OFCCP's Compensation Analysis
Have You Heard of the
"2% - 30 & 3" Test?

If the OFCCP could require contractors to prepare their own databases for regression analysis, their scientists could examine these results and pass judgment on their accuracy. As things now stand however, there is no requirement for contractors to do that much compensation analysis in conjunction with Affirmative Action Plan preparation. And so if the OFCCP wants to use this technique, it has to prepare its own database using contractor-supplied data. That has effectively brought the scientific staff at the OFCCP to a screeching halt. They are overwhelmed by the clerical task of database management. As a result, not much has been getting through the process.

The agency is attempting to resolve this problem by determining how it can examine preliminary data from a contractor’s compensation program and determine if that contractor is a good candidate for full-blown regression analysis. Like the "80% rule" in disparate impact testing, the agency has been searching for a tool to allow front-end examination of data. If no preliminary problems are suspected after applying the test, then the contractor would be exempt from the full-blown regression analysis process.

The Initial Screening Tool

As a gross qualifier, OFCCP statisticians have developed what is being called the "2% – 30 & 3" test. We asked Dan Biddle, Ph.D. at Biddle Consulting Group, Inc. to lend a hand in explaining what the agency is doing with this approach. Here is what he told us:

2% - 30 & 3 Test The OFCCP standard is to further investigate the compensation practices of a facility [establishment] if all the following criteria are met [or determined by applying the test]: 1. Differences between gender [groups] or ethnic based groups with regard to their average compensation levels are identified to be two percent or higher for a submitted job category, grade or salary band, referred to as a pay division. 2. The gender [group] or racial group bearing the negative difference in the identified pay divisions must represent 30 percent or more of the entire workforce in that gender or ethnic group. 3. The percentage of the workforce in the potentially affected group must be at least three times that of the percentage associated with the comparison group (if the compared group also presents negative differences) to ensure the differences identified are focused on a particular gender or ethnic class. So, there we have it: the 2% (Item 1), 30% (Item 2), and 3 (Item 3) rule. (Dan Biddle can be contacted at dbiddle@biddle.com)

This rule or test does not appear in the OFCCP’s regulations, and you won’t find it in the agency’s Compliance Manual either. Since it is a test designed by the agency to assist in managing its own workload and there is no additional burden placed on contractors, the OFCCP is not then obligated to publish a new regulation. At least, that’s the argument agency management makes.

There are still huge disagreements between contractors and agency officials however, and one key disagreement is in the definition of "Similarly Situated Employee Groups" (SSEG), which is the approach the agency encourages to create job groupings large enough to analyze with some statistical significance. The OFCCP has proposed contractors lump together similarly compensated jobs across job titles until a sufficient number of incumbents are reached that can then be analyzed. Contractors, on the other hand, are saying compensation analysis should be limited to incumbents within a specific job title. This difference of opinion may well have to be resolved in the courts. In any event, SSEG is a new category for organizing jobs and data in addition to EEO-1 Category, Job Group, Compensation Grade and Job Title.

Conclusions

For all of these reasons we believe it is important that contractors are aware of this new screening tool and how the OFCCP is using it. In a broader sense, if time demonstrates -- through support gained in the courts -- that the application of this screening process works, other employers beyond government contractors may need to be concerned.

The "2% - 30 & 3 test" will screen out many contractors and hopefully no further work will be required on their part. In those instances however, where preliminary results suggest a full-blown regression analysis of the company’s compensation practices are warranted, that should be taken seriously. While regression analysis results won’t constitute proof that illegal discrimination has taken place, it will be highly suggestive and require more time and effort to reconcile or dispute. A better use of time and money would be to proactively do a self-analysis utilizing the formula and if problems are surfaced, immediately investigate to determine the reasons.

Further Suggestions

1. Carefully analyze and plan how you configure your SSEGs, determining for yourself how you will create these groupings. Document your thinking and decision-making in a written record and be prepared to defend your choices with the OFCCP. We recommend making SSEGs synonymous with Job Title. The OFCCP and some other consultants do not agree with that suggestion. They believe job titles should be combined into SSEGs based on the similarities of compensation.
2. If applying the "2% - 30 & 3" test yourself leads to reason for concern, perform a regression analysis on data for those SSEGs that are showing potential problems.
3. Monitor any compensation treatment for individuals in any questionable group. It might not hurt to review every compensation decision made for people in these groups. One way to ensure compensation decisions are reviewed properly is to raise the approval authority needed to make compensation changes in SSEGs where problems have been detected.

It’s ultimately better to perform a bit of self-enforcement than to have to live with the worst-case scenario of a court determining how much your employees will be paid.

Good luck with your program.

Copyright, 2006 by William H. Truesdell, SPHR
Mr. Truesdell is president of The Management Advantage, Inc. in Walnut Creek, California. He can be reached at billt@hrwebstore.com.