Is that "random" drug test selection really random?

Many employers -- particularly, those in the transportation industry -- use third party vendors to do the random selections for federally mandated drug and alcohol tests. I think it's a great idea, because it prevents employees from claiming that they were selected for "random" testing in a not-very-random process.

Generally, the employer will provide to the vendor a list of the eligible employees without any information about race, sex, national origin, or any of those other protected categories that can cause a problem. The third party feeds the data into a program, and the program does its proprietary algorithm thingy and spits out the names of the lucky individuals who get to be tested each period.

What could be easier? Or less subject to legal challenge?

Well, a federal magistrate judge in California has denied summary judgment to a transportation employer in a race discrimination case* brought by an African-American supervisor who felt that he'd been "randomly" selected too many times. He'd been selected three times in nine months, which is indeed a lot.

*The discrimination claim was brought under the California Fair Employment and Housing Act, not Title VII.

But if you know anything about "random selection," you know that this sometimes happens. If you flip a coin 100 times, you may get heads 50 times and tails 50 times. But you may also get heads 100 times. (Well, ok, maybe 55 times?) In any event, the same idea applies with random selection for drug tests: Sometimes the same individual will be chosen multiple times while a co-worker may escape altogether.

If the testing entity doesn't even know the races of the individuals, how can the selection be discriminatory? The judge in this case apparently based his decision on the fact that a white manager (who I don't believe had anything to do with selections for random drug tests) allegedly made comments about the plaintiff's "player-mobile" and "pimpmobile." But the real killer, I think, was the fact that a white female who should have been in the testing pool was left out and did not remember having been tested since 2009.

(In the company's defense, other white employees were apparently in the testing pool, and the company said that the omission of the white female was a clerical error. But the judge wasn't buyin' it. He said that a jury should decide whether the employer "manipulated" the pool.)

Which reminds me of another statistical principle: garbage in, garbage out. Employers, make sure that your drug testing rosters are complete, correct, and current. Otherwise, you may not have the defense that the selections were truly random.