On May 7, 2014, the First Circuit issued its decision in Jones, et al. v. City of Boston, et al., No. 12-2280 (1st Cir. May 7, 2014), and vacated an order granting summary judgment in favor of the City and, instead, entered partial summary judgment in favor of Plaintiffs.
In doing so, the First Circuit found that Plaintiffs’ failure to demonstrate the “practical significance” of a statistical disparity did not preclude them from establishing a prima facie case of disparate impact. Id.at 30. The First Circuit’s rejection of practical significance sets it apart from others, notably the Tenth Circuit. (Read our post discussing the Tenth Circuit’s view here.)
This is a key issue for employers in workplace class actions.
With a large enough data set, even very small differences can be statistically significant. (For example, if you flip a coin one million times and get tails 50.1% of the time, the deviation from the expected 50% result is statistically significant.) As a result, statisticians generally recognize that not all statistically significant results are practically significant or “practically meaningful or important.” Id. at 20-21.
Nevertheless, after struggling with a method for gauging practical significance as a matter of law, the First Circuit ultimately found that any “theoretical benefits” of inquiring as to practical significance were outweighed by the difficulty of doing so “in any principled and predictable manner.” Id. at 30. In other words, because it could not define an acceptable test, the First Circuit opted to jettison the requirement altogether.
Between 1999 and 2006, the City of Boston subjected its officers and cadets in its Police Department to annual drug tests using samples of their hair. Id. at 4. Pursuant to a collective bargaining agreement, the City analyzed employees’ hair for exposure to five substances — cocaine, marijuana, opiates, PCP, and amphetamines. Id. at 5.
Although the City tested more white officers and cadets, it found more black employees positive for cocaine during the relevant period. For instance, in 2003, 6 of 529 black officers and cadets tested positive, or 1.1%, while 3 of 1260 white officers and cadets tested positive, or 0.2%. Id. at 6-7.
Plaintiffs argued that the racial differential in outcomes did not result from chance. Id. at 12. They asserted that black individuals tend to have higher levels of melanin in their hair and that melanin causes cocaine and associated chemicals to bind to hair at a higher rate. Id. Plaintiffs also claimed that hair treatments more common in the black community made it more likely that “aerosolized powder” from snorted or smoked cocaine would deposit on the hair of non-users. Id. at 12-13.
The City disputed any correlation between positive test results and melanin levels, asserting that many Caucasians and most Asian -Americans have melanin levels as high or higher than those of Plaintiffs. Id. at 13. Pointing to studies, the City also claimed that the relative rates of positive drug tests for black and white individuals remain materially constant across different methods of testing, specifically hair, urine, and blood. Id.
On September 28, 2012, the district court granted summary judgment to the City on all claims. Id. at 14. Plaintiffs appealed.
The First Circuit’s Opinion
The First Circuit reversed. The First Circuit found that plaintiffs identified the challenged employment practice as the “hair test” and presented evidence that the results had a statistically significant correlation with race. Plaintiffs showed that, in at least three of the eight years during the relevant period, as well as during the eight years aggregated, the differential between positive results for black and white employees was statistically significant. Id. at 16.
Rather than challenging Plaintiffs’ math, the City attacked Plaintiffs’ methodology. The City apparently did not question Plaintiffs’ showing of causation or claim that Plaintiffs failed to control for any number of other potentially contributing factors such as age. Id. at 19-20. Instead, the City raised other objections.
First, the City claimed that employees who opted to avoid termination in the wake of a positive test by undergoing drug rehabilitation were “correctly identified” and, therefore, should have been excluded from Plaintiffs’ statistical analysis. Id. at 17. The First Circuit found no evidence that altering the raw numbers in such a way would produce any material difference. It also found the accuracy of the identification process irrelevant to disparate impact – accuracy “is a different matter, perhaps relevant to the business necessity defense.” Id. at 17-18.
Second, the City objected to plaintiffs’ aggregation of data from the first eight years of the program. Id. at 18-19. The First Circuit noted that some form of aggregation was necessary to sustain the claims of plaintiffs who tested positive in years with no statistically significant disparity. The City, however, did not raise the argument in support of its motion for summary judgment and did not sufficiently develop it in the district court. Id.
Third, the City argued – and the district court found – that the statistically significant disparities did not show disparate impact because they were not sufficiently large or “practically significant.” Id. at 20. The First Circuit acknowledged that, with a large enough data set, even small differences can be statistically significant, therefore, not all statistically significant results are “practically meaningful or important.” Id. at 21.
The First Circuit nevertheless refused to apply a practical significance requirement due to the practical difficulties. The concept of practical significance is “impossible to define in even a remotely precise manner.” Id. at 25. Accordingly, with no objective measure, “courts would find it difficult to apply such an elusive, know-it-when-you-see-it standard, let alone instruct a jury on how to do so, and parties may find it impossible to predict results.” Id. at 25.
The First Circuit also rejected the City’s reliance on the EEOC’s “four-fifths rule” as a gauge for measuring practical significance. Id. It noted that it, as well as its sister circuits, have minimized the rule’s importance and criticized it directly. The rule can lead to anomalous results, and its consequences can vary in “seemingly arbitrary ways” depending on the magnitude of the selection rates at issue. Id. at 28. The rejection of the four-fifths rule left “no statute, regulation, or case law proposing any other mathematical measure of practical significance.” Id. at 29.
The First Circuit noted that its confidence in rejecting a practical significance requirement was “bolstered” by two other requirements. First, although the appellate court rejected the City’s aggregation argument, it noted that the need to show statistical significance “will eliminate small impacts as fodder for litigation in many instances because proving that a small impact is statistically significant generally requires large sample sizes.” Id. at 31. Second, a prime example of unpredictability, the First Circuit pointed to job relatedness – an employer “may rebut a prima facie case of disparate impact by showing that its use of the challenged practice is ‘job related . . . and consistent with business necessity.’” Id.
Implications For Employers
Given the inconsistencies within the decision – and its inconsistencies with the approach of other Circuits – the decision seems prime for a motion for reconsideration or motion for en banc review. If the decision stands, it smooths the road for plaintiffs looking to establish disparate impact in the First Circuit. If that occurs, employers are sure to face the citation of the case by plaintiffs in other circuits as they argue for this analysis in other disparate impact lawsuits. As the First Circuit itself noted, when the size of a disparity is negligible, it is not meaningful. Nevertheless, because neither the parties nor the appellate court found an acceptable way to quantify “negligible,” at least for the time being, the decision opens the door for plaintiffs to pursue “practically meaningless” statistical disparities in the First Circuit.