

Our measure uses term record data from the National Corrections Reporting Program (NCRP), available for 44 states and Washington, DC. To categorize prison terms by offense, we use the variable indicating the offense with the longest sentence, which we then group based on standard Bureau of Justice Statistics offense classifications. NCRP data present serious challenges, particularly around race and ethnicity. Thus, our analysis on this topic is limited to two groups: “black” (all people identified as black based on the NCRP race variable) and “all other” (people identified as being in other racial groups and people for whom the race variable is missing).

The range of years for which each state has reported data to the NCRP varies greatly. In most of our analyses, we focus on 2000–14 and include all available years for each state. Some analyses use the most recent data available. For most states, this means data from 2014. However, we are limited to data from 2013 for Alaska, Illinois, Michigan, New Jersey, Ohio, Oregon, and Washington, DC, and data from 2012 for Maryland and South Dakota.

Our method underestimates time served in certain ways. For example, we do not count time spent in a local jail before being transferred to a state prison. Data quality and reporting on jail terms are not consistent across states, so we chose to focus on time spent in state prison. We eliminate term records with a total sentence length of less than one year to filter out jail terms that may be included in the NCRP, which can occur in states with unified jail/prison systems. And because of data limitations, we cannot accurately link parole revocation data to the original prison term. A person on parole could return to prison and be released multiple times, but our measure considers each stay in prison separately rather than summing up their cumulative time served.

Throughout this feature, we refer to people with the “longest” prison terms. By this, we mean people in the top decile by time served, or the tenth of the prison population that has been in prison the longest. The cutoff point (i.e., 90th percentile) that delineates who is in the top 10 percent varies significantly by state and year. In some analyses, we also focus on people serving 10 or more years as a measure that is standardized across states and years.

Project credits

RESEARCH: Leigh Courtney, Elizabeth Pelletier, Sarah Eppler-Epstein, Ryan King, and Leah Sakala

AUDIO EDITING: Lydia Thompson and Michelai Graham

DESIGN: Christina Baird and John Wehmann

DEVELOPMENT: Ben Chartoff, Vivian Hou, and Jerry Ta

EDITING: Serena Lei and Daniel Matos

INTERVIEWS: Leigh Courtney, Serena Lei, and Matthew Johnson

PHOTOGRAPHY: Matthew Johnson and Logan Cyrus

WRITING: Leigh Courtney, Sarah Eppler-Epstein, Elizabeth Pelletier, and Ryan King (Intro, Trends, Demographics, Policies, Reform); Serena Lei (Narratives, Reform)

We would like to thank the following current and former Urban Institute scholars for their contributions:
Nancy La Vigne
Samuel Taxy
Cybele Kotonias
Brian Elderbroom
Julia Durnan

Special thanks to those we interviewed:
Stanley Bailey
Ramona Brant
Mujahid Farid
Elvin Garcia
Elizabeth Gaynes
Howard Harris
Samantha Harvell
Monica Jahner
Virginia Lasoski-Nepa
Barbara Levine
Stanley Mitchell
JoAnne Page
Nelson Rivera
Danielle Sered
Jonathan Simon
Dionne Wilson

We are also grateful to the following people for their support:
Jalon Arthur, Cure Violence
Daryl Atkinson, Southern Coalition for Social Justice
Norris Henderson, Voice of the Experienced
William Johnston, Open Society Foundations
Glenn E. Martin, JustLeadershipUSA
Bruce Reilly, Voice of the Experienced
Michael Romano, Stanford Justice Advocacy Project
Danielle Rosario, The Fortune Society
Laura Sager, Citizens Alliance on Prisons and Public Spending
Elizabeth Smith, University of Maryland
Nkechi Taifa, Open Society Foundations

This project was funded by the Open Society Foundations. The views expressed are those of the authors and should not be attributed to the Urban Institute, its trustees, or its funders. Funders do not determine research findings or the insights and recommendations of Urban experts.