Gather statistical evidence to show you don’t discriminate

by on
in Discrimination and Harassment,Employment Law,Hiring,Human Resources

When it comes to proving a pattern and practice of discrimination, courts rely heavily on statistics. Employees who can show that a company routinely discriminates against members of a particular protected class will have a much easier time showing that, as members of that class, they were discriminated against, too.

Perform your own statistical analyses to test your hiring practices for hidden discrimination.

A good statistical analysis involves looking at the demographic composition of your potential applicant pool, breaking it down by protected classes. Then calculate how many qualified potential candidates who are members of a protected class you actually hired. The numbers should roughly match.

If they don’t, it’s time to call in a statistical expert. He or she can perform the specific statistical tests (such as multiple regression analyses) to determine whether there is hidden discrimination.

Recent case: Melinda Carney, who is black, applied for a job as a Denver police officer.

After several tries—and an EEOC complaint—she was finally allowed to start training. She never completed the training, however, because she was injured several times. After going out on disability, she sued, alleging race discrimination was to blame for the long time it took for the department to hire her.

Carney tried to prove her case by showing that just 1% of Denver officers are black females, while the general population of black females in the Denver area is far higher. That, she claimed, was statistical proof of discrimination.

Not so, said the court. It pointed out that the relevant comparison would be to the number of qualified black females in the Denver area, not the entire population of black females in Denver. (Many might be unqualified for the job, given their ages, physical condition and education.) The court dismissed her case. (Carney v. City and County of Denver, No. 06-1490, 10th Cir., 2008) 

Leave a Comment