Here’s a weird one.
Employers originally began using standardized tests to try to find a more objective and hopefully accurate way to select the best candidates for promotion and hire. Over time, however, such tests have also been used to avoid disputes, contractual, statutory, or otherwise, over whether they have selected the right candidate. Put another way, one reason employers may test is to avoid claims of discrimination. So how can an employer administer tests in such a way as to discriminate against BOTH whites and African-Americans? Somehow the City of Akron, Ohio managed that and to discriminate on the basis of age as well.
In Howe v. City of Akron, Case No. 11-3752 (6th Cir. July 22, 2013), the City of Akron’s fire department developed and implemented a procedure to promote candidates to lieutenant and captain positions. The core of the procedure was a 100-question multiple-choice test. The top three candidates were chosen based upon the test for interviews, but apparently the interviews proved largely irrelevant because candidates were promoted in rank order based on their test performance.
The results were all over the map. The pass rates for each of the categories of race and age were each 75% or higher. In the case of lieutenant positions, however, whites were promoted at the rate of 36% while blacks were promoted at the rate of only 20%. For captaincy positions, the numbers were reversed, with blacks promoted at the rate of 71% while the rate for whites was only 27%. With respect to age, those under 40 were promoted to lieutenant at the rate of 38%, while those over 40 had only a 38% rate. The opinion does not disclose the reason for the disparities or the parties’ arguments, if any as to the sufficiency of those reasons.
In 2006, twenty-three unsuccessful promotion candidates brought suit against the city, relying primarily upon disparate impact claims under Title VII and the ADEA. The case was tried before a jury, which found in their favor. Because the jury awarded essentially the same amounts to every plaintiff, the trial court ordered a new trial as to damages, but did award injunctive relief in the form of promotions to the positions they had sought. The damages aspect of the case was not part of the following appeal, but the jury’s decision to award the same amount to each plaintiff in such an unusual case very well should have raised some eyebrows.
It was odd enough that the same test procedure was found by a jury to discriminate almost simultaneously against both blacks and whites. The Sixth Circuit’s decision too, however, was equally strange.
First, because the district court had not yet decided the issue of damages, technically there was no final judgment, and the Sixth Circuit treated the injunctive relief under preliminary injunction standards. It ultimately concluded that in light of the jury verdict, the firefighter’s careers, and the City’s own inaction in filling open positions, the injunction was proper.
Second, along the way, the Sixth Circuit uncritically noted that the district court had permitted the plaintiffs to use the “four fifths” rule to establish their disparate impact claims. That rule, however, has questionable validity if it can even be called statistical analysis. See, e.g., Isabel v. City of Memphis, 404 F.3d 404 (6th Cir. 2005) (rejecting results of four fifths rule in favor of multiple regression analysis; stating“we are grateful for statistics beyond the four-fifths rule analysis because we prefer to look to the sum of statistical evidence to make a decision in these kinds of cases”) The Sixth Circuit did not even cite its own prior cases, and noted that “the rule is not dispositive,” but went on to draw distinctions between pass/fail and other kinds of examinations. That analysis simply misses the point – the four-fifths rule simply isn’t a very good or accurate test.
Third, nothing in the opinion addresses the very real difference between simply passing a test and doing well on a test. As the facts above suggest, the claimed disparity was not between who scored the highest on the test, but rather comparing the pass rate to the selection rate. If the test did not accurately measure the best candidate, the opinion does not disclose why.
Unfortunately, much of the remainder of the court’s opinion is mired in objections it found to have been waived and practical issues it concluded were of the City’s own making. It ultimately affirmed the decision of the district court.
The Bottom Line: Testing is an area that remains fraught with danger for employers and can result in both multi-plaintiff and class action claims. Courts continue to struggle with statistical arguments.