Under the Administrative Remedy Program, inmates held by the U.S. Bureau of Prisons (BOP) are permitted to seek "formal review of an issue relating to any aspect of [their] own confinement." Since the year 2000, prisoners have filed nearly two million such cases, made available to the public for the first time in July after two years of FOIA requests undertaken by the Data Liberation Project.
The new data brings to light thousands of concerns raised by inmates that were never addressed. Of the 41,617 unique cases to reach a final resolution in the year 2019, 41% were initially rejected and sent back to inmates without resolution. The most common reason for rejection (excluding "other") was improper level — an attempt by the inmate to file a complaint directly with the BOP Regional Director without first filing with their own warden.
2,292 complaints were rejected after reviewers determined that they did not qualify as 'sensitive.' The BOP allows inmates to mark complaints as sensitive and submit them directly to the Regional Director if they believe that "the inmate’s safety or well-being would be placed in danger if the Request became known at the institution."
An additional 27% of complaints were reviewed as valid filings but denied without relief.
Only 1.8% of complaints were granted on their first attempt.
For every five cases rejected without a decision, only two were refiled.
Those that did choose to appeal — sometimes several times — could expect to wait an average of two to three months for a final decision. By the end of the appeal process, only 3% of cases were ultimately granted remediation.
In early May 2019, an inmate at the Schuykill Correctional Facility in Pennsylvania filed a complaint against prison authorities for alleged "forced medical treatment." The complaint was rejected before the end of the day for being untimely: according to BOP policy, institutional complaints must be filed within 20 days of an incident.
Ten days later, the inmate refiled the complaint. This time it was rejected with custom remarks not released to the Data Liberation Project. The complaint was rejected 10 more times over the next four months, for reasons ranging from "YOU SUBMITTED YOUR REQUEST OR APPEAL TO THE WRONG LEVEL OR WRONG OFFICE"* — to simply "OTHER." The inmate appealed their case to the Schuykill facility seven times, the regional office three times, and the Bureau of Prisons itself two times.
Twice, in July and August respectively, the inmate filed complaints under the same case number regarding "LEGAL MATTERS / ADMINISTRATIVE REMEDY PROCEDURES", presumably objecting to his inability to receive consideration for the original incident. In September, after a 12th rejection, the inmate finally gave up. Despite a dozen attempts, his complaint never received a formal decision, having been rejected for consideration by the BOP every time it was filed.
The Schuykill case was not unique. Since the year 2000, inmates have filed more than 130 thousand complaints alleging forced medical treatment, the third most common complaint type after housing appeals and staff misconduct.
For every hundred such medical complaints, only about seven were ultimately granted. Yet this is a relatively high rate compared to most other types of cases. Complaints relating to "SEARCHES AND USE OF RESTRAINTS," for example, are granted only about 2% of the time.
Cases with the highest odds of being granted are those related to dental care: with nine positive resolutions per hundred cases. The least succesful, with less than one grant per hundred cases, are marked simply "FSA." While the Bureau of Prisons provided no official translation of this code, it likely relates to the 2018 First Step Act, a bipartisan bill that implemented a number of sentencing reforms intended to lower the number of inmates confined in federal prisons. Secondary subjects tied to FSA cases included "FSA ELIGIBILITY" and "FSA - PROGRAMMING / INCENTIVES."
FSA Related Cases are Least Likely to Be Granted
According to BOP documentation obtained by the Data Liberation Project, at any given time an inmate complaint is assigned one of five statuses: Accepted, Rejected, Closed Granted, Closed Other, or Closed Denied. Accepted cases are exceedingly rare, and appear to mark only filings that were actively under consideration at the moment the BOP exported data from its internal system. Granted cases indicate that the inmate "will be granted the relief," while denied ones will not.
Of the 1.78 million case tickets filed since 2000, nearly seven hundred thousand were rejected — each with one or more of thirty seven "Status Reason Codes" designated by the BOP.
A hand categorization of these codes and their associated narrative descriptions shows that nearly 40% of rejected cases reference paperwork mistakes made by inmates. Such mistakes include "YOU DID NOT SUBMIT YOUR REQUEST OR APPEAL ON THE PROPER FORM (BP-9, BP-10, BP-11)" and "YOU DID NOT SUBMIT A COMPLETE SET (4 CARBONIZED COPIES) OF THE REQUEST OR APPEAL FORM." Nearly 28,000 complaints were rejected from consideration with the reason: "YOU DID NOT SIGN YOUR REQUEST OR APPEAL."
Of the hundreds of thousands of complaints rejected for paperwork mistakes, nearly 30% were never refiled. For those that did choose to refile, only five in every hundred were ultimately granted.
Methodology
*Many sections of the original BOP records are written in all caps. They are preserved here for authenticity.
This story for the Lede Program at Columbia University is based on BOP inmate complaint data made available to the public thanks to the tireless work of Jeremy Singer-Vine and the Data Liberation Project. (Disclosure: I am an unpaid volunteer for the DLP.)
Filings: Each row in the BOP complaints dataset represents a single time an inmate filed a complaint with the BOP, with its subject, start and end dates, status (accepted, rejected, etc), and reasons for that status.
Case Numbers: Case numbers track a single incident that occurred in a BOP facility. They may appear across one or more rows if a prisoner refiles multiple complaints about the same incident at different times.
Case Subject: Each complaint has both a primary and secondary subject. A single case number may appear with differing subjects in different filings, but this is unusual.
This story began with volunteer work I undertook for the DLP tidying the original raw dataset for easier use by researchers and journalists, as well as geocoding the complaints to the real-world locations of their subject facilities. That work is ongoing. However, the analysis presented here is my own and should not be taken to represent the findings of the DLP or its other volunteers. This repository contains the complete supporting code for my analysis, with each calculation marked with its associated conclusions found in the above article.