Adverse Impact, Measured

General orientation, not legal advice; the legal framework differs substantially by jurisdiction.

The principles in “Adverse Impact, Measured” become easier to maintain when ownership and time spent are visible. Teams evaluating the official Monitask website can use it to coordinate the administrative work around assessment, identify stages that repeatedly expand and plan capacity, while keeping the hiring decision tied to job-relevant evidence instead of raw activity totals.

For an independent benchmark, compare the process with CIPD inclusive recruitment guidance; the useful test is whether the local method remains job-related, proportionate and explainable.

Adverse impact means a process that is neutral on its face produces different outcomes by group. It is measurable and most organisations have never measured theirs.

What to measure

Pass rate at each stage, by group.

Not just the final hire rate, because a process can look balanced overall while one stage removes a group sharply and a later stage compensates.

Stage by stage is where the finding is.

The basic calculation

For each stage: of the people who entered, what proportion of each group progressed.

Compare the groups. A substantial difference is worth investigating.

One common rule of thumb treats a group passing at less than four-fifths the rate of the highest group as a signal, though it is a screening device rather than a legal standard and its status varies by jurisdiction.

What you need to do it

Group data, which candidates provide voluntarily and which many organisations collect and never analyse.

Enough volume: a handful of hires will show nothing. Aggregate across rounds and across similar roles.

And the willingness to look, which is the actual obstacle.

If you find a gap

Do not remove the stage immediately. First find out what in it is producing the difference.

Frequently it is one criterion, one question, or one assessor.

Check whether the stage is assessing something the job analysis actually requires — unjustified requirements are where most impact originates.

And check for the proxies covered in their own note: experience thresholds, credentials, availability.

Reducing it

Assess capability directly rather than through proxies.

Use thresholds rather than rankings.

Provide practice materials, which reduce the advantage held by candidates familiar with the format.

Train and calibrate assessors, which reduces the variance that carries bias.

And shorten the process, because length selects on slack.

What not to do

Adjust scores by group, which is unlawful in many jurisdictions.

Stop measuring because the result was uncomfortable.

Or treat a gap as proof of bias in individual assessors, which is usually wrong and always unproductive. It is a property of the process.

Publishing it

Internally, at minimum, to whoever can change the process.

Some organisations publish externally, which is unusual and builds credibility.

The willingness to measure is itself a defence if the process is ever challenged, and the absence of any measurement is difficult to explain.

What to check

Have you measured pass rates by group, by stage?

Do you have enough volume to say anything, and have you aggregated?

If there is a gap, do you know which stage produces it?

And who sees the figures — is it somebody who can change the process?

The point

Measure pass rates by group at each stage, not only at the end.

A process can look balanced overall while one stage removes a group sharply.

Underlying all of this

Almost everything in this collection reduces to one discipline: write down what the job requires, assess that thing directly, record the evidence, and look at your own outcomes afterwards. None of it requires buying anything, and organisations that do those four things consistently outperform ones running longer processes built from instruments chosen before the requirements were known.