Despite resistance from the employer community, EEO-1 “Component 2” compensation reporting requirements will most likely be upheld by federal courts. As employers scramble to assemble their reports, many are also concerned with the potential risk exposure these reports may introduce. First and foremost, how will the OFCCP and EEOC use this information in their pay equity investigations?
Strategically, the OFCCP has largely shared their two-part game plan:
To measure internal pay equity, the OFCCP is clear1:
The mechanics of these statistical tests are not important for the average Human Resources (HR) professional. What is important, however, is for HR professionals to conceptually understand what these statistical tests are evaluating.
In concept, these analyses can determine whether there is a bias in the distribution of men and women along the 12 pay bands (ranging from lowest 1 = less than $19,239, to highest 12 = $208,000 or greater). For example, if women tend to dominate lower-paying salary bands while men are in the higher-paying salary bands, then women are being paid less than men. The Mann-Whitney U and Kruskal-Wallis can determine if there are differences in the distribution of gender or race across the 12 bands, respectively.
To help make sense of all of this, here is a sample report by gender for EEO Category professional:
Per the EEOC reporting requirements, the counts of male and female employees were reported by the pay bands. The statistical analyses will examine the actual counts. In concept, the analyses are looking for patterns in the distribution of males and females across the pay bands. In this example, note that females dominate the lower 1-5 pay bands, while men dominate the higher 7-12 pay bands. The statistical methods will identify this group difference and estimate its statistical significance.
The OFCCP is less clear on their strategy to measure external pay equity with the EEO-1 pay data. This is understandable as Title VII is primarily concerned with internal pay equity, and the law is less concerned with compensation philosophy where a company may choose to pay below-labor market standards as long as everyone is impacted equally.
Contractors need to be aware that there is an analytical approach that could assist the OFCCP and EEOC to identify employers with potential pay equity problems. Consider the following analysis example of EEO Category 1.1:
To simplify this analytical method, pay bands 1-6 are collapsed into “1-6” and pay bands 7-12 are collapsed into “7-12”. Note the interesting pattern when comparing internal composition (%) against external composition (%):
These results draw three important conclusions:
In effect, women do worse than expected among lower pay bands while men do better than expected among higher pay bands. In and of itself, this pattern is sufficient to raise concerns for the OFCCP and EEOC. Additionally, if this pattern is observed across multiple EEO categories, concerns for systemic pay discrimination will naturally be raised.
One of the most compelling arguments against EEO-1 Component 2 reporting has been the concern that no meaningful conclusions can be drawn from analysis of broad EEO categories; the results will primarily be a mix of false positives and false negatives. Analyses that do not make proper apples-to-apples comparisons among similarly situated employees will most likely suffer this very likely consequence. Having said that, it is important to keep in mind that the OFCCP and EEOC will most likely use the EEO-1 pay data to flag employers who may have pay equity problems rather than to establish prima-facia claims of pay disparity. The methods discussed in this paper serve this purpose well.
By measuring internal and external pay equity with a lens for patterns, the OFCCP and EEOC can more efficiently identify employers who may be considered for more thorough and complex Title VII-oriented pay equity investigations. Employers should not dismiss the very likely danger that lurks once their EEO-1 Component 2 pay data is reported.
1. Federal Register, Vol 80 (20), p. 5118, footnote 47
