by Anthony W. Accurso

Concerns about the accuracy of criminal background check reports led two researchers to study this issue, and the results undermined commonly held beliefs about the reliability of such reports and their failure to align with American values and public policy goals. 

Researchers Sarah Lageson and Robert Stewart—of the School of Justice, Rutgers University in Newark, New Jersey, and University of Maryland, College Park, respectively—analyzed criminal background information on 101 participants known to have at least one prior criminal charge. Not everyone had a conviction, though most did. All were residents of New Jersey and were referred to the study by legal aid organizations. 

For all participants, researchers obtained a report from “three sources: (1) the official state record; (2) a background check report from a consumer reporting agency; and (3) a report generated from an open-source, public records data aggregator, also known as a ‘people search’ service.” Researchers also “asked respondents to review their criminal records in a qualitative interview.”

Qualitative interviews were used to verify the official state record, since it sometimes lacks information on case dispositions. Official records were most accurate because they require what are known as “ground truth identifiers,” such as a social security number, fingerprints, or other biometric data. Consumer reporting agencies (“CRAs”) and private data aggregators “rely on matching algorithms using combinations of alternative data points” such as “name, alias, birthday, [and] state.”

The results of the study showed how three types of errors manifest: “as mismatched data that create false negatives in third-party records, as missing case dispositions that create incomplete and misleading criminal records, and as incorrectly reported criminal records that lead to false positives in third-party reports.” 

A false negative is any instance where a charge or conviction is present in the state report but missing from a private sector report. Of the 1,323 charges present in the sample, CRAs failed to report 672 (50.8% false negative rate), and the data aggregators failed to report 1,065 (80.5% rate). The private sources located charges for only 77 of the participants, meaning 24 participants appeared to have no criminal record despite having entries on the official state reports. 

A false positive is any instance where a charge or conviction is present in a private source but is absent from the official state report. This often occurs when an algorithm creates a “match” between a record and non-matching person, for example where the name and state match a criminal record, but the person’s birthday is different. The private sector reports listed a total of 1,776 charges. That’s 453 charges more than the official records. Additionally, the CRAs reported 472 charges not present in state reports, while data aggregators reported 715. Fully 73.5% of charges that appeared on private data aggregator reports were not present on state reports. 

As troublesome as these errors are, at least some employers and landlords are willing to look past convictions for individuals who have paid their debt to society, but the absence of a case disposition “can cause significant problems for the subject of a background check as it implies that a case is still open or pending,” which is a level of uncertainty many forgiving employers aren’t willing to risk given the costs of training new employees.

Some individuals in the study were disproportionately impacted by errors. Participant #83’s official state report “listed only two drug felony charges (cocaine possession in 1986 and cocaine sales in 1991) that both resulted in convictions.” Yet, the CRA reported “78 additional charges filed from 1986 to 2009, with convictions that included armed robbery (5 separate charges), burglary (5), receiving stolen property (7), forgery/fraud (2), weapons offenses (6), and arson (1).” 

Participant #97 was “arrested in 1990 for sexual assault but was charged and convicted with misdemeanor simple assault, and this is all that appears in his state report.” Yet, his CRA report shows “11 charges, including seven counts of drug trafficking, two counts of drug possession, one count of receiving stolen property, and one count of felony conspiracy.”

This person was working at the same company for almost 30 years when it instituted a new background check policy, which included current employees, and he was fired. He found a new job but was let go after the “background check included a conviction for an assault on a police officer in another state,” which was actually a false positive “match” for “another person with the same name, birthday, and a match on three of the last four digits of his social security number.”

Participant #61 was arrested in 2010 “in a pharmacy parking lot after attempting to fill a forged prescription and getting into a minor car accident with a responding police officer.” The DA charged her with assault on an officer and attempted murder, along with the prescription forgery charge. The judge dismissed the charges relating to the accident, substituting them for resisting arrest. 

