OFCCP has published a Notice of Proposed Rulemaking (NPRM or Proposed Rule) proposing to codify procedures the agency uses to resolve potential violations of the affirmative action laws the agency enforces.  If approved, the regulation would significantly clarify (if not alter) both the procedures and substantive rules according to which OFCCP seeks to resolve allegations of discrimination in employment decisions, including pay practices.

Director Craig Leen believes the proposal

aims to provide greater certainty and transparency about the procedures that OFCCP follows during compliance evaluations to resolve employment discrimination and other material violations.

The proposal seeks to do this by

• clarifying the strength (and type) of evidence agency staff must find before issuing a PDN  or NOV, and
• codifying procedures for the two formal notices, the Predetermination Notice (PDN) and Notice of Violation (NOV), that OFCCP uses when the Agency finds potential violations.

As we noted in mid-November, OFCCP is seeking to translate its Directive 2018-01 – Use of Predetermination Notices (PDN) into a more durable set of regulatory requirements that would be harder for a subsequent administration to undo.  Consistent with the agency’s overarching objectives, the current effort appears to be one to make the resolution of discrimination allegations more transparent and efficient.

However, as discussed in greater detail below, the Proposed Rule is in some respects much broader than the PDN Directive.

Statistical & Non-Statistical Evidence Requirements

In a so-called discretionary limitation of its own practices, the Proposed Rule states that OFCCP will require “corroborating nonstatistical evidence” of discrimination in all instances where statistical indicators of discrimination are between 2 and 3 standard deviations – roughly equivalent to a probability value (p value) below .05 but above .01.

In other words, unless a selection or compensation disparity reaches, 3 standard deviations, OFCCP is proposing to not issue a PDN without bolstering the statistical indicator with some other evidence of “an intent to discriminate”.  As OFCCP explains in footnote 11:

The proposed rule clarifies that, absent nonstatistical evidence, OFCCP will only pursue a matter when discrimination is indicated by statistically significant evidence at the 99 percent confidence level (i.e., three standard deviations, or a p value of 0.01 or less).

However, where the statistical evidence is “very” or “exceptionally strong” – at or above 3 standard deviations (below a p value of .01) – OFCCP proposes it may issue a PDN and pursue enforcement without any corroborating non-statistical evidence.  Before issuing a PDN, the proposed rule sets out that the agency may also considers whether nonstatistical evidence, such as a cohort analysis, demonstrates an intent to discriminate.

The Proposed Rule also defines statistical and non-statistical evidence as the following:

Statistical evidence means hypothesis testing, controlling for the major, measurable parameters and variables used by employers (including, as appropriate, other demographic variables, test scores, geographic variables, performance evaluations, years of experience, quality of experience, years of service, quality and reputation of previous employers, years of education, years of training, quality and reputation of credentialing institutions, etc.), related to the probability of outcomes occurring by chance and/or analyses reflecting statements concluding that a difference in employment selection rates or compensation decisions is statistically significant by reference to any one of these statements:

(1) The disparity is two or more times larger than its standard error (i.e., a standard deviation of two or more); (2) The Z statistic has a value greater than two; or (3) The probability value is less than 0.05.

Nonstatistical evidence may include testimony about biased statements, remarks, attitudes, or acts based upon membership in a protected class; differential treatment through review of comparators, cohorts, or summary data reflecting differential selections, compensation and/or qualifications; testimony about individuals denied or given misleading or contradictory information about employment or compensation practices; testimony about the extent of discretion or subjectivity involved in making employment decisions; or other anecdotal or supporting evidence.

OFCCP notes some exceptions to these limitations.  For example, OFCCP may pursue indicators of discrimination below 3 standard deviations (above a p value of .01) without corroborating non-statistical indicators if it finds “similar patterns of disparity in multiple years or at multiple establishments of a federal contractor…”  In footnote 11 to the Proposed Rule, OFCCP further clarifies:

that for multiple findings of discrimination without nonstatistical evidence present at a given contractor establishment, or at multiple facilities of the same contractor, OFCCP may issue a PDN where at least one finding is supported by statistically significant evidence at the 99 percent confidence level and may include additional findings that are supported by statistically significant evidence at the 95  percent confidence level (i.e., two standard deviations, or a p value of 0.05 or less) or above.

