Adverse Impact

Adverse impact analysis for RIFs

Q: What is the 4/5ths rule for adverse impact?

The 4/5ths rule (also called the 80% rule) is the EEOC's standard screening test for adverse impact. A protected group has been adversely impacted if its selection rate is less than 4/5ths (80%) of the highest selection rate among any group. For example, if 60% of white employees are retained but only 40% of Black employees are retained, the ratio is 40/60 = 67%, which fails the 80% threshold.

Q: When should you use Fisher's exact test instead of the 4/5ths rule?

Fisher's exact test is a statistical significance test used when sample sizes are small, typically when any group has fewer than 30 members in the at-risk pool. The 4/5ths rule alone is unreliable with small samples because a single termination decision can swing the ratio dramatically. Courts and the EEOC both give weight to Fisher's exact test results.

Q: Does adverse impact analysis apply to age?

Yes. The Age Discrimination in Employment Act (ADEA) protects employees 40 and older. Courts have recognized disparate impact claims under the ADEA (Meacham v. Knolls Atomic Power Laboratory, 2008). Analysis should compare the selection rate for employees 40+ versus employees under 40. Additionally, any RIF involving employees 40+ triggers OWBPA requirements for the waiver and release of ADEA claims.

Q: What should you do if your RIF fails the adverse impact analysis?

A statistical flag does not mean you cannot proceed, but it requires action before execution. First, audit the selection criteria to identify what is driving the disparity. Second, determine whether you can apply a legitimate, non-discriminatory reason to each selection decision. Third, consider whether adjustments can be made to the selection pool that do not compromise your business objective. Fourth, document every step. If you cannot eliminate the disparity, consult employment counsel before proceeding.

Q: What is the correct pool to use for adverse impact analysis in a RIF?

The correct comparison pool is the at-risk population: the group of employees who were considered for the RIF, not the entire workforce. If the RIF is limited to a specific department, job family, or level, your analysis should use that group as the denominator. Using the entire company workforce as the denominator when only a subset was at risk is a common and significant error.

Before you execute a reduction in force, you need to verify that your selection criteria do not disproportionately affect a protected class. This guide explains the EEOC 4/5ths rule, Fisher's exact test, age analysis under the ADEA, and what to do when the numbers flag a problem.

Title VIIADEA4/5ths RuleFisher's Exact

Overview When it applies The at-risk pool 4/5ths rule Fisher's exact Protected classes Age and ADEA If you fail Documentation Common mistakes FAQ

What is adverse impact?

Adverse impact (also called disparate impact) occurs when a facially neutral employment practice disproportionately excludes members of a protected class. In the context of a RIF, it means your selection criteria, even if applied uniformly, result in a significantly higher termination rate for one group than another.

The legal framework comes from two primary sources: Title VII of the Civil Rights Act (covering race, color, religion, sex, and national origin) and the Age Discrimination in Employment Act (ADEA, covering employees 40 and older). The EEOC's Uniform Guidelines on Employee Selection Procedures (1978) establish the 4/5ths rule as the standard screening test.

Critically, adverse impact is a statistical showing. You do not need discriminatory intent for liability to attach. If the numbers indicate a disparity and you cannot justify the selection criteria with a legitimate business necessity, you are exposed.

Run the analysis before you announce

Adverse impact analysis must be completed before the RIF is executed, not after. Once termination letters go out, you have lost the opportunity to restructure the selection pool to eliminate a disparity. Courts have been skeptical of post-hoc adjustments.

When adverse impact analysis applies

There is no statutory threshold that triggers the obligation. As a practical matter, any RIF affecting five or more employees warrants at least a preliminary review. The risk scales with size: a RIF of 50 employees in a department with clear demographic patterns carries substantial exposure.

