In addition to adopting a new “Ban the Box” statute, California also recently issued innovative regulations applying general employment discrimination law principles to criminal record screening. The previous post provided some background and discussed how the regulations introduce a “bright-line” approach to when criminal screens are unjustified. This post turns to another innovation, one of fundamental important to all disparate impact litigation, including that involving criminal records: how to establish a prima facie case of disparate impact. In particular, what is the role of general population statistics demonstrating pervasive, deep racial disparities in our criminal justice system?
The evidentiary requirements for a prima facie case are a critically important but somewhat technical and quite neglected topic. Once a disparity is established, the remaining question of justification converges with that asked directly by more narrowly targeted laws like “Ban the Box,” or the Family and Medical Leave Act for that matter.
Some policies subject to disparate impact challenge, like a standardized test designed for use by a specific employer, intrinsically require an analysis tailored to the defendant employer’s application of that policy. That analysis—comparing the racial (or gender, etc.) composition of those who pass versus those who fail the test—requires substantial discovery and invites a “battle of experts.” This significantly raises the costs of litigation and thus deters enforcement. It also makes it particularly difficult to challenge policies at small- or medium-sized employers where smaller applicant pools make statistical analysis less reliable. That becomes all the more true if one starts slicing and dicing the employment policy into sub-components, as some district courts have done in criminal records cases.
Criminal records screening, however, is less like a civil service exam and more like other classic disparate impact settings, like employer screening based on high school graduation or on height or weight. In those contexts, the employer makes a decision based on a characteristic of the worker that exists prior to and independent of her application to a specific employer, unlike a test administered during the application process. Indeed, that is precisely what make these criteria attractive—they are readily ascertainable without significant expense to the employer.
Once the employer’s criterion exists apart from its own application process, however, two important features follow. First, the likely disparate impact of the policy can be estimated by using general population statistics without reference to the employer’s actual practice. Thus, the Supreme Court in Griggs used data on racial disparities in high school graduation in North Carolina to determine that Duke Power’s graduation requirement had a disparate impact; it did so without analyzing the racial composition of those who actually applied but were rejected under that policy. Similarly, the Supreme Court in Dothard used national data on sex differences in height and weight to determine that the Alabama Board of Corrections’ height/weight requirements had a disparate impact. The Court specifically rejected the argument that employer-specific (or even labor market-specific) data was needed. The Court sensibly noted that “reliance on general population demographic data was not misplaced where there was no reason to suppose that physical height and weight characteristics of Alabama men and women differ markedly from those of the national population.”
The second feature of criteria that exist independent of the employer process is that workers can know where they stand before they apply. If an employer announces that it will not hire people without high school degrees, or with a felony record, then people without degrees or with records will tend not to go through the “futile gesture” of applying. The result is that the actual applicant pool fails to reflect who would have applied in the absence of a potentially discriminatory policy, and it is the latter that provides the relevant reference point for assessing disparate impact. Therefore, not only is employer-specific analysis less necessary in these cases, but it is also more error-prone, making it even more expensive and technically difficult to perform well.
Criminal record screens are in these respects analogous to high school graduation and height/weight rules, as was argued by my own and others’ public comments like during the FEHC’s rulemaking process. The racial disparities are massive, and they are robust. Although the precise level varies substantially, this variation occurs within a range that is far above conventional thresholds for disparate impact. This disparity persists across states, across counties within states, across forms of criminal justice contact—arrest vs. conviction vs. incarceraton—, across types of offense charged, and across subpopulations defined by additional characteristics like age, sex, and educational attainment.
Therefore, disparate impact plaintiffs should be able to rely on such general population statistics to establish a prima facie case of disparate impact without introducing employer-specific proof. To rebut such evidence, employers should have to provide evidence that something about the specific form of their policy (such as the offenses or time period covered) or the composition of its otherwise qualified applicant pool (such as its location, education level, etc.) only significantly changes the level of the disparity. Moreover, there must be reason to believe that change is large enough actually to affect whether the more employer-specific disparity is actionable. For instance, in Alaska the black:white incarceration rate ratio is only about a third that of Iowa. Sounds impressive. But the former is 4:1 while the latter is 11:1. So Alaska’s disparity itself remains vastly larger than the roughly 20% threshold readily accepted in disparate impact cases.
Happily, California’s Fair Employment and Housing Commission (FEHC) was persuaded by these arguments and provided by regulation that “State- or national-level statistics showing substantial disparities in the conviction records of one or more categories enumerated in the Act are presumptively sufficient to establish an adverse impact. This presumption may be rebutted by a showing that there is a reason to expect a markedly different result after accounting for any particularized circumstances such as the geographic area encompassed by the applicant or employee pool, the particular types of convictions being considered, or the particular job at issue.”
This principle tends to make employment discrimination law converge with Ban the Box in practice. In California, the new, more specific Ban the Box statute now gets to the essentially same place in a classic criminal records application case as does disparate impact analysis under the state’s general Fair Employment and Housing Act.
Despite this overlap, forcefully articulating the general principle under employment discrimination law remains important. It governs criminal records cases outside the application context, and it should be influential in other domains that raise similar problems, including employer screening based on credit ratings. Hopefully, it will also serve as a beacon both to other states and to federal courts charged with interpreting Title VII—and following the Supreme Court precedent established by Griggs and Dothard.
The FEHC’s actions also provide a case study in the virtues, long urged by scholars, of having an empowered administrative agency that can address via rulemaking emergent, recurrent forms of discrimination. As with criminal records screening, such issues often require thoughtful application of general principles and benefit from empirical evidence. They do not necessarily require a new statute for every problem nor burdensome relitigation in case after case.