Is “racial balance” always discriminatory?

In a recent case out of the Sixth Circuit, the court addressed the concept of “racial balance,” finding that an effort to achieve racial balance in disciplinary measures constitutes direct evidence of discrimination. While this is by no means the first case to deal with “racial balance” and discrimination, I am wondering: are all “racial balance” cases created equal?

In Ondricko v. MGM Grand, the plaintiff, a white woman, claimed reverse race discrimination (and sex discrimination) after she was fired from her job as a floor supervisor in the casino. Ondricko was ostensibly fired for participating in a “bad shuffle” at a blackjack table that she supervised. This type of incident is apparently not uncommon, and the court had at least six other similar incidents to compare involving white and black men and women who had engaged in similar conduct and whose discipline varied from several-day suspensions to terminations.

The “smoking gun” in this case is the interesting part. Four months before Ondricko’s incident, a black woman was terminated for her involvement in a similar incident involving unshuffled cards put into play. Around the time that supervisors were discussing the appropriate discipline for plaintiff, two managers had a conversation in which one noted that the black woman’s lawyers had called and wanted to know how the casino was going to handle Ondricko’s case (presumably because they viewed her as a similarly situated comparator). The other manager responded by saying, “do you think I wanted to fire [Ondricko], I didn’t want to fire [her], how could I keep the white girl?” The Sixth Circuit determined, based on this statement, that a reasonable jury could conclude that race was a motivating factor in the decision to terminate. “[I]t is certainly reasonable to conclude . . . that MGM was motivated by a desire to be racially balanced in its terminations for misconduct related to shuffling.” In support of this proposition, the court cited another Sixth Circuit case involving a school board’s attempt to be racially balanced in the hiring of school employees. And that was essentially the end of the court’s analysis.

But Ondricko was not an affirmative action case nor was it a case about achieving racial balance in hiring. Instead, the case was about insuring racial balance in the employer’s discipline of its employees. The Sixth Circuit did not see a distinction between these two types of “racial balance” cases, but I think that is a flawed view. This case may not be the best example because the desire to mete out the same discipline across races was expressed in response to a call from a lawyer but what if that had not been the case? Although race is technically a motivating factor when an employer attempts to be “racially balanced” in its approach to disciplining employees, is that the type of case Title VII is intended to cover? Shouldn’t we, on some level, be encouraging employers to be mindful about race when meting out discipline and to insure that they are treating employees of all races the same? If they don’t, they risk disparate treatment claims for treating employees differently based on race. To call race a “motivating factor” in this type of case and not discuss the potentially legitimate reason for consideration of race seems to be a flawed or, at least, an incomplete analysis.

I think there is an analogy here to Ricci v. DeStefano, in which the Supreme Court recently concluded that an employer’s fear of disparate impact litigation is only a legitimate basis for intentional discrimination when the employer possesses a “strong basis in evidence” for believing that a valid disparate impact claim can be asserted. The Court in Ricci may have made my argument about Ondricko and racially balanced discipline more problematic, but I would be interested to hear others’ views on this issue.


You may also like...

13 Responses

  1. Michael Yuri says:

    I think this could go either way depending on the precise facts.

    Suppose a black employee was accused of certain misconduct and fired as a result. Later a white employee is accused of the same type of misconduct.

    Category 1: The white and black employees’ alleged misconduct is basically identical (or the white employee’s conduct was worse).

    Here, an employer who decides that the company cannot retain the white employee when the black employee was fired for equivalent (or even less culpable) conduct, might fairly be characterized as just being “mindful about race when meting out discipline [] to insure that they are treating employees of all races the same.”

    Category 2: The white employee’s misconduct is less culpable than the black employee.

    Here, the same statement could fairly be viewed as prioritizing “racial balance” over fair treatment of individual cases.

    Imagine a very stark example. A black employee in a retail store with a string of prior disciplinary warnings in his file is caught on video pocketing several hundred dollars from the register. As a result he’s immediately fired. The next day a white employee’s till comes up $20 short. A review of store video shows that she gave a customer who had paid with a $100 bill an extra $20 in change. The customer happened to be a personal friend of the employee.