But her CRA report shows “two counts of attempted murder and assault on a police officer,” and the data aggregator report erroneously notes “20 felony convictions for obstruction, forgery, and aggravated assault and two unknown dismissed records.”

She reported being routinely denied housing and employment due to the errors on her report. “I take responsibility,” she said in her interview. “I should not have been there, but for it to say attempted murder is just crazy; it wasn’t like that.”

At this point, it’s important to note that official state reports could contain errors of fact, mainly as a consequence of states shifting from paper records to digital and records not being entered correctly. These would not have been labeled as “errors” in this study because the state report information was used as the baseline for comparing the two private reports. 

Private companies fall into two main categories: CRAs and data aggregators. CRAs are regulated, and they must comply with The Fair Credit Reporting Act (“FCRA”) of 1970 to offer a background check product for use in making housing or employment determinations. 

“Unregulated” companies, commonly known as “data aggregators” or “people search” companies, charge a one-time search or monthly subscription fee, require no consent from the subject of the search, and do not have a common method for objecting to erroneous data in the reports. Further, these companies have a monetary incentive to over-report including false positives because their users are less likely than employers and landlords to face legal action for discrimination in a personal context. 

Ironically, CRAs are often more accountable to justice-impacted people then even states are because these people can dispute inaccurate data on a report—15 U.S.C. § 1681(o)(5)(A))—and companies can be fined for failing to “follow reasonable procedures to assure maximum possible accuracy of the information concerning the individual about whom the report relates.” 15 U.S.C. § 1681e(b). 

Compare this to state workers “tasked with collecting and organizing criminal history data” who “receive broad, qualified immunity for any errors they create.”

Data errors are difficult to challenge even through CRAs, leading to what researchers call “systems avoidance,” where impacted persons would rather suffer the consequences of the erroneous data than deal with the Byzantine system to resolve them. One federal judge damningly described the FCRA challenge process as “imposing such burdensome procedural requirements that no lay person could possibly be expected to comply.”

All of this collective mess is occurring against a backdrop of pervasive background checking by employers and landlords—reportedly, up to “94% of employers conduct background checks” and 93% of those use only private reports. For this reason, “[t]he relationship between criminal records and employment access has been studied extensively during the last two decades, including how serious and less serious records affect employer callbacks, whether credentials and reentry programs mitigate criminal records effects, and how employers view applicants with records.” And criminal records continue to “have an outsized role in people’s lives, conditioning access to opportunities and resources across a wide range of areas.” 

Various movements have sought to address the situation. “Clean Slate” initiatives in some states have allowed impacted persons the opportunity to expunge or seal records after some amount of time, but this has been undermined by private data companies that are unregulated and retain and report this data. 

Likewise, “Ban the Box” and “Fair Chance” were once touted as a solution, but these simply postpone the background check until a later stage of the hiring process. 

One less mentioned movement is “Open Hiring.” The Greyston Bakery in New York pioneered this approach in 1982. Under this approach, the business “accept[s] an individual based on current actions and future potential, not judge them on their past.” Around 80% of the Greyston staff—close to 80 workers in 2020—have come through their open hiring model. Another more recent addition is Dave’s Killer Bread, a company based near Portland, Oregon, which practices what it calls “Second Chance Employment: hiring the best person for the job, regardless of criminal history.”

Just prior to the pandemic, when the unemployment rate was lingering around 3.6%, employers faced the challenge of finding workers to fill positions and began to explore their options. The Body Shop, a multinational beauty brand with around 2,500 stores in 80 countries, made headlines in 2020 by adopting the Open Hiring model after consulting with Greyston Bakery. Requirements for positions were modified, with entry-level jobs limiting requirements to “the ability to lift over 50 lb, stand for 8 hours, and be authorized to work in the United States.” This was on a much bigger scale than Greyston Bakery, since “The Body Shop employs around 1,000 retail workers during peak seasons, with 10,000 employees in total with annual revenue close to $1 billion dollars.”

