EEOC's Proposed Age Discrimination Regulation Introduces Practical Significance

On February 18, 2010 the EEOC released new regulations that will address the meaning of "Reasonable Factors Other than Age" (RFOA) under the Age Discrimination in Employment Act (ADEA). Dennis Westlind of Stoel Rives has an excellent post on the law behind the proposed regulation.

To establish the RFOA defense under the new rules, an employer would have to demonstrate that the employment practice was both (a) reasonably designed to further or achieve a legitimate business purpose, and (b) administerred in a way that reasonably achieves tha purpose in light of the particular circumstances and facts that were known - or should have been known - to the employer. The rule also provides a non-exhaustive list of six factors to assess whether an employment practice is "reasonable":

1. whether the employment practice and the manner of its implementation are common business practices;
2. the extent to which the factor is related to the employer's stated busienss goal;
3. the extent to which the employer took steps to define the factor accurately and to apply the factor fairly and accurately;
4. the extent to which the employer took steps to assess the adverse impact of its employment practice on older workers;
5. the severity of the harm to individuals within the protected age group (both in terms of the degree of injury and the number of individuals adversely affected) and the extent to which the employer took preventative or corrective steps to minimize the severity of the harm, in light of the burden of undertaking such steps;
6. whether other options were available and the reasons the employment selected the option that it did.

From an economic and statistical perspective, the most interesting factor in this list is the severity of the harm, both in terms of the degree of injury and the number of individuals adversely affected. Based on this criterion, it would appear as though the finder-of-fact would be considering not only statistical significance, but also practical significance. This introduces an element of ambiguity and arbitrariness into the decision.

Unlike statistical significance, which has a discrete threshold (two units of standard deviation), practical significance has no such threshold. Practical significance refers to whether a difference is "big enough to matter". There is no formal definition of "how big is big enough".

Consider the following two examples:
Example #1: a formal analysis of a reduction in force finds that there is a surplus of two "older" workers selected for layoff relative to the eligible population. This surplus is statistically significant at four units of standard deviation.

Example #2: a formal statistical analysis of a reduction in force finds that there is a surplus of twenty "older" workers selected for layoff relative to the eligible population. While directionally adverse, the surplus is not statistically significant.

Which of the two examples is more problematic from an employer's perspective? The first example, where a statistically significant disparity is found but only two individuals are affected, or the second example, where there is a directionlly adverse outcome affecting twenty individuals, but the disparity is not statistically significant? Would your opinion change if you knew that the surplus found in the second example was equivalent to 1.90 units of standard deviation? 1.95 units of standard deviation?

There are potential implications for other questions outside of the reduction in force arena. Assume that a statistical analysis of compensation finds a $1 disparity between the annual salaries of men and women, and this disparity is highly statistically significant. Should the employer remediate this $1 difference? Is $1 "big enough" to matter? What if the difference was $10? $100? Where is the threshold over which a statistically significant disparity becomes big enough to matter?

While I do not think the concept of statistical significance will be (or even can be) dismissed, reasonableness factor #5 on the above list may blur the line of how to interpret a statistically significant disparity. As always, matters will be decided on a case-by-case basis, examining the particular facts of the situation at hand. But if the gray area of practical significance is given more weight in the decision-making process, additional ambiguity may be introduced and finders-of-fact may have more latitude in determining whether an employer's actions were "reasonable".