    Now imagine the store manager says the following: “Look, [white employee] Jane’s a great employee. She’s worked here for years, and I’ve never had any problems with her before, and she’s a conscientious hard-worker. I’m sure the short till was an accident. But how will it look if I keep the white girl on when I just fired [black employee] John yesterday? My hands are tied – I have to let her go.”

    That sure looks like someone getting a raw deal solely on the basis of race.

    The problem is, in real life, it will very often be unclear whether a situation belongs in category 1 or category 2. Different employees are almost never in an identical position, and there will frequently be plenty of room for reasonable argument over whether two people are really similarly situated. I don’t really see any solution apart from careful parsing of the particular details of each case, but I don’t have much confidence that this will reliably lead to the right results.

  2. TJ says:

    I think Ricci made your argument both easier and harder. It made it easier in the sense that it says that racial balancing is permissible when there is a strong basis in evidence that a disparate impact claim could be made. That might be a tough standard to meet, but it is extremely helpful in establishing your conceptual argument that avoidance of disparate impact litigation is a permissible non-discriminatory motivation.

  3. Brett Bellmore says:

    “Shouldn’t we, on some level, be encouraging employers to be mindful about race when meting out discipline and to insure that they are treating employees of all races the same?”

    But this presumes that their employees of all races are acting the same. Because only if they’re acting the same, would treating them the same result in the same discipline rates. We want the employers to treat similarly situated employees the same. Not levy excessive punishments to one group, and extend excessive leniency to another, just to make statistics look good.

    But there’s no particular reason to suppose that employees of different races will, in fact, be similarly situated in the same proportions. Crime statistics, including victimization surveys, demonstrate massive differences in crime rates between different races. Why, then, would we expect employee misconduct to be equal across these same groups? Rather, it would be irrational to have any such expectation.

    Applying pressure to achieve discipline rates that are the same between racial groups can hardly be expected, regardless of any ambiguity in given cases, to improve the fit between treatment and behavior. It can only increase injustice.

    It is, in short, discriminatory.

  4. Michael Yuri says:

    @Brett Bellmore: “Applying pressure to achieve discipline rates that are the same between racial groups can hardly be expected, regardless of any ambiguity in given cases, to improve the fit between treatment and behavior.”

    But this case, as described in the post, has nothing to do with aggregate statistics — it’s about the reaction to specific cases of employee misconduct. If an employer fired a black employee for misconduct, and then said “now we need to find three white employees to fire to keep our numbers proportional”, of course that would be blatantly discriminatory.

    But here we have an employer faced with a specific case of alleged misconduct who says essentially: “I can’t keep the white employee after firing the black employee.”

    This could mean: “I need to be conscious of the way I’ve treated other employees and not treat this employee differently due to (possibly unconscious) racial favoritism.” That seems unobjectionable.

    But it could also mean: “Regardless of any actual differences between the incidents, I need to impose the same punishment on the white employee to avoid an appearance of discrimination.” That seems very problematic.

    I just think it will be very difficult in practice to distinguish between the two.

  5. Anon says:

    I’m having a hard time articulating it in my head, but I think there’s an important distinction between Ricci and Ondricko (as described in the post; I haven’t read it). In Ricci, the alleged discrimination against the white firefighters (discarding the results of exams on which whites scored higher) was intended to remedy alleged discrimination against the minority firefighters (resulting from the disparate impact of the exam). Thus in Ricci, we’re faced with a problem of balancing the rights of white firefighters to rely on the results of an exam with the rights of minorities not to have a biased exam. (The best solution, of course, is not to use a bad exam in the first place). But the situation in Ondricko seems different. The firing of Ondricko doesn’t seem to remedy the allegedly discriminatory firing of the minority employee; it feels like a blatant attempt to avoid creating a bad litigating position in the suit involving the previous employee. Ondricko doesn’t feel as zero-sum to me as Ricci does. The two terminations in Ondricko are theoretically unrelated. Whether the termination of either employee was discriminatory has nothing to do with whether the termination of the other was discriminatory. In Ricci, on the other hand, it seems like (once the test was given) there had to be discrimination against one set of firefighters or the other (assuming, as the majority does, that the white firefighters had a right to rely on the results of the exam; for the record, I tend to agree with Ginsberg’s dissent that there was no legitimate expectation to upset by throwing out the test results).

