Wednesday, July 01, 2009
Strong Basis in Confusion
I don’t want to address the specific facts of the Ricci case, since specific facts aren’t what Supreme Court decisions are (supposed to be) about. I want to try to figure out, based on this case, what employers are supposed to do when they use a criterion – any criterion – that may have a ‘disparate impact’ on minority candidates.
According to Title VII of the Civil Rights Act of 1964 (as amended in 1991), there are two varieties of unacceptable discrimination: disparate treatment and disparate impact. "Treatment" is the straightforward kind of discrimination that announces itself clearly, the "we don't serve your kind" variety. Treatment is assumed to be intentional. "Impact" refers to outcomes that may or may not have been intentional, but that have the effect of disadvantaging one group as against another. The Act stipulates that criteria that have disparate impacts are presumptively invalid, unless and until the employer can show 'business necessity' and a lack of better alternatives. (The employer also has to show that the business necessity is not 'pretextual' -- in other words, that it's not just a fig leaf to mask another agenda.) In this case, the City of New Haven threw out a written test it used to determine promotions within the fire department, on the grounds that the results of the test showed a disparate impact on minority candidates. The City feared that it would be held liable under the 'disparate impact' standard, so it threw out the test after it had been administered. Some white candidates who had done well on the test sued, claiming disparate treatment based on race -- arguing that whites were singled out on purpose -- and won, 5-4.
As Justice Kennedy correctly put it in the majority opinion, "[O]ur task is to provide guidance to employers and courts..." (p. 20) Exactly so. I'm looking for guidance. Let's say that I want to comply with the law, as delineated by the Court. What would compliance look like?
Justice Kennedy holds that actions taken to remediate disparate impact are themselves disparate treatment. Drawing on 14th amendment rulings -- although at pains to say that this case wasn't about the 14th amendment -- he allowed only a "strong basis in evidence" threshold for exceptions. In other words, unless you can show a "strong basis in evidence" that you're guilty of disparate impact, you can't engage in disparate treatment to remedy it. (As Justice Scalia correctly notes in his concurrence, Kennedy's opinion doesn't address whether 'impact' trumping 'treatment' can ever make sense in the first place.)
He doesn't define "strong basis in evidence," but it must be a pretty high threshold. I know that because the majority decision didn't remand the case for reconsideration under the new rule. Instead, it simply declared that the city couldn't possibly meet the standard, so it declared a winner and closed the argument. If you know already that the threshold couldn't possibly be met, it looks less like a threshold and more like, well, a pretext. After all, appellate jurisdiction isn't supposed to be about weighing the evidence. Since I have to assume that the Court knows that, I can only conclude that it decided that no amount of evidence could possibly suffice, by definition. It's pretextual.
The point of the pretext, as near as I can tell, is to render the Civil Rights Act unenforceable without actually overturning it. This becomes clear in the application. Let's say that my college does a search, and the applicant pool turns out to be almost entirely white. What, if anything, can the college do about it? If anything remedial amounts to disparate treatment by definition, and if the threshold for an exception is so high that no amount of evidence could possibly suffice, then what, exactly, is left?
I'm at a loss.
It gets worse. Later in the opinion Kennedy makes a point that the "strong basis in evidence" standard that might satisfy the Civil Rights Act, "we...do not hold that meeting the strong-basis-in-evidence standard would satisfy the Equal Protection Clause in a future case." (p. 25) So even if you somehow manage to thread the needle of the pretextual standard, the Court reserves the right to yank that away, too, using a different argument. The precedent is allowed to lean only in one direction.
Justice Ginsburg's dissent is a mixed success, but the line that jumps off the page is her confident, if somewhat resigned, declaration that "[t]he Court's order and opinion, I anticipate, will not have staying power." (p. 2 of dissent) To the extent that the Court's job is to provide "guidance," a declaration that the guidance won't have staying power doesn't inspire confidence.
As a hiring manager, I literally don't know what to do with this. I'm compelled by law to ferret out disparate impact, but forbidden by law from doing anything about it. Pre-emptive compliance with disparate impact will fail to meet the "strong basis" standard, since I can't prove I'd lose a lawsuit until I actually lost it. (As Kennedy put it, "[f]ear of litigation alone cannot justify an employer's reliance on race to the detriment of individuals who passed the examinations and qualified for promotions." (p.33) I can't just be afraid of losing; I have to actually lose.)
Left unaddressed, tellingly enough, is whether the reverse would also be true. Could I defend a disparate impact claim by asserting a strong basis in evidence that I'd get nailed for disparate (compensatory) treatment? Who knows?
