The Trump administration for years has used asylum data and grant rates to argue for stricter enforcement at the U.S. border and less processing of asylum requests nationwide. But records about asylum claims are notoriously complicated, and experts have questioned the data’s reliability, accuracy and effectiveness in tracking a complex legal system.
According to a San Diego Union-Tribune review of records, immigration court data — which is tracked and released by the Executive Office for Immigration Review (EOIR) within the Department of Justice and used by the government to calculate statistics for policy decisions — is hamstrung by errors and inconsistencies and is missing information.
“There are so many inaccuracies at every level,” said former immigration Judge Jeff Chase. “I don’t see how anyone, even someone within EOIR, could analyze the data and get an accurate result or accurately use that analysis to shape public policy.”
Based on the current records, which are released monthly, some fields, including gender and whether an asylum seeker already has a green card, are rarely filled out. The raw data itself has misaligned rows and columns, and uninterpretable entries in incorrect fields.
It’s also not always clear what asylum seekers’ situations were when their cases began. Data fields that indicate if applicants were already in the United States when they requested asylum, or if the asylum applications are part of a request for permission to be in the country, are incomplete.
The data often does not distinguish whether an asylum seeker has an attorney or an accredited representative.
Separate “lookup tables,” which are designed to help explain and describe codes entered by immigration court staffers, are not published for some entries seen in the master table, leaving their meaning open for interpretation. EOIR also declined to answer questions from the Union-Tribune on background about how to interpret certain entries.
Because of privacy concerns, the published immigration court records are entirely anonymous. But one table, which contains bond information for applicants, lists someone’s full name where a proceeding identification number should be.
EOIR did not respond to a request for comment from the Union-Tribune about the quality of its data.
The agency’s director, James McHenry, has defended the agency’s data in a letter to a research group that criticized EOIR’s recent releases.
“EOIR has committed to an unprecedented data transparency initiative over the past two years, including the monthly posting of data from its Case Access System for EOIR (CASE) database pursuant to the Freedom of Information Act,” McHenry wrote.
In addition to technical issues, immigration court data are subject to human error. Chase said judges and their support staff manually input case information and have little-to-no training on what codes should be used and when.
“I remember having a handbook that tells you what the codes mean, but nobody sits you down and says when it’s appropriate to use one code or another,” Chase said. “There’s absolutely no guidance in that regard. It was entirely left up to us.”
“Identical cases could be categorized in entirely different ways,” he said.
Among other things, asylum data is used to calculate the rate in which applications are granted, a method that has changed in recent years.
EOIR previously calculated this grant rate by dividing the number of granted applications by all cases decided on merits — cases that received a definitive yes or no.
The agency changed this calculation in 2018, when it published statistics of immigration court data for fiscal 2017.
Email records between agency officials obtained through a Union-Tribune public records request show this was one of the last data calculations officials tackled as they prepared the new report. Most of the discussions about how to calculate grant rates were redacted by the agency.
In the report’s final version, grant rates were calculated by dividing granted applications by all cases with asylum applications that ended in a year’s time, regardless of whether a decision was made on merits.
Since cases can close without reaching a conclusion, often allowing an applicant to remain in the United States, the grant rate shrank.
Records show the Trump administration often publishes this newly calculated grant rate without the context of other possible outcomes, using the minimized grant rate to support the administration’s view that most asylum seekers’ cases are not legitimate.
In July 2019, the Department of Justice and Department of Homeland Security jointly announced a plan to place further restrictions and limitations on eligibility for those seeking asylum in the United States.
The agencies said in a statement that the number of immigrants entering the southern border, the number of people claiming fear of persecution or torture when apprehended by officials, and the number of cases referred to the agency for judicial review have increased dramatically in recent years.
“Only a small minority of these individuals, however, are ultimately granted asylum,” the statement said, arguing that the new restrictions would alleviate the strain on the country’s immigration system by identifying those who plan to abuse the system.
Recently, researchers found even more alarming inconsistencies.
The Transactional Records Access Clearinghouse, or TRAC, a research center at Syracuse University, regularly requests and analyzes EOIR data and publishes its findings online. The nonprofit group released in October a report documenting “gross irregularities” in the federal data.
According to the report, TRAC found that more than 1 million records of U.S. immigration cases were missing from recently released databases — and some of the absent records appeared to be purposefully deleted.
TRAC officials also said EOIR appeared unwilling to explain what happened or correct the issue.
“It is deeply troubling that rather than working cooperatively with TRAC to clear up the reasons for these unexplained disappearances, the agency has decided to dig in its heels and insist the public is not entitled to have answers to why records are missing from the data EOIR releases to the public,” said a statement released by TRAC at the time.
Judge Ashley Tabaddor, president of the National Association of Immigration Judges, issued a statement saying the group’s members were “deeply disturbed” by the systematic flaws uncovered in the report.
“The Supreme Court has made important rulings based on EOIR data that has turned out to be inaccurate and miscalculated,” Tabaddor said. “This is not a dispute about numbers on a spreadsheet. It’s about getting the facts we need to make decisions affecting real people in our courtrooms.”
In March, TRAC co-director Susan B. Long said EOIR was working with TRAC to correct mistakes identified in previous releases, adding that the agency has been more cooperative than most federal agencies she’s encountered.
“It’s unusual to even be able to have discussions like this with any agency about their administrative database, period,” Long said. “If we didn’t have (the Freedom of Information Act), we would be totally dependent on what the agency decided to publish, so it’s really quite revolutionary to have access to internal databases that are so useful.”
Long said she’s familiar with the abnormalities, errors and complications that come with immigration court data. Her team has spent years parsing through millions of records and deciphering internal judicial codes to better understand the asylum system and its processes.
She said her team does not expect any agency’s data to be pristine or understandable to the untrained eye.
“When we’re talking about administrative databases, they’re developed for their own internal use. It’s not designed to be communicated to the public,” Long said.
Still, in a June statement, Long and TRAC co-director David Burnham said the April 2020 release of immigration data was too unreliable to be accurate or meaningful to the public. The statement said any future statistics or reports on asylum applications filed in immigration court published by the EOIR may be equally suspect until the deficiencies are explained or rectified.
McHenry responded to the researchers with his firmly worded letter.
“As EOIR has explained to you previously, data that you allege has been ‘deleted’ actually reflects either real-time updates to information in CASE, technological errors in the extraction and posting of the data which have been corrected, or data that was appropriately withheld from disclosure pursuant to FOIA,” McHenry responded in his June letter.
Not satisfied, TRAC replied to McHenry with another letter, faulting him for not addressing the data’s deficiencies.