  6. Brett Bellmore says:

    “Thus in Ricci, we’re faced with a problem of balancing the rights of white firefighters to rely on the results of an exam with the rights of minorities not to have a biased exam. ”

    Of course, the problem with this view is that there wasn’t any basis for thinking the exam biased except that the results weren’t what were wanted. But if you’re going to use mere statistical disparity as proof of bias, you have neatly transformed an anti-discrimination rule into a simple quota.

  7. Anon says:

    I’m not sure it’s right to say “that there wasn’t any basis for thinking the exam biased except that the results weren’t what were wanted,” but maybe I was sloppy in my original use of the word “biased.” What I meant by biased exam is an exam that has an unjustified disparate impact on minorities, emphasis on the word unjustified. Under this understanding of “biased,” I would agree that just the fact that the results “weren’t what were wanted” (i.e., scores that varied significantly by race) wouldn’t make an exam biased. But that’s because unwanted results alone don’t make the disparate scores necessarily unjustified. However, it is certainly debatable whether the exam in Ricci resulting in the disparate scores was justified, in the sense of accurately testing the skills required to perform the job.

  8. Joe says:

    “wasn’t any basis for thinking the exam biased except that the results weren’t what were wanted”

    From Ginsburg’s dissent in Ricci:

    Bridgeport, Day observed, had once used a testing process similar to New Haven’s, with a written exam accounting for 70 percent of an applicant’s score, an oral exam for 25 percent, and seniority for the remaining five percent. CA2 App. A830. Bridgeport recognized, however, that the oral component, more so than the written component, addressed the sort of “real-life scenarios” fire officers encounter on the job. Id., at A832. Accordingly, that city “changed the relative weights” to give primacy to the oral exam. Ibid. Since that time, Day reported, Bridgeport had seen minorities “fairly represented” in its exam results

    If two possible exam approaches are available, approaches based on merit, why wouldn’t we “want” the one that is more likely to include more diversity? If the result are skilled personnel, encouraging various groups to be included is something basic to our country. By race, religion, gender, location, nationality etc., diversity was a means from the very beginning to bring everyone together while also promoting equality in the long run. Diversity, e.g., is what Madison hoped for in the House of Representatives.

    There are any number of ways to do things and some ways unnecessarily burden certain groups. Furthermore, certain groups have burdens because of illicit discrimination and other struggles (thus, certain tests are crafted to help those with disabilities). We can “want” to craft tests to help them.

    A “quota” is a set number of slots & like Asian groups have signed on the briefs in Fisher v. Texas to reject the argument they are being blocked by that policy, merely crafting a test to address the above is not a “simple quota.” Words have meanings. A “simple quota” would be a hard and fast 100 slots or whatever. This isn’t occurring.

    It is not a problem to look at a situation and determine some sort of discrimination is occurring in part based on lack of numbers. Brett might not like our civil rights laws that use that as a red flag — it isn’t the ONLY thing used & can be overridden — but there is nothing “simple” about as a complete issue.

  9. Veracitor says:

    What Ginsburg doesn’t tell you is that industrial psychologists have determined that subjective oral exams like Bridgeport’s are much less predictive of employee performance as well as less reliable (in the sense of consistently ranking candidates) than objective exams (commonly written tests). The valid purpose of oral exams is to rank candidates on gregariousness, and the improper purpose is to give examiners a chance to exercise their biases. It is proper to use oral exams to weed out people with strange affects or poor communication skills, because those are disruptive in the workplace. However, above the threshold of acceptable social competence oral exams don’t tell you much. The only practical reason to weight oral-exam scores more than written test scores is to give supervisors a way to hide racial discrimination (and sexual-orientation discrimination, and sex discrimination, and so-forth)– because oral exams conveniently generate no evidence for independent review.

