Statistics, competing interests, job-relatedness and business necessity — all this and more from two experienced attorneys who have provided invaluable commentary on our post this week on “disparate impact” discrimination.
We wrote that “While courts have often held that although a standardized test may be neutral and non-discriminatory on its face, and therefore demonstrate no discriminatory intent, nonetheless the impact of such a test may disproportionately fall upon members of a protected class, and therefore be found to be discriminatory.”
Our experts weigh in:
Daniel Kinsella, an attorney in the Chicago area:
“Richard, first, thank you for identifying “disparate impact” as “unintentional discrimination.” That, I believe, is a very important distinction with “disparate treatment” cases. We must go back to Griggs v. Duke Power to understand the statutory basis for such claims of unintentional discrimination.
Second, statistics are too often misunderstood and misused, even in litigation. You toss around the term “statistically significant” like a throw-rug. Since I, like most lawyers, slept through at least one statistics course as an undergraduate, arguing something is or is not “statistically significant” requires a statistician, i.e., someone who actually took more than one statistics course, stayed awake and really enjoyed it. (That was the guy, or girl, you went to get class notes when you slept through yet another class; thick glasses, pocket protector, slide-rule — showing my age — or calculator on a holster on the belt.)
A statistician can interpret and analyze the data and determine how far off the median (or mean) a given distribution is. That distance is the number of standard deviation units.
In short, a high percentage difference between numbers does not automatically make a “statistically significant” difference. Conversely, a low percentage difference does not automatically translate to no statistical significance. Other factors, such as sample size, are important. Back of the envelope math doesn’t cut it.”
Richard Seymour, an attorney in the Wash., D.C. area:
“This is a useful posting. For much of my professional life, I’ve prosecuted disparate-impact cases. Once the practice is identified and its impact shown — both sometimes very difficult — the focus turns to the question whether the practice is job-related and consistent with business necessity.
For example, police hiring tests used to be reading comprehension tests heavily dependent on knowing the vocabulary of police work. That was a built-in advantage to white applicants with family members already in the police department, but involved terms quickly learned in training.
That kind of test is hard to defend and easy to challenge.
Take the example one step further, and look at the next generation of police hiring tests, which did not use job-specialized vocabulary but were tests of reading comprehension. Up to a point, that is fine. Recruits have to be able to read well enough to understand statutes and regulations and some court decisions and all their training materials. It does no good to get African-American and Hispanic recruits into a Police Academy that uses fair graduation criteria and that they cannot finish.
Often, such a case is resolved when one side or the other identifies a practice, often as simple as a different way of scoring the same test, that serves the employer’s interests as well as the original practice, but with a meaningful reduction in adverse impact. For example, instead of ranking all candidates by orders of their scores and hiring form the top down, one can use the test as a hurdle, identifying the score that represents reading comprehension sufficient to satisfy the Police Academy’s needs and the Police Department’s operational needs, but not give additional credit for higher scores unrelated to those needs. Reading twice as well as needed does not make one twice as good a police officer, so why pretend that it does by rank-ordering people?
What is next most important is situational judgment, so that needs to come into the selection process. Tests of situational judgment tend to have little or no adverse impact, and having twice as much situational judgment really does relate to being twice as good an officer, so rank-ordered scores on such tests tend to be very job-related.
Put the two together and it’s a win for both sides: a selection process that works much better for the employer, has much less adverse impact against members of minority groups, and tends to select the best-qualified persons of all races.
Disparate-impact cases frequently involve competing interests, such as physical-ability tests and the needs of police and firefighting jobs. There, too, there are ways to reduce adverse impact. Letting everyone know what the physical testing will involve, and giving them a chance to take gym classes and work out, to meet the tests, both lets applicants know what is expected and significantly reduces adverse impact against women. Experts may also be able to work with the particular elements of the physical ability test to make sure they are calibrated to the job and not just a bright idea someone had.
I confess that I have found working with experts and employers in a joint effort to get the job done extremely enjoyable. It’s one of the areas of the law where one can be creative, and see one’s efforts winding up benefiting both sides.”