Skip to content

Adverse impact

Adverse impact is a substantially different rate of selection that works to the disadvantage of a group defined by a protected characteristic. It is a property of outcomes rather than of intent, so a procedure applied identically to everyone can still produce it and still require justification.

A selection rate below four-fifths of the highest group's rate is generally treated as evidence of adverse impact.

Why it matters when the plan changes

Because it is measured on outcomes, adverse impact can be produced by a procedure nobody designed to discriminate and everybody applied consistently. The Uniform Guidelines on Employee Selection Procedures set a rule of thumb for spotting this, a selection rate below four-fifths of the highest group's rate, precisely because intent cannot be read off a process. That is what makes it a testing question rather than an intent question: the only way to know whether a procedure produces it is to measure the rates by group, which most organisations never do until a dispute forces the calculation.

The tension is that a procedure showing adverse impact is not automatically prohibited. It has to be justified by evidence that it predicts what the job actually requires, the predictive validity question, which returns the matter to validity for the specific use rather than the procedure in general. Skipping either half, the fairness testing or the validity evidence, leaves a decision that cannot be defended on either ground.

In practice

An assessment stage is added to a hiring process and applied identically to every applicant. Nobody checks the pass rates by group for two years. When they are checked, one group passes at just over half the rate of the highest, and two years of decisions rest on a stage that was never tested for the effect it was having.

Evidence

What it cannot tell you

Adverse impact identifies a disparity in selection rates; it does not explain what causes the disparity or whether the procedure is job related. A finding says a group is selected less often than the four-fifths threshold suggests it should be, not which stage produced the effect, why it occurred, or whether the difference reflects the procedure or the applicant pool it drew from.

Questions

A rule of thumb set out in the Uniform Guidelines on Employee Selection Procedures (1978): if one group is selected at less than four-fifths of the rate of the group with the highest rate, that is generally treated as evidence of adverse impact. It is a screening threshold, not a legal verdict.

No. It is measured on outcomes, so a procedure designed without any discriminatory purpose and applied consistently to everyone can still fall below the four-fifths threshold set out in the 1978 Uniform Guidelines on Employee Selection Procedures. That is why it has to be tested for, not reasoned about from how the procedure was built.

Not automatically. It means the procedure has to be justified by evidence that it predicts what the role actually requires, and that less discriminatory alternatives were considered. Without that evidence, the finding stands and the decisions resting on the procedure are exposed.

Those protected in the jurisdictions where the procedure is used, which varies. Fairness work on the Atlas instrument is planned across gender, age, role and country. Examining only the categories that are convenient to obtain is a common way to produce a reassuring result.

Before high-stakes use, and periodically afterwards, because the applicant population changes even when the procedure does not. Testing only at launch answers a question about the people who applied that year, which may not describe the people applying three years later.