When to run the analysis

5+ employees

Run a preliminary 4/5ths check by protected class

15+ employees

Full analysis required: Title VII applies at this employer size

20+ employees

ADEA applies: add age (40+/under 40) to your analysis

Any size with small subgroups

Use Fisher's exact test when any group has fewer than 30 employees in the at-risk pool

Defining the at-risk pool

The single most common error in RIF adverse impact analysis is using the wrong denominator. The correct comparison pool is the at-risk population: the specific group of employees who were considered for reduction.

Incorrect

Using the entire company workforce as your denominator when the RIF only affected one department or job level. This dilutes the true selection rates and can mask a real disparity.

Correct

Using only the employees who were actually considered for the RIF, defined by the same criteria you used to make the selection (department, job family, grade level, location, or some combination).

If your RIF was conducted in phases or across multiple business units with different criteria, you may need a separate analysis for each discrete selection event. Combining pools with different selection rationales can obscure disparities.

The 4/5ths (80%) rule

The 4/5ths rule is the EEOC's primary screening test for adverse impact. A protected group shows adverse impact if its selection rate (the rate at which members of that group are retained) is less than 80% of the highest selection rate for any group.

The formula

Selection rate (group A) = retained ÷ at-risk

Adverse impact ratio = rate(A) ÷ rate(highest group)

Flag if ratio < 0.80

Source: EEOC Uniform Guidelines on Employee Selection Procedures, 29 C.F.R. § 1607.4(D)

Worked example

Group	At-risk	Retained	Rate	Ratio
White	50	40	80%	1.00 (ref)
Hispanic	20	14	70%	0.88
Black	15	9	60%	0.75 ⚠

Black employees have a 60% retention rate vs. the 80% highest rate (White). The adverse impact ratio is 60/80 = 0.75, which is below the 0.80 threshold. This flags a potential adverse impact issue that must be addressed before executing the RIF.

Important limitation

The 4/5ths rule is a screening test, not a legal standard. Courts do not consider it dispositive on its own. A ratio above 0.80 does not guarantee no liability, and a ratio below 0.80 does not automatically establish discrimination. Statistical significance (Fisher's exact or standard deviation analysis) matters equally.

Fisher's exact test

Fisher's exact test measures whether a disparity between groups is statistically significant or could have occurred by chance. It is most important when sample sizes are small, because the 4/5ths rule is unreliable at small n: a single termination decision can swing the ratio from passing to failing.

When to use each test

Any group has fewer than 5 members

Fisher's exact test only. The 4/5ths rule is meaningless at this size.

Any group has 5 to 29 members

Run both tests. Weight Fisher's exact more heavily given the small sample

All groups have 30+ members

4/5ths rule is reliable. Supplement with Fisher's exact for completeness

Fisher's exact test produces a p-value. The conventional threshold is p < 0.05 (5% probability the result occurred by chance). Some courts and practitioners use p < 0.10 for employment discrimination cases given the policy stakes. A p-value above 0.05 with a failing 4/5ths ratio is still worth documenting and explaining.

Protected classes to analyze

Federal law requires analysis across five primary protected characteristics. Many states add additional protected classes. Run a separate 4/5ths calculation for each.

Race and color

Title VII, 42 U.S.C. § 2000e

Analyze each racial category separately where sample size permits. Common EEO categories: White, Black/African American, Hispanic/Latino, Asian, American Indian/Alaska Native, Native Hawaiian/Pacific Islander, Two or more races.

Sex / gender

Title VII; Bostock v. Clayton County (2020)

Compare male vs. female retention rates. Post-Bostock, gender identity and sexual orientation are also protected. If you collect this data and have sufficient sample sizes, include in your analysis.

National origin

Title VII, 42 U.S.C. § 2000e

Analyze if you have employees from distinct national origin groups with meaningful representation in the at-risk pool. Often correlates with race/ethnicity analysis.

Age (40 and older)

ADEA, 29 U.S.C. § 623

Compare employees 40+ versus under 40. The ADEA also permits more granular analysis by age subgroups (e.g., 50+, 60+). Triggers separate OWBPA disclosure obligations for waivers.

