Uniform Guidelines on Employee Selection Procedures

What the guidelines cover

UGESP was adopted jointly in 1978 by the EEOC, the Department of Labor, the Department of Justice and the Civil Service Commission (now the Office of Personnel Management), and it is codified for the EEOC at 29 CFR Part 1607. It applies to any selection procedure used to make employment decisions: written tests, pre-employment assessments, interviews, application screening rules, minimum requirements such as education, and the scoring logic of automated tools.

The guidelines do not ban any type of tool. They set out what an employer should do if a tool screens out one group at a substantially higher rate than another, and they are the reference federal agencies and courts often use when judging selection practices under Title VII.

Adverse impact and the four-fifths rule

UGESP introduced the four-fifths rule as a practical rule of thumb: a selection rate for any race, sex or ethnic group that is less than 80% of the rate for the group with the highest rate is generally regarded as evidence of adverse impact. If 60% of male applicants pass a screening step and 42% of female applicants do, the ratio is 0.70, below the threshold.

The guidelines also say smaller differences can matter when they are statistically and practically significant, and larger ones may not when sample sizes are small. In high-volume hiring, with thousands of applicants, statistical significance shows up quickly, so monitoring by stage is worth doing routinely.

Validation

When a procedure has adverse impact, the employer should be able to show it is valid for the job. UGESP describes three types of evidence, covered in more depth under assessment validity:

• Content validity: the procedure samples important parts of the job, like a typing test for a data entry role.
• Criterion-related validity: scores statistically predict job performance or another important outcome.
• Construct validity: the procedure measures a trait shown to matter for the job.

Employers are also expected to consider alternative procedures that are about as valid with less adverse impact. Validation needs a job analysis and documentation; vendor claims alone are not enough.

Record-keeping

UGESP expects employers to keep information on the impact of their selection procedures by race, sex and ethnic group, broken down by job. Federal contractors do this through the applicant flow log. Rules on AI tools are evolving at the state and local level, so check the requirements where you hire.