It’s tricky to use tests to determine which employees are eligible for promotion.
Reason: Employment tests generally are a legal minefield. If members of a protected class score significantly lower than other groups, you can expect a lawsuit alleging the test had a disparate impact. And as the following case shows, even court-approved testing procedures are no protection.
Recent case: A group of Akron firefighters sued for race and age discrimination when they lost out on promotions to lieutenant and captain after scoring low on a qualifications test.
Because at least 80% of the protected-class group scored as well as other candidates, it met the so-called “4/5ths rule,” which courts long ago created to determine if a test was discriminatory.
However, the judge in this case allowed more advanced statistical tests. Based on those statistics, a jury decided the test did have a disparate impact on minority and older firefighters. (Howe v. City of Akron, No. 5:06-CV-2779, ND OH, 2010)
Stats that overwhelm juries
The job-testing world is a complex one. Designing tests that are “job related” and “of business necessity” is just the first step. Employers must also ensure the tests withstand charges of disparate impact discrimination.
In other words, do minorities, women or older workers perform significantly worse than other workers?
In the Akron firefighters’ case, the plaintiffs analyzed test results to determine if the exam disparately impacted blacks and older workers.
This complex realm can be a gold mine for lawyers trying to convince jurors that their clients got a raw deal. Because most jurors lack a sophisticated understanding of statistics, a good attorney can use expert testimony to so overwhelm a jury that it can’t help but rule against the employer.
The crucial margin of error
Traditionally, Ohio courts have relied on the 4/5ths rule. But the 4/5ths rule is a creation of the courts, not statisticians. Most significantly, the 4/5ths rule does not account for different sample sizes—and sample sizes determine the margin of error for each test.
We often hear the term margin of error in the context of election polls—for example, that a poll had a margin of error of +/- 3%. The smaller the group polled, the greater the margin of error.
The same concept applies to job tests: the fewer the test results, the greater the margin of error when applying those results to a larger group.
Statisticians use many methods to calculate margins of error—T-tests, Z-tests, chi-squares and binomial distribution. Presented with alternatives to the relatively simple 4/5ths rule, the jury in the Akron case concluded that the promotion exam was almost certainly biased.
Advice: If you use tests to decide whom to hire or promote, have your attorney spearhead regular reviews of the tests’ vulnerability to charges that they are biased. To win lawsuits like this one, you will need to consult statisticians to anticipate the challenges that are likely to show up in court.
Like what you've read? ...Republish it and share great business tips!
Attention: Readers, Publishers, Editors, Bloggers, Media, Webmasters and more...
We believe great content should be read and passed around. After all, knowledge IS power. And good business can become great with the right information at their fingertips. If you'd like to share any of the insightful articles on BusinessManagementDaily.com, you may republish or syndicate it without charge.
The only thing we ask is that you keep the article exactly as it was written and formatted. You also need to include an attribution statement and link to the article.
" This information is proudly provided by Business Management Daily.com: http://www.businessmanagementdaily.com/13333/bias-lawsuit-alert-if-you-use-job-tests-your-legal-team-better-include-statisticians-too "
- Confidentiality rules protect EAP counselors
- Are we vulnerable to reverse discrimination claims because of our 'early out' program?
- OK to reject applicant who volunteers that disability can't be accommodated
- Class actions exploded in '08, employers continue to pay the price
- Don't rely on written policy to prevent harassment