Jim Higgins - Rescinding The 2006 Compensation Analysis Guidelines

While Dr. Patricia Shiu of the OFCCP announced that the current OFCCP Compensation Guidelines were to be "rescinded" at the 2010 National Industry Liaison Group (NILG) meeting in Las Vegas, employers should be aware that this announcement will have little effect on how they will defend themselves during compensation discrimination-related litigation or an OFCCP audit. This is because the question in such situations is "after taking into account the legitimate reasons why two groups (e.g., males and females) differ in terms of the pay they receive, are there unexplained differences that could be attributed to their gender (ethnicity, age, etc.)?" This question is most effectively answered through the careful application of the statistical tool known as Multiple Regression Analysis--the very tool that the OFCCP has settled upon in the 2006 Compensation Analysis Guidelines.

It is the power of Multiple Regression Analysis to "control for" legitimate job-related factors while determining the effect of a factor like gender that makes in the perfect tool for use in compensation analysis. It is also the reason that the courts have used Multiple Regression to both determine whether compensation discrimination exists and, if so, how much the disadvantaged group would need to be paid to eliminate the statistically significant disparity.
"SO," you might ask, "if all of this is true, why does it appear that the OFCCP may be reducing their reliance on Multiple Regression Analysis?" The answer is simple. The OFCCP spent a significant amount of time researching the tools and techniques necessary to conduct a valid compensation analysis. You only have to review the discussion published in the Guidelines to see this. The fact of the matter is they did a great job of identifying the relevant issues and choosing the right tool for the job.

What's the problem, then? Essentially, that it takes a significant amount of expertise to do Multiple Regression correctly. The OFCCP understood this problem well which is one of the reasons that they frequently used a "trigger" or "red flag" test as a pre-screen to identify those federal contractors whose data suggested they were systemically discriminating based on a protected characteristic. Even this pre-screening technique, however, failed to help them uncover significant amounts of actionable pay disparities.

Now, the OFCCP is looking for alternative analytical methodologies. However, they did such a great job (from a statistical standpoint) that they will be hard pressed to find anything better.

So, what is the bottom line? Federal contractors would be best served by continuing to conduct proactive compensation analysis using Multiple Regression Analysis as the cornerstone--where appropriate. In cases where unexplained pay differences are found, employers should conduct a thorough cohort analysis as a follow-up. Only after finding unexplainable differences in pay using Multiple Regression that were confirmed through a cohort analysis should the contractor make adjustments to employee pay. In this manner, federal contractors are likely to realize maximum protection from OFCCP audits or pay-related litigation.

To learn more about compensation analysis and multiple regression, visit me on the web at www.AffirmativeActionServices.com.