1O03 Statistical Analysis Summary

When Investigators identify evidence of disparity against members of a protected group, they must request additional data from the contractor for further analysis. If the analysis confirms that the statistical disparities exist and further investigation is needed, Investigators must describe the problem in CMS and indicate that the matter needs further investigation on-site to determine whether there is discrimination.

a. Contractor Adverse Impact Determinations (Contractors with 100 or More Employees).

• Maintenance of Records. As noted earlier in the discussion of the recordkeeping requirements in 1C02, the UGESP requires contractors with more than 100 employees to maintain specific records by job title that are sufficient to disclose whether their selection procedures have an adverse impact on the employment opportunities of each sex, and each race or ethnic group, or both. The racial and ethnic groups are defined by 41 CFR 60-3.4B as Black, Hispanic, Asian, American Indian, and white other than Hispanic. However, OFCCP also permits contractors to keep their records concerning impact by using the racial and ethnic categories on the EEO-1 report.
• Analysis of Impact. The UGESP also requires that contractors with more than 100 employees analyze this data annually to determine whether the total selection process for each job is having any adverse impact. These determinations are required by sex and for each race and ethnic group that constitutes 2% or more of the labor force in the relevant labor area, or 2% or more of the applicable workforce for jobs filled internally.

b. Requesting Contractor Determinations. When an Investigator identifies a job group with evidence of adverse impact and the contractor employs more than 100 people, the Investigator will ask the contractor to furnish the adverse impact determinations prepared during the review period as a part of its in-depth analyses for the job titles that fall within the job group. This assumes, of course, that the contractor has not already provided this information. For example, if the Investigator finds an adverse IRA for female hires into a Professionals job group, the contractor should submit its adverse impact determinations for the hiring of women in each title within the Professionals group. This action will assist the Investigator in determining whether the IRA and other statistical analyses need to be refined before investigating further. An Investigator may also request the contractor’s adverse impact analyses in other areas. For example, a review of the workforce analysis showed a concentration or underrepresentation suggesting a potential placement problem. The contractor provided personnel activity data by job group and it does not show placements into the titles of concern. In response, an Investigator must request the contractor’s adverse impact analyses for hires, promotions and transfers into the job titles at issue. When appropriate, the Investigator may also ask to review the contractor’s adverse impact analyses and/or may ask the contractor to identify those jobs where its analyses showed an adverse impact.

c. Need for Information about the Selection Process. When the hiring, promotion or termination analysis for a job group or job title indicates that further investigation is needed, the Investigator must ask the contractor to provide a description of how the employment selections are made for positions in the job group or job title at issue. The description should include the steps in the process. The contractor should provide the following data for each selection step: the decision- makers, the criteria used, a description of how the criteria are used, and the records maintained. In addition, when a job group is under investigation, the Investigator should ask whether the selection process, as a whole or individual steps, applies to all of the job titles in the job group, and whether the process was the same throughout the entire duration of the review period. It is also helpful for the Investigator to know whether the process applies to other job groups or titles in addition to the group or title at issue. Steps may include, for example, review of application forms by the human resources representative, written tests, formal or informal interviews, physical examinations and on-the-job tests. The Investigator will plan to verify the contractor’s statements through review of records, interviews with applicants and employees, and, if possible, observation of the process by which applicants are screened and selected.

d. Multi-Component Selection Processes - Contractor Obligations.

• Adverse Impact in Total Selection Process. When the contractor submits its adverse impact analyses for desk audit and if the analyses shows an adverse impact in the total selection process for a job group or job title, the UGESP requires the contractor to evaluate the individual components of the total selection process for an adverse impact. Therefore, in a multi-step and/or multi-criterion selection process with an adverse impact, the Investigator must plan to request the contractor’s records showing at what step(s) and/or by what criteria members of the nonfavored race, ethnic group or sex the contractor is disproportionately screening out, as well as any validity studies the contractor conducted of its selection procedures.
• No Adverse Impact in Total Selection Process. Generally, if the total selection process for a job does not have an adverse impact, a contractor will not usually be expected to evaluate the individual components for an adverse impact or to validate those individual components, and enforcement action will not usually be warranted based on a component.

There are, however, several exceptions: when enforcement appears warranted and with individual complaints of discrimination.

• Exception 1: When Enforcement May Be Warranted. In some circumstances, even though there is no adverse impact in the total selection process, further investigation and, possibly, enforcement action may nonetheless be appropriate when an individual component has an adverse impact (e.g., height and weight requirements or criminal background checks), and its use is not justified as job-related and consistent with business necessity.
• Exception 2: Individual Complaints of Discrimination. The “bottom line” standard – that a contractor need not evaluate the individual components of a selection process if the process as a whole does not result in adverse impact – does not preclude the investigation of complaints alleging race, ethnicity or sex discrimination caused by a component of a selection process. Investigators and contractors can find more information at https://www.eeoc.gov/laws/guidance/questions-and-answers-clarify-and-provide-common-interpretation-uniform-guidelines, “Adoption of Questions and Answers to Clarify and Provide a Common Interpretation of the Uniform Guidelines on Employee Selection Procedures,” Questions and Answers Nos. 25 and 26.

e. Analysis of Selection Process Tools, including Artificial Intelligence. The Investigator must examine any measure, combination of measures, or procedure that a contractor uses to make employment decisions, including decision-making tools developed using AI. Examples of these types of tools or procedures include resume scanners, virtual assistants, and video interviewing software that evaluate candidates based on their facial expressions and speech patterns. If a contractor’s use of an AI-based selection procedure has adverse impact, the contractor is required to validate the selection procedure using an appropriate validation strategy that demonstrates the procedure is job-related and consistent with business necessity. Like other selection procedures, AI systems must comply with OFCCP’s laws and regulations on nondiscrimination in employment. 41 CFR part 60-3, Uniform Guidelines of Employee Selection Procedures (UGESP) includes additional information about validation. If the hiring or other selection analyses indicate there are findings of potential discrimination, Investigators should request more specific information on the related policies and practices to investigate the indicators and conduct an on-site, if necessary.