---
title: "Adverse impact ratio"
description: "The adverse impact ratio is one group's selection rate divided by the selection rate of the group selected most often. A ratio below 0.80 is the threshold at which United States federal enforcement agencies generally treat a selection procedure as showing adverse impact and expect the employer to justify it."
canonical: https://www.assessall.com/guides/glossary/a/adverse-impact-ratio
updated: 2026-09-14
source: AssessAll
---

# Adverse impact ratio

_Also known as: Impact ratio, Selection rate ratio._

The adverse impact ratio is one group's selection rate divided by the selection rate of the group selected most often. A ratio below 0.80 is the threshold at which United States federal enforcement agencies generally treat a selection procedure as showing adverse impact and expect the employer to justify it.

<!-- #how-it-is-computed -->
## How it is computed

Take one stage of a hiring process and one demographic dimension. For each group, divide the number of people who passed that stage by the number who entered it — that is the group's selection rate. Divide every group's rate by the highest rate observed. The lowest resulting figure is the one people mean when they say 'the impact ratio'.

The ratio is computed per stage and per dimension. A process can be clean at the screening stage and fail at the interview stage, and the aggregate over the whole funnel can hide both.

<!-- #what-the-number-does-not-tell-you -->
## What the number does not tell you

A ratio is a point estimate with no sample size attached, so it says nothing about how much evidence sits behind it. Two selections out of five against five out of five gives a ratio of 0.40 and means almost nothing; two hundred out of five hundred against five hundred out of five hundred gives the same 0.40 and means a great deal. Always report the group sizes beside the ratio, and pair it with a significance test.

The ratio is also silent on cause. A flagged stage may reflect the instrument, the sourcing that produced the pool, the order of the stages, or a genuine difference in the applicant population. It is a prompt to investigate, not a finding.

<!-- #the-validity-cost-of-acting-on-it-is-smaller-than-the-field- -->
## The validity cost of acting on it is smaller than the field assumed, and that is a 2023 result

The standard objection to changing a procedure that shows adverse impact is that accuracy will suffer, because the predictor with the largest subgroup differences — cognitive ability — was for twenty-five years held to be the strongest one available. That premise has been re-examined and it does not survive.

Sackett, Zhang, Berry and Lievens report that in their illustrative selection composite, giving cognitive ability a weight of zero reduces validity by 0.05, from 0.61 to 0.56. Put through the Taylor–Russell conversion at a base rate of 50% and one hire in ten, that is a move from a 90.2% success rate among those hired to 87.4% — under three good hires per hundred.

That changes the shape of the conversation rather than settling it. Reweighting is not the only remedy, it is not automatically the right one, and a composite is not a procedure. But the cost of considering it can now be stated, priced and weighed, instead of being used to close the discussion before it starts.

---

Source: https://www.assessall.com/guides/glossary/a/adverse-impact-ratio