Disability

ADA, 42 U.S.C. § 12112

Disparate impact claims under the ADA are available. If you track disability status, include it. Selection criteria that tend to screen out employees with disabilities (e.g., physical performance metrics) warrant particular scrutiny.

State-specific classes

Varies by state

Many states protect additional characteristics: pregnancy/caregiver status (CA, NY), sexual orientation (many states), military status, credit history, criminal history. Check your state laws.

Age analysis and ADEA obligations

Age deserves its own section because it carries obligations beyond the statistical analysis. Any RIF that includes employees aged 40 or older and involves a waiver of ADEA claims as part of a severance agreement triggers the Older Workers Benefit Protection Act (OWBPA).

OWBPA group waiver requirements

When two or more employees are terminated as part of the same program and asked to waive ADEA claims, each employee must receive a disclosure listing every job title and age of all employees in the decisional unit who were selected and not selected for the program.

45-day consideration period (not 21 days) for group terminations
7-day revocation period after signing
Written waiver that specifically refers to the ADEA
Disclosure of job titles and ages for all employees in the decisional unit
Advice in writing to consult an attorney

A defective OWBPA waiver is void

Unlike other employment waivers, an OWBPA waiver that fails to meet all statutory requirements is void, not voidable. The employee keeps the severance and can still sue. There is no way to cure a defective OWBPA waiver after the fact. Get the disclosure right before severance letters go out.

For the adverse impact analysis itself, compare the selection rate of employees 40 and older versus employees under 40 using the same 4/5ths framework. If your RIF disproportionately affects older workers and you cannot demonstrate a legitimate, age-neutral business reason, the OWBPA disclosure will not protect you from a disparate impact claim.

What to do when the analysis flags a disparity

A statistical flag is not a stop sign. It is a mandatory pause point. The goal is to understand what is driving the disparity and determine whether it can be addressed before execution.

Audit the selection criteria

Review each individual selection decision in the flagged group. Is there a legitimate, non-discriminatory, documented reason for each person selected? Common legitimate criteria include: performance ratings, skill redundancy, role elimination, and seniority (with caution on ADEA). Criteria that are vague, undocumented, or applied inconsistently are red flags.

Test the business necessity defense

If the criteria that caused the disparity are genuinely necessary to the business objective, document that necessity thoroughly. The business necessity defense requires showing both that the criterion is job-related and consistent with business necessity, and that no equally effective alternative practice with less adverse impact exists.

Explore restructuring the pool

Ask whether the RIF objective can be achieved with a different selection pool composition that produces less disparity. This must be done on legitimate business grounds. You cannot simply add protected-class members back to hit a statistical target. But if there are legitimately equivalent employees who could be included or excluded, this is worth exploring.

Consult employment counsel

If you cannot eliminate the disparity through legitimate means, consult employment counsel before proceeding. This conversation should happen under privilege. Counsel can help assess whether the business necessity defense would hold, whether the statistical disparity rises to the level of actionable risk, and how to structure the documentation.

Document everything

Your analysis, the criteria review, the business necessity rationale, and any adjustments made must all be documented contemporaneously. Post-hoc explanations receive far less weight in litigation than decisions that were clearly made and recorded before execution.

Documentation checklist

In litigation, the document trail often matters more than the underlying numbers. Every step of the adverse impact analysis process should produce a contemporaneous record.

Pre-execution documentation

Definition of the at-risk pool, including who was included and why

Selection criteria applied to the pool, documented before selection began

Per-employee selection rationale for each person selected

4/5ths rule analysis results for each protected class

Fisher's exact test results where sample sizes warrant

Age cohort analysis (40+ vs. under 40) if any affected employees are 40+

Review by employment counsel or HR leadership before execution

Any adjustments made to the selection pool and the business rationale for them

OWBPA decisional unit disclosure (if ADEA waivers will be sought)

Common mistakes

Running the analysis after announcing the RIF

Once employees receive termination letters, you have lost the ability to restructure the selection to fix a disparity. The analysis must happen during the planning phase, while adjustments are still possible.

