Ban-the-Box “Statistical Discrimination” Studies Draw the Wrong Conclusions

Behind every study and headline that casts doubt on ban-the-box policies is a common—but often misleading—economic concept: “statistical discrimination.”

Here’s the basic idea: when an employer has limited information about a job applicant, the “rational” employer will rely on easily identifiable traits—like race, gender, and age—to make generalizations about the applicant and predict how that person would perform as an employee.

Let’s be clear about what that means in practice, in the context of ban-the-box. Unless people of color—particularly young black men—can indicate on a job application that they have never been involved in the criminal justice system, the default assumption will frequently be otherwise. In other words, when ban-the-box policies are in place, “employers use race to proxy for convictions,”[i] which is racial profiling. This, in turn, is sometimes offered as a rationale for repealing these laws.

But racial discrimination in employment is neither new nor caused by ban-the-box. In study[ii] after study,[iii] white applicants receive more interviews and job offers than black applicants, even when the candidates have no criminal history and are identical in every other respect. One study even found that white applicants with a record received more callbacks from employers than black applicants without a record.[iv] These findings remind us that the nation’s legacy of racial discrimination remains entrenched in employer hiring decisions, as it is elsewhere in our society.

What’s also true is that ban-the-box policies are accomplishing what they set out to do: boost job prospects for people with a criminal history. After Washington, D.C. adopted ban-the-box, employment of people with records jumped by 33 percent.[v] And there are similar success stories to tell in cities across the country, from Durham, North Carolina to Atlanta, Georgia.[vi]  Even in the same studies that have been skeptical of ban-the-box, one concluded that the policy increased employer callback rates for people with criminal records,[vii] and the other found that it increased employment by about 4 percent for older black men without a college degree, and black women with a college degree.[viii]

Still, criticisms of ban-the-box invariably rest on statistical discrimination. But as a recent article in the American Society of Criminology pointed out, “there is something deeply fatalistic about wavering on ban-the-box policies because of statistical discrimination.”[ix] There are at least three reasons why this is so.

In This Context, Statistical Discrimination Means (Illegal) Racial Discrimination

First, across the studies, there’s one point of consistency: employers—when unable to ask about an applicant’s criminal history on an application—often use skin color as a shorthand for criminality, which does the most damage to the employment prospects of young black men.

It’s important to call this what it is: racial discrimination. Racial profiling is also plainly illegal under Title VII of the Civil Rights of 1964.[x] And if the most forceful argument against ban-the-box is that it exposes racial discrimination against young men of color, then our focus should be on expanding and enforcing anti-discrimination laws rather than unraveling policies that effectively serve people with criminal records. We can and must tackle both problems at the same time.

There’s also no need for an employer to speculate, on the basis of race or anything else, about whether a job applicant has a record. This is because ban-the-box policies merely delay when that kind of information can be considered in the application process, instead of shielding it entirely.

Far From Making Rational Decisions, Employers Are Getting It Wrong About Young Black Men

Second, in some settings, statistical discrimination is held up as an integral part of rational decision-making. For example, before a bank approves a mortgage, it may be interested in understanding the overall statistical relationship between household income and default rates. So long as the information is accurate, the rational banker may have a slight edge toward a better decision.

But what happens when the information isn’t accurate? The data used in one study indicated that 15.7 percent of white men in the sample had criminal convictions by age 22, compared to 18.5 percent for black men. That small gap approached zero once differences in education levels were factored in.[xi] Nevertheless, the authors found that employers “greatly exaggerated” the difference and assumed that more than half of young black applicants had a felony conviction, even though this was not based on any “plausible, real-world measure.”[xii] Again, this points to racial discrimination as the deeper problem, not ban-the-box policies.

There’s another limitation to keep in mind, too: even where a general statistical relationship is found in the aggregate, it may provide little guidance in making individual decisions. So, even if 18.5 percent of black men had convictions by age 22, that by itself is of limited use to an employer making an individual decision about a specific black applicant. To base a decision solely on these general trends—as opposed to information about the candidate that’s more nuanced and tailored to the job—is tantamount to racial profiling and a recipe for poor decision-making.

Our Values Matter

Finally, there are domains in which our collective values must carry the day, regardless of how “rational” or predictive statistical discrimination may be. In these areas, the larger questions of who we are and what we believe in as a society come to the foreground and tip the balance.

For example, we know that men and women, over the course of their lifetimes, use health care in different ways. But—because we hold fairness as a core value—the law forbids health insurance companies from setting premiums based on gender. Likewise, the rational business might be apprehensive about hiring too many women, as they could become pregnant. But—because we value gender equity—we have made this kind of discrimination illegal.

That ban-the-box policies have been enacted in 29 states, and embraced by both Democrats and Republicans, is also a marker of our values.[xiii] It signals a firm belief in rehabilitation and giving people second chances.

But what we cannot afford to do is choose one value over another. To the extent that ban-the-box has more fully revealed the cancer of racism in America, that’s a call for more action, not less. After all, ban-the-box has never been advertised as a cure-all and it’s misguided to judge it on that basis.

Instead, ban-the-box is one part of what must be a comprehensive racial justice and economic opportunity agenda. This includes strategic and well-resourced enforcement of anti-discrimination laws, clean slate reforms that seal old and minor criminal records, policies that tear down barriers to employment and support holistic assessments of people with criminal records, and investments in jobs programs.



[i] Amada Agan & Sonja Starr, Ban the Box, Criminal Records, And Racial Discrimination: A Field Experiment, Abstract at 24 (June 7, 2017),

[ii] Devah Pager, The Mark of a Criminal Record, 108 Am. J. of Sociology 937, 957-958 (2003),

[iii] Uggen, C., Vuolo, M., Lageson, S., Ruhland, E., & Whitham, K. H., The Edge Of Stigma: An Experimental Audit Of The Effects Of Low-Level Criminal Records On Employment, Criminology 627, 637 (2014),

[iv] Pager, supra, 958.

[v] Office of the D.C. Auditor, The Impact of “Ban the Box” in the District of Columbia at 16 (June 10, 2016),

[vi] See Terry-Ann Craigie, Ban the Box, Convictions, and Public Sector Employment, 3 (January 27, 2017),

[vii] Agan, supra, 24.

[viii] Jennifer Doleac & Benjamin Hansen, Does Ban the Box Help or Hurt Low-Skilled Workers? Statistical Discrimination and Employment Outcomes When Criminal Records Are Hidden, 22-23 (July 2016),; see also National Employment Law Project, Racial Profiling In Hiring: A Critique Of New “Ban The Box” Studies (August 2016),

[ix] Naomi Sugie, Criminal Record Questions, Statistical Discrimination, and Equity in a “Ban the Box” Era, Criminology & Public Policy 167, 172 (2017),

[x] See generally 42 U.S.C. § 2000e et seq.

[xi] Agan, supra, 25.

Back to Top of Page