    “Since [switching to oral exams], Day reported, Bridgeport had seen minorities “fairly represented” in its exam results…”

    You bet! The whole point of switching to oral exams was to ensure there would be no paper trail revealing that supervisors were selecting candidates by racial quotas (probably using unlawful “race norming” to fill each quota with the best candidates of each race).

    Remember, the New Haven firefighter exam in Ricci had been prepared with the utmost care by costly “diversity” consultants in a long process under the direct supervision of a panel overstuffed with minority firefighting managers. That exam produced “disparate impact” only because ALL objective tests with any job-related validity partly measure general cognitive ability– and major races have different average cognitive ability levels.

    The more complex the job, the more job performance depends on general mental ability (see for a handy primer and survey). So long as each major racial group displays a different average level of mental ability, objective employment screening tests (which are not racially-biased in any way) will continue to produce “disparate impact.”

  10. Joe says:

    Oral exams have various “valid” purposes beyond determining “gregariousness” or to determine “social competence” though that is a value when your job includes dealing with people. One value would be an ability to think on your feet, another would be to find out if some show ability even if they might not do quite as well on written tests, thinking on your feet,* including during the stress of orals, part of being a firefighter.

    And, the oral exam was just part of the score at any rate while written exams were also found to have problems, thus the value of a certain mix of the two.

    Red flag alert: “the whole point.”

    Also, “remember.” That word is repeatedly used to state debatable things that are cited as clear as day. If so, one would need not to be “reminded.”

    The last point references “general mental ability,” which is mighty vague, especially when followed by the argument that major racial groups different levels of mental “ability,” not specific differences in training or such.

    * Per the dissent, quoting factual finding, “addressed the sort of “real-life scenarios” fire officers encounter on the job.”

  11. Joe says:

    The conclusionary “which are not racially-biased in any way” is also repeatedly shown to be false.

    Also, yes, the law in place is in part to deal with “disparate impact,” which occurs in various ways that do not match apparent differences in “general mental ability.”

    Differences in “general mental ability,” e.g., didn’t stop women from joining various professions and employment opportunities and so-called “objective” tests that were set based on “norms” (at times, just taken for granted, even if alternative ways to judge competency were present) with traditional male applicants in mind helped.

  12. Veracitor says:

    Joe, perhaps you should read Gottfredson (linked above), since you seem uninformed on what constitutes general mental ability. I also recommend you to read an article by Prof. Amy Wax of UPenn law school, available online in draft:

    The way we can tell whether a test is racially biased is by checking whether it predicts (academic or job) performance equally well for people of all races. That check is part of “test validation.” Just for example, the SAT test is NOT racially biased. The SAT produces “disparate impact” because test-takers of some races score better or worse on the average than test-takers of other races. The SAT is substantially a test of general mental ability and predicts individual academic performance fairly well (and equally well no matter what the individual test-taker’s race), but when you tabulate candidates’ scores by race, you find disparities in group performance. The test does not cause such disparities, it merely reveals them.

    No evidence* was adduced in the Ricci (New Haven) case to show that the test was biased. The city claimed (at the last minute) and supporting amici simply assumed that the test was biased because it produced disparate impact, which is logically invalid and legally dubious (the EEOC’s 4/5 rule and so-forth notwithstanding).

    *Some amici in the business of implementing “assessment centers” for civil service testing criticized the New Haven test, not as biased, but as inadequately prepared and validated. Their assertions were criticized in turn:

  13. Brett Bellmore says:

    If your only basis for asserting that a test is biased is that it doesn’t produce the numbers you’re trying to achieve, and you’re willing to manipulate the test until it does produce those numbers, then, like it or not, you are implementing a quota. With the test as just a way to obscure that. It’s always a quota if the only way you can accept that bias wasn’t present was getting the ‘right’ numbers.

    And the idea that oral exams are more objective than written tests, a way to avoid bias in testing, is laughable. Genuinely laughable. Really, the only difference between an oral exam and a written exam is that the former gives you a chance to observe the race of the applicant. Aside from a limited utility in permitting you to confirm applicants are capable of communicating verbally, they are to be avoided any time you want to eliminate bias.

    The problem in New Haven wasn’t that the test was biased, it was that they didn’t LIKE the results of an unbiased test.