As coronavirus began to take hold across the country, The Marshall Project recognized the outsize impact it would have on the people living and working in America’s prisons and jails—where social distancing is impossible and many basic preventative measures are against the rules. An astonishing amount of data on criminal justice already goes uncounted or is deliberately obscured from the public. During the course of the pandemic, the need for strong data reporting behind bars has only grown more urgent and we focused on telling those stories.
We were able to draw on our reporters’ expertise to write groundbreaking stories on COVID-19’s impact on the juvenile justice system, courts, policing and juries. We exposed how North Carolina prisons—despite being on lockdown—were still allowing hundreds of incarcerated people to work for local industries like chicken-processing plants, potentially bringing the virus back with them. After that exposé ran in local papers, the state closed the program. Our story on how Texas prison meals had become even more disgusting than usual prompted local lawmakers to press corrections officials to improve the quality of the food.
Working with the Associated Press on covid tracking project means that our data is sent to newsrooms across the country, with more than 670 citing our work. Sen. Amy Klobuchar used our data in a letter to former-Attorney General William Barr, while Sen. Cory Booker and Rep. Ayanna Pressley relied on our numbers in a letter to governors of five hard-hit states, urging them to release people from prison who are over 50 or have pre-existing health conditions.
Our reporting also reveals how little the Federal Bureau of Prisons has done to protect vulnerable prisoners. Rep. Hakeem Jeffries referred to our work while questioning BOP director during a House hearing, our work was cited in a federal court ruling allowing district courts to reduce sentences for incarcerated people in “extraordinary and compelling” circumstances—essentially, to counteract the Bureau of Prison’s reluctance.
Our August investigation into the role of the U.S. Marshals in spreading COVID-19 while transferring federal prisoners around the country led Senators Elizabeth Warren and Cory Booker, and Congressman Ted Deutch, to write letters to both the Marshals Service and Prisoner Transportation Services, urging them to immediately start COVID-testing before transferring incarcerated people between facilities.
For our weekly tracker, we monitor websites for all of our state and federal prison agencies using Klaxon, an open-source reporting tool built by The Marshall Project. This alerts us to changes in data for many of the states. We have also built our own program that takes screenshots of most of the websites’ published data, such as it is.
We manually log all of the figures from the web and our shoe-leather reporting in a shared Google Spreadsheet, accessed by reporters at both The Marshall Project and The Associated Press. After the data has been manually checked, we run it through another battery of scripts to check for any discrepancies that need to be examined more closely.
Ultimately, our data analysis for all the projects is written in R, run through the RStudio IDE.
What was the hardest part of this project?
To build and maintain the database that informs our covid tracking project has been enormously difficult. Most states provide only a fraction of the information we’re seeking. They might post positive cases and deaths, but not testing information, which makes it hard to put the number of cases into context. They might not report deaths at all. Most are not publicly reporting vaccine numbers. And some will report one set of numbers on their website with a current snapshot, without revealing that certain subsets of prisoners are being excluded. Only one or two of the departments post historic data on their site, so it’s extremely difficult to track changes over time. Very often numbers will be revised without any indication of why, for example, the number of recovered prisoners is 1,000 fewer today than it was a week ago. It requires creating our own historic snapshots and a series of checks to ensure the data’s integrity.
We try to surmount these problems by heavily employing the most powerful tool in a data journalist’s arsenal: the telephone. We contact the departments themselves every week and ask them to fill in the blanks and to answer questions about strange changes in any of the numbers. This requires persistence and a good system for examining the data as it comes in.
What can others learn from this project?
By design, prisons and jails are among the most hidden institutions in American life, and an astonishing amount of data on criminal justice already goes uncounted or is deliberately obscured from the public. Even journalists who manage to gain access are only ever shown a highly edited version of life inside.
When we are given data, it’s often incomplete or comes out years after it’s relevant. We’ll receive thousands of pages of scanned, hand-written reports, forcing us to find time-consuming and laborious ways to pull the information into digital databases that we can use to make sense of it all and to share our knowledge with readers.
These stories, which rely on creating new data that doesn’t exist anywhere, have been incredibly demanding on our staff and our AP partners as well. To fill out the latest data from 51 agencies that are generally reticent to answer questions and have a habit of changing their answers when they do give them, demands a great deal of time and effort. To do it every week for ten months and counting is a huge undertaking. But without this data, we, and journalists and policymakers everywhere, would be operating completely in the dark as we try to assess conditions in prisons systems, and thus believe it’s crucially important that we continue to create the data and to make it available to everyone.