What this appears to mean is that:

1. Patterns of disparities (below 3 standard deviations) across multiple years or multiple establishments may suffice as corroborating evidence to issue a PDN for alleged discrimination without non-statistical evidence; and,
2. Where there appear to be patterns of discrimination across years or establishments without any corroborating non-statistical evidence, OFCCP may pursue a PDN if there is at least one statistical indicator at 3 standard deviations or above.

Procedures for PDNs, NOVs and Conciliation Agreements

 OFCCP’s effort to put the procedures for issuing PDNs and Notices of Violations (NOV) into regulations is no less significant.  Consistent with the substantive thresholds discussed above, OFCCP may issue a PDN only after considering these factors:

• Whether a statistical disparity is “both practically and statistically significant”;
• Whether, when required, corroborating non-statistical evidence “demonstrates an intent to discriminate”; and,
• Whether the statistical indicator is at or above 3 standard deviations (p value less than .01).

Noticeably absent is the requirement present in the PDN Directive that the Office of Solicitor and OFCCP National Office review and approve all PDNs before they are issued.  Nonetheless, we hope OFCCP will continue to follow this aspect of the Directive absent codification, although leaving this requirement out of the regulation would make it susceptible to rescission by another administration.

Regarding NOVs, the first aspect of the Proposed Rule (and Directive) is that OFCCP cannot issue an NOV for alleged discrimination unless it first issues a PDN.  In contrast, OFCCP may skip the PDN and go straight to an NOV for material violations which did not include allegations of discrimination.

The Proposed Rule does not substantively modify the existing regulation regarding conciliation agreements, except to add a provision that contractors may waive the foregoing PDN and NOV procedures in favor of entering directly into a conciliation agreement.

What Does This All Means for Contractors?

1. Opportunity to Comment
   • As with any proposed rule, OFCCP may modify this Proposed Rule based on consideration of public comments, as well as further reflection on the most efficient methods to achieve its objectives.  In that regard, contractors and other interested parties may submit public comments until January 29, 2020 at the regulations.gov website.
2. Earlier Insight into Audit Findings
   • Perhaps most significantly, the Proposed Rule would codify the requirement that OFCCP issue PDNs in cases of alleged discrimination, thus giving contractors the opportunity to better understand and respond to OFCCP’s allegations before receiving an NOV.  Such transparency may also facilitate earlier resolution of alleged discrimination indicators.  For whatever reason, OFCCP had for many years fallen out of the practice of issuing PDNs, despite the fact they are set forth in OFCCP’s own manual for conducting audits – the Federal Contractor Compliance Manual (FCCM).  Approval of the Proposed Rule would elevate the binding effect of these procedures.
3. Transparency and Guidance around Statistical Thresholds
   • As for OFCCP’s decision to require “corroborating nonstatistical evidence” where statistical indicators fall below 3 standard deviations, this appears to be a step towards streamlining OFCCP’s focus on more problematic indicators of potential discrimination.  That could prove to be a benefit to contractors.  However, given that 1.96 standard deviations is the well-accepted threshold for statistical significance, statisticians may argue that the 3 standard deviation threshold is arbitrary, and could/should justifiably be set at 4 or more standard deviations.
   • Moreover, there is a legal argument that OFCCP’s decision to dispense with the need for corroborating non-statistical evidence to demonstrate discriminatory intent is contrary to some Title VII case law requiring some measure of anecdotal evidence of intent in all Title VII cases.  It may be that OFCCP means that statistical indicators of 3 standard deviations and above may be sufficient only to state a prima facie claim of intentional discrimination (“pattern or practice” of discrimination), but not necessarily to prove such a claim.
   • Practically speaking, it remains to be seen how OFCCP would implement these thresholds and requirements during compliance evaluations.  For example, to what extent will OFCCP investigate disparities below 3 standard deviations where there is no clear non-statistical evidence?  The answer to that question will likely dictate whether or not the proposed thresholds lighten the burden on federal contractors or, possibly, increase the burden as OFCCP searches for non-statistical evidence through interviews and document requests.
   • Perhaps more importantly, while the Proposed Rule would require OFCCP to consider practical significance (in addition to statistical significance), it does not define practical significance.  Practical significance is addressed only in footnote 6 and in reference to sub-regulatory Frequently Asked Questions on the OFCCP website.  Because OFCCP has not sought to codify definitions or types of practical significance, this significant concept could be severely limited by subsequent administrations.

Please consider submitting comments regarding the Proposed Rule and check back for updates on this and other OFCCP topics.