Monday, January 16, 2017

Karlo v. Pittsburgh Glass Works: The Infinitude of Disparate Impact Claims

In Karlo v. Pittsburgh Glass Works, LLC, No. 15-3435 (3d Cir. Jan. 10, 2017), the court held that the Age Discrimination in Employment Act (ADEA) covers claims of disparate impact (unintentional discrimination) involving a subset of the protected group of workers under the ADEA. This means that although the ADEA prohibits age discrimination against all workers 40 and older, an ADEA disparate impact claim can be based on the effects of an employment practice on, for example, workers 50 and older and need not be based on the effects on all workers 40 and older. This is significant because the same employment practice can have an adverse effect on older workers in the protected group but not on younger workers in the protected group.

In an ADEA disparate impact claim, the plaintiff is required to show that a practice adopted by the defendant employer resulted in an adverse impact based on age. The Karlo plaintiffs challenged a reduction in force that resulted in the termination of about 100 employees in more than 40 locations. Unit directors had broad discretion as to whom to select for termination, but they did not receive training on how to implement the RIF.

Critical to the Karlo court's analysis is the fact that the ADEA only prohibits discrimination based on old age, meaning that someone must have been treated worse than a younger individual. Although the ADEA protects all workers 40 and over, it does not, for example, prohibit treating a 60-year-old better than a 45-year-old. Because the ADEA only prohibits older-than discrimination, age is a continuous variable rather than a categorical variable like race or sex. Thus, in an ADEA disparate impact claim, the issue is whether an employment practice generally treats individuals worse as they get older whereas in a disparate impact claim under Title VII of the Civil Rights Act of 1964 the issue is whether an employment practice generally treats individuals belonging to one protected group, such as African Americans, worse than workers belonging to another protected group, such as whites.

Consistent with age being a continuous variable, the plaintiffs' expert in Karlo compared the termination rates of workers older than 45, 50, and 55 to that of younger workers and found that increasing age resulted in an increasing chance of termination. By contrast, if the treatment of workers 40 and older is compared to that of workers under 40, the comparison merely establishes whether the 40-and-over group, as a whole, was treated worse than the under-40 group, as a whole. It does not answer the essential question of whether workers are treated worse as they get older. As the court illustrated:

For example, if a substantial disparate impact is experienced only by individuals sixty-five and older, the effect can show up in the forty-and-older aggregate statistic, creating the misimpression that forty-year-old plaintiffs were disparately impacted. 
In a 1999 decision, EEOC v. McDonnell Douglas, the Eighth Circuit reached the contrary conclusion. The court reasoned that the ADEA does not prohibit discrimination against subgroups because "the consequence would be to require an employer engaging in an RIF to attempt what might well be impossible: to achieve statistical parity among the virtually infinite number of age subgroups in its work force."

In my view, by treating age as a continuous variable in which the issue is whether workers are treated worse as they get older, the Karlo court effectively undercuts any objections that ADEA subgroup claims would require employers to achieve subgroup parity. In addition, plaintiffs will not be able to "'gerrymander' arbitrary age groups" in order to show that an employment practice has a disparate impact based on a particular age subgroup. 

Title VII, however, is an entirely different story. As noted, unlike the ADEA, Title VII disparate impact variables are categorical, so different groups are compared with one another. For instance, under the ADEA, old age is a continuous variable, so the relevant comparison looks at how workers are treated as they get older, not merely at how workers in one age bracket are treated compared to workers in another age bracket. The equivalent under Title VII would be to see if workers are treated better (or worse) depending, for example, on how male they are, e.g., the more male an applicant, the more likely he is to be hired. But unlike older age, maleness is a binary trait. Either you are, or you are not, male. Thus, men as a whole are compared to women as a whole.

For the most part, this means that Title VII comparisons are more straightforward than those under the ADEA, but it also means that there are many more potential comparisons under Title VII. For disparate impact claims involving sex or race, there are not so many categories, but for national origin, the categories number in the hundreds at the very least. Given the large number of potential comparisons, almost any employment practice will, as a theoretical matter, likely have a disparate impact on a protected group if there are enough affected workers in different groups. 

To bring a disparate impact claim, however, a plaintiff must be able to show that a specific employment practice actually had the effect of disproportionately excluding workers of his protected group.

What this boils down to in most cases is the need for a sufficiently large sample size. Consider the example of coin flips. Suppose someone hands you a coin, and asks you whether you think it is biased in favor of a particular side. If you flip the coin four times and get three heads, that's 75% heads, but the odds of such a result are 25%, so you'd be foolish to conclude based on such evidence that the coin is biased. On the other hand, if you flip the coin 100 times and get 75 heads, then you would have good reason to conclude that the coin is biased in favor of heads since the odds of this happening is less than 1 in 5 million.

Because of the need for such statistics, disparate impact claims are inherently systemic cases that require sufficient numbers of employees in the classes being compared. This imposes a practical limitation on disparate impact claims. For example, if an employer administers an entrance examination, a Native American applicant who fails the exam may have difficulty establishing a disparate impact on his protected group if only a few Native Americans took the exam.

Oddly, while concerns about the infinitude of ADEA subgroups appear to be vastly overblown, similar issues about Title VII have entirely escaped notice. This is all the more striking because, as I think is obvious, the real danger lies not with the ADEA but with Title VII, given its large number of protected groups.

Moreover, there are even some kinds of Title VII disparate impact claims in which the courts and the EEOC appear to have entirely overlooked the need for a plaintiff to establish an actual adverse effect on his protected group. For example, if an employer requires English fluency for a certain position, the EEOC's view is that the employer must show that the requirement is justified by business needs. The EEOC presumes that such a practice has a disparate impact and does not require any statistical evidence. The EEOC applies the same kind of automatic and irrefutable presumption of disparate impact if an employer adopts an English-only policy.

Of course, it is undoubtedly true that an English-fluency or an English-only requirement could have a disparate impact if there were enough employees affected by it. Indeed, it may even be likely that it would have a disparate impact. On the other hand, it is also certainly possible that it would not have a disparate impact. In the absence of evidence one way or the other, the plaintiff, bearing the burden of proof to show impact, must lose. 

This blog reflects the views solely of its author. It is not intended, and should not be regarded, as legal advice on how to analyze any particular set of facts.