As a citizen, I have my preferred outcome, but that's secondary. As a hiring manager, my primary need is clarity. If I'm going to be held accountable for following the law, I need to know what the law is. I need guidance. At that -- at the first task of the Court -- this decision is a manifest failure.
People forget that for every advantage given to anyone, it means a disadvantage is being applied to someone else.
Suppose you had a student complaint that a professor had decided not to count a particular assessment (which they had worked hard on and did well on) because, according to the professor, the results showed that the assessment wasn't valid? How would you rule on that?
IMHO, it seems like these two decisions are the Court's way of saying "we need to rethink race law, but we're not sure what comes next." Ginsburg is probably right--not because the decision was badly written, but because it deliberately provokes debate and litigation. HR folks may not like lawsuits, but remember the Common Law tradition is designed to operate this way; over the coming years, lawsuits will bring new ideas and arguments about race law in 21st century America, giving the courts opportunities to sort the issues out better than they could if forced to issue a clear-cut decision today.
Why should the "first task" of the Supreme Court be to impose "clarity" on a society that is still working things out? Any clarity it found would be illusionary at best, and artificially imposed at worst. The question in Ricci is genuinely hard, and personally, it's refreshing to see national figures with the humility to say "you know, I don't have all the answers right now--let's talk about this some more."
Second, I'm not a legal scholar, but share Ginsburg's suggestion that the case be remanded. If the SC is really establishing a new burden of proof, why not send the case back to the lower courts to let the particular facts of this case come to light?
This ruling is pretty simple, DD. You cannot win under any circumstances; you should hire whichever group has the best lawyers.
Seriously, the SCOTUS was like, "You mean we've been limiting civil rights law to the point where white people can't sue to get jobs? That's crazy talk! Roll 'em back!"
As you clearly explain, it seems to be making a particularly obscure policy that will result in a multitude of appeals until it is straightened out. Let a thousand lawsuits bloom!
Anonymous@6:20 - I have seen that done. What would you do if the page order on a multiple choice exam got scrambled when some of them were printed, and some students answered them in the order the questions appeared on the test? There is no way to know if the ones who did well actually did well or got lucky ... except for the one student who pointed out the problem after the exam was over.
I think the example falls simply because the scrambling of "some" tests is not an equitable testing of the students. Some were handicapped by the misprint, while perhaps others were not.
In this case, the situation seems fairly straight-forward. If the process was presented in advance, and administered evenly to all candidates, then changing the criteria after the fact "simply" because we didn't get the balance we had hoped for seems, well... unfair.
Interestingly I haven't heard any argument made in the general discussions that presented a case for the test itself being "unfair" or biased. The whole discussion (and perhaps the argument?) is based on the outcome.
What is that likelihood that this result happens to be one of many statistically likely outcomes that could be achieved if we give the test enough times to enough different groups? That there is a "distribution" that at times will result in "outlier" results simply through the stochasticity of the process?
Hard cases make bad law.
Someone on IHE quoted a couple questions that seemed to be checking for fact-based knowledge about the city.
I can't be sure of the whole test, but that certainly seems like applicable knowledge.
Also, I have not followed the case closely, but wasn't some sort of preparation necessary for the test? Tossing out the results because some people scored poorly may have meant that the test measured exactly what it was intended to measure: general knowledge based upon study of the relevant materials.
Off-tangent: This issue of tossing out a test because a certain group scores poorly has me a bit peeved. As a white instructor, I have often been accused by black students of assigning lower grades by to them because they were black when they scored poorly on exams. The material is the material. I can't change testing of basic factual material just because some students (of various demographics...it's NEVER just one sub-group) don't score well on the exam.
I mean, sometimes people do poorly on tests because they're not capable of doing well. That's why most tests for promotion can be retaken...just like college courses.
Kennedy doesn't go that far. What the opinion does say is that disparate statistics are not enough. You have to find "substantial evidence" that the policy/practice was not a "business necessity" and that less discriminatory alternatives were available.
The message is that you can't change the rules midstream just because the stats came out wrong.
First, the standard set by Title VII is "disparate impact", so you only have to see a problem, not diagnose it, according to long-standing law. That has been changed by this ruling.
My example was deliberately extreme. Where I have actually seen it done was in a class where good students did badly on a calc 2 exam. Were several problems harder than intended, for some reason that was not anticipated? The prof had no idea what led to some form of mass hysteria on those problems, but just re-did the exam. In that case, I'm pretty sure the one student with a good score was allowed to keep it.
