Adverse impact testing
Screen your hiring funnel against the EEOC four-fifths rule by protected class.
Adverse impact testing applies the EEOC's four-fifths rule to your hiring funnel: it flags a group whenever its selection rate falls below 80% of the highest-selection-rate group in the same comparison.
This is a statistical screen, not a legal determination. A flagged group means the numbers cross a commonly-used threshold — it does not mean discrimination occurred, and an unflagged result does not mean your process is legally sound. Work with employment counsel to interpret and act on any result.
What it measures
Three steps of the funnel, for each protected-class dimension you collect:
| Step | What it compares |
|---|---|
| Applied → Interviewed | Of everyone who applied, who was invited to an interview |
| Interviewed → Hired | Of everyone interviewed, who was hired |
| Applied → Hired | Of everyone who applied, who was ultimately hired |
Dimensions: gender, race/ethnicity, veteran status, disability status — the same voluntary self-identification data used elsewhere in Compliance & DEI.
"Interviewed" comes from scheduled interviews on the application, not from an application's current status — a rejected application still counts as having reached the interview step if one was scheduled.
Before you begin
- Confirm that your organization has the DEI & Compliance package.
- Collect voluntary demographic responses under a policy reviewed by legal or compliance.
- Confirm that enough candidates responded to support a meaningful comparison — see suppression below.
Read the results
- Open the compliance reporting area in InsightHire.
- Find Adverse impact (four-fifths rule).
- Review each dimension's three steps.
For each group, InsightHire shows:
- Applicants and Selected — the raw counts behind the rate
- Rate — the group's selection rate for that step
- Impact ratio — this group's rate divided by the highest-rate group's rate
A ratio below 0.80 is flagged. The highest-rate group in a comparison always shows a ratio of 1.00 and is never itself flagged.
Understand small-group suppression
Groups with fewer than five applicants at a given step are suppressed (shown as "< 5") and excluded from the comparison entirely — not just hidden, but left out of the highest-rate calculation too, so a tiny group can't distort the comparison for everyone else. A step with fewer than two groups that have enough data shows as "not enough data" rather than a misleading comparison.
Validate before acting
- A flagged result is a starting point for investigation, not a conclusion — look for confounding factors (role type, location, tenure of the requisition) before assuming the cause.
- Small absolute differences in count can swing a ratio dramatically when volumes are low — read the raw applicant/selected counts alongside the ratio, not the ratio alone.
- Have employment counsel review any flagged result before it informs a hiring-process change or a response to a regulator.
Related
- OFCCP export — the data export this testing complements
- EEOC export
- Demographics