If Open Hiring seems to work for these companies, why hasn’t it been more widely adopted? In the U.S., this is primarily due to liability for “negligent hiring claims.” According to Forbes.com, “[n]egligent hiring occurs when an employer fails to verify that a prospective employee may present a danger to the organization.” States define the elements necessary to prove a negligent hiring claim, with most being that an aggrieved party must establish: “(1) the employer owed a ‘duty of care’ to others when hiring the worker; (2) the employer breached that duty; (3) the breach was the cause of the injury or harm; (4) the injury or harm was reasonably foreseeable; and (5) damages resulted from the employer’s action.”

Take the case of Trusted Health Resources, Inc. of Boston, Massachusetts. The company failed to perform a background check, which would have revealed that the man applying for a position as a home healthcare worker had six previous felonies, including physically abusing an elderly client. The employee then murdered a disabled man and his grandmother, and the estate won $26.5 million, including $18 million in punitive damages. 

The Equal Employment Opportunity Commission provides guidance that employers should avoid blanket policies that exclude individuals with a criminal record, should screen workers proportional to the role they are filling, and consider the specific nature and gravity of the crime. But with negligent hiring claim “settlements averaging more than $1 million,” it’s understandable that employers face difficult choices.

There has been some movement in the housing space as well. Responding to concerns that many people were homeless in Seattle because of felony background checks, that municipality passed the Fair Chance Housing Ordinance in 2017, barring landlords from asking about a prospective tenant’s criminal history and prohibiting “taking an adverse action based on that history.”

The city of Seattle wrote that “[t]he legislation caps a decade-long effort to address bias against people who have served their time, are seeking to provide for themselves and their families, and yet have faced barriers to accessing safe, stable housing.”

A conservative legal group helped area landlords challenge the law, and in June 2023, the U.S. Court of Appeals for the Ninth Circuit struck down the provision barring landlords from asking about a prospective tenant’s history but left in place the prohibition on taking adverse action based on it. But Seattle is an outlier, and landlords across the country regularly deny access to housing even after a person has paid their debt to society. 

This is not simply about employers and landlords having access to accurate data on prospective employees and tenants, especially since this and other studies have highlighted inaccuracies in background check reports. It’s also about systemic issues, such as how criminal history is often used as a proxy for race, allowing for a loophole in laws designed to protect against discrimination in housing and employment for people of color. 

The Consumer Financial Protection Bureau has even warned that name-only matching algorithms heightened the likelihood of erroneous matches “because there is less surname diversity” in “Hispanic, Black, and Asian populations.”

Inaccurate reports also undermine basic principles of American society and justice, such as the right to be innocent until proven guilty and the right to due process. And let’s not ignore that the companies must pay for these reports—even if landlords often require tenants to reimburse them—and for their money, they are not getting good information. 

The study’s authors note that, however inaccurate and damaging private background reports are, even this tells an incomplete picture of the harms these reports cause. The study was limited to 101 participants in one state, and it entirely omits the number of false positives for people who have never been charged or convicted of any crime. They also note that the pervasiveness of background checks has led to “function creep,” where “administrative criminal record data on individuals are collected for specific internal purposes and uses but have migrated to other purposes and uses that extend and intensify surveillance and invasion of privacy beyond what was originally understood and considered socially, ethically and legally acceptable.”

“The problem with criminal records is that they regularly contain errors, are easily misinterpreted, and are thus unreliable tools for evaluating individuals,” wrote the study’s authors. It is clear that inaccurate background check information is ruining lives every day and costing people and companies billions of dollars.  

Sources: Lageson and Stewart, “The Problem with Criminal Records: Discrepancies Between State Reports and Private Sector Background Checks” 2024; forbes.co; daveskillerbread.com; cnn.com; seattle.gov

Subscribe today

Already a subscriber? Login