To reduce bias in interviews, use twelve practical controls: define job-