5 ways to reduce bias in recruitment
Ben Schwencke walks through the five practical changes that do most to take bias out of a hiring process.
FAIRER HIRING FOR EVERY CANDIDATE
Diversity and inclusion
It's important to build a hiring process that gives every candidate a genuine chance. We should aim to reduce bias and make assessments accessible to all applicants.
What it means
Diversity in recruitment means your process gives candidates from all backgrounds a genuine, equal chance of success. Inclusion means the process itself doesn't put unnecessary obstacles in anyone's way.
Why it matters
Diverse teams draw on a wider range of perspectives and reach better decisions. Inclusive processes widen your talent pool rather than filtering out capable people for reasons that have nothing to do with the job.
Equality Act 2010
In the UK, the Equality Act 2010 makes this a legal obligation: employers must not discriminate against candidates with protected characteristics, and must make reasonable adjustments for disabled applicants.
Many common hiring methods (CVs, interviews) are where bias thrives
Bias is hardly ever deliberate, it's unconsciously favouring familiar universities or candidates who are similar to yourself. Structured assessments remove the opportunity for that bias to operate. This is why psychometric tests consistently outperform CV sifting and other methods on fairness.
1. How to improve fairness and equality in online testing
Online testing only improves fairness if it's set up properly. From practice materials, pass marks, to differential item functioning analysis.
Improve fairness in online testing2. How to use psychometric tests to improve diversity
Aptitude tests measure cognitive ability directly, letting you drop socioeconomically loaded requirements like UCAS points, degree classification and institution.
Improve diversity3. How to reduce bias in recruitment
From managing assessors' unconscious bias, designing assessments to reduce adverse impact, meeting technical accessibility standards and more.
5 steps to reduce biasThe science of fair assessment
Not all assessments are equally fair or valid. Assessments need to be tested and monitored against things like adverse impact and validity.
Adverse impact
Adverse impact occurs when a selection method disproportionately screens out candidates from a particular group, even without any discriminatory intent.
More on adverse impactReliability and validity
Reliability (does the test measure consistently?) and validity (does it measure what actually predicts job performance?) are the foundations that fairness is built on.
Reliability and validityReasonable adjustments
Fairness doesn't always mean treating everyone identically. For disabled / neurodivergent candidates identical conditions produce unequal opportunity.
Why you should allow extra time
Neurodiversity
Neurodiversity describes the natural variation in how people think, process information, and learn. An estimated one in seven people in the UK is neurodivergent, which means every sizeable candidate pool includes neurodivergent applicants, whether they've disclosed or not.
Find out how to improve accessibility for the following conditions:
Improve accessibility for neurodivergent candidates
Using psychometric tests to support diversity in hiring
Why the assessments employers worry about most are usually fairer than the CV sift and unstructured interview they'd replace.
Diversity and inclusion FAQs
Poorly made tests might. But well-made tests are fairer than what you'd use instead. CV sifting and unstructured interviews are the most bias-prone selection methods available, and nobody measures their impact. A well-built assessment is checked for adverse impact before it's published and monitored afterwards, and individual questions are screened with differential item functioning analysis. Personality questionnaires in particular show no adverse impact across most protected groups.
Adverse impact happens when a selection method screens out candidates from one group at a disproportionate rate, regardless of whether anyone intended it. You detect it by recording diversity data at each stage of your process and comparing pass rates between groups. Most employers find their largest drop-offs at the CV sift and interview stages, simply because those are the stages nobody measures. Our explainer on adverse impact covers how it's identified and corrected.
Under the Equality Act 2010 you must make reasonable adjustments for disabled applicants. In testing, the most common and most effective adjustment is additional time, because timed conditions usually measure processing speed rather than underlying ability. Others include adjustable display settings such as background colour, contrast and font size, compatibility with screen readers and speech recognition software, and a quiet room. An adjustment has to be reasonable, not unlimited: removing a genuine job requirement isn't required, but changing the format it's assessed in usually is.
No, and the assumption usually points at the wrong culprit. Requirements like UCAS points, degree classification and institution track socioeconomic background as much as capability, so dropping them in favour of a direct measure of ability tends to raise predictive validity rather than lower it. The same goes for practice materials: giving every candidate the chance to prepare removes an advantage that previously went to those already familiar with testing, and makes the resulting scores a better predictor of job performance.