Using the entire company as the at-risk pool

If the RIF only affected one department, comparing against company-wide demographics dilutes the selection rates and can hide a real disparity within the affected group. The pool must match the actual selection universe.

Skipping age analysis

Teams that routinely run race and gender analysis sometimes skip age entirely. The ADEA applies to any employer with 20 or more employees, and older workers are frequently overrepresented in reductions that target higher-cost or senior employees.

Not using Fisher's exact with small groups

When any subgroup has fewer than 30 employees in the at-risk pool, the 4/5ths rule alone is unreliable. A single termination can swing the ratio dramatically. Fisher's exact test measures whether the result is statistically meaningful.

Treating a passing analysis as complete protection

Passing the 4/5ths rule does not immunize you from an individual disparate treatment claim, an OWBPA defect, or a failure to accommodate under the ADA. The adverse impact analysis is one layer of a multi-layer compliance review.

Applying selection criteria inconsistently

If your stated criteria are 'performance and role redundancy' but some employees were selected based on manager discretion, the inconsistency becomes a problem in litigation even if the overall numbers pass the 4/5ths test.

Free download

Get the full RIF compliance checklist

66 steps across 9 phases, from selection criteria through post-RIF close-out. Formatted as an Excel workbook your team can track in real time.

Frequently asked questions

What is adverse impact in a RIF?

Adverse impact (also called disparate impact) occurs when a facially neutral selection criterion disproportionately excludes members of a protected class. In a RIF, it means the selection rate for a protected group (by race, gender, age, etc.) is significantly lower than the selection rate for other groups, creating potential liability under Title VII, the ADEA, or other anti-discrimination statutes.

What is the 4/5ths rule for adverse impact?

The 4/5ths rule (also called the 80% rule) is the EEOC's standard screening test for adverse impact. A protected group has been adversely impacted if its selection rate is less than 4/5ths (80%) of the highest selection rate among any group. For example, if 60% of white employees are retained but only 40% of Black employees are retained, the ratio is 40/60 = 67%, which fails the 80% threshold.

When should you use Fisher's exact test instead of the 4/5ths rule?

Fisher's exact test is a statistical significance test used when sample sizes are small, typically when any group has fewer than 30 members in the at-risk pool. The 4/5ths rule alone is unreliable with small samples because a single termination decision can swing the ratio dramatically. Courts and the EEOC both give weight to Fisher's exact test results.

Does adverse impact analysis apply to age?

Yes. The Age Discrimination in Employment Act (ADEA) protects employees 40 and older. Courts have recognized disparate impact claims under the ADEA (Meacham v. Knolls Atomic Power Laboratory, 2008). Analysis should compare the selection rate for employees 40+ versus employees under 40. Additionally, any RIF involving employees 40+ triggers OWBPA requirements for the waiver and release of ADEA claims.

What should you do if your RIF fails the adverse impact analysis?

A statistical flag does not mean you cannot proceed, but it requires action before execution. First, audit the selection criteria to identify what is driving the disparity. Second, determine whether you can apply a legitimate, non-discriminatory reason to each selection decision. Third, consider whether adjustments can be made to the selection pool that do not compromise your business objective. Fourth, document every step. If you cannot eliminate the disparity, consult employment counsel before proceeding.

What is the correct pool to use for adverse impact analysis in a RIF?

The correct comparison pool is the at-risk population: the group of employees who were considered for the RIF, not the entire workforce. If the RIF is limited to a specific department, job family, or level, your analysis should use that group as the denominator. Using the entire company workforce as the denominator when only a subset was at risk is a common and significant error.

People Plan

Adverse impact analysis built into your RIF workflow

People Plan runs the 4/5ths rule and Fisher's exact test across every protected class as you build your selection criteria, before you lock in the list. If a disparity is detected, it flags the issue in context so you can investigate and adjust while it is still possible to do so.