Some instructors also "curve" an exam when the results are unexpected.
Where I have learned to watch for disparate impacts is with language. I get a substantial number of foreign students in my physics classes, and they always surprise me with what they don't know about regular cultural items that I tend to take for granted when writing problems. And that doesn't even scratch the surface for kids who learned English English as a second language.
For the record, I think that the Civil Rights act was trying to fix a problem that was as much about culture (a culture originally imposed by racists) as treatment based on skin color, and those cultural legacies have proven much harder to eliminate than any of us expected in the 60s.
In the middle of a recent search in my department, one of my colleagues complained about having too few female applicants. In our field, there is one place that essentially all ads appear. That's where we published our ad, so everyone had access to it. Should we have started over? What could we have done to get a different outcome? Why are we responsible for who applies?
Mangement is about process not product. 1) Everybody goes to college for free, 2) except for the objectively qualified who are rejected for not being diverse, students and faculty both. An advocate makes virtue of failure. The worse the cure the better the treatment - and the more that is required.
I sailed away on that little boat to heaven
And by some chance found a bottle in my fist
And there I stood, Nicely passin' out the whisky
But the passengers were bound to resisist.
For the people all said beware,
You're on a heavenly trip.
People all said beware
Beware, you'll scuttle the ship!
And the devil will drag you under
By the fancy tie 'round your wicked throat.
Sit down, sit down, sit down, sit down
Sit down, you're rockin' the boat.
Nicely-Nicely Johnson (Stubby Kaye), Guys and Dolls
Faced with bad law, the SCOTUS can either a. send the bad law back to Congress to be fixed; or b. take a stab at making law from the bench.
Thankfully, the SCOTUS this time chose to do a. which is arguably better than trying to do b.
[the law (1964) calls for race neutral opportunity- which the testing process in this case had out the wazoo- while simultaneously calling for (1991) race neutral results. Apparently, sometimes color blind opportunity does not result in colorblind results. This is the elephant in the room that *nobody* wants to deal with- but HR folks and supervisors have to deal with every day. Congress can try to "wish reality away" (we can all try to wish reality away) but sometimes a society has to choose between equality of opportunity and equality of outcome. The latest SCOTUS decision avoids taking a choice . . . ]
Equality of opportunity is a sick joke. Nobody steps up to the plate with the same number of pitches. Any sensible system makes sure people get at least a few swings and medical care for the occasional beanball.
Afterall, firemen are just brutes who haul hoses, right?
Many of the issues we grappled with in the 1970s* have slowy worked their way "into the mainstream" of occupations related to the military (like Law Enforcement and Public Safety).
The stakes are higher in these occupations . . . so the issues have been more closely scrutinized.
Oddly enough, for this forum of intelligent academics,none of the actual social science scholarship on this issue has been introduced yet.
Back in the 70s and early 80s we used to talk openly about sexism and racism in this country, and the challenges of achieving gender and race neutral opportunities vs. outcomes.
I think the conversation ended a couple of decades too early.
We have devolved into a dysfunctional argument set of assumptions and jingoism . . . many intelligent people take the first whiff of any meaningful discussion and run away as fast as they can. That is sad. There is so much more work to be done.
* What do you do when rational and relevant job requirements conflict with desired societal outcomes? What do you do when "enough" women can't meet the physical requirements of occupational specialties? What do you do when color blind promotions prevent the achievement of diversity goals?
Don't project your issues onto us. It's not exactly hidden knowledge that fire departments around the country are moving from written exams to "assessment centers" which have more holistic opportunities for assessment. It was a hard case, and it was a mess. But the thing to remember is -- would conservatives have been rooting for the plaintiffs if they had been black? Hispanic? Asian? Yeah, maybe. That's the issue.
yacp -- we used to talk honestly about race, but then we started talking about Cadillac-driving welfare queens, and so we stopped talking about race at all. It's not like Archie Bunker would've been part of Carter's base. And it's not like the military handles gays well.
No one has argued that the test was not relevant, no one has argued that it had a hint of racial bias to it. The city deliberately decided to not pursue that angle after the test, because they invested so much money before to make certain that the test was fair and relevant.
It really came down to, who studied hard for the test? Some did, some didn't, it wasn't a race thing.
Disparate impact analysis is no longer tenable, nor should it be.
*(can't say institutional racism anymore, unless we are talking about the "reverse" kind- we changed the language on that one very carefuyl a couple of decades ago for obvious reasons)
Have a nice political wilderness, and please try to avoid winning elections; too many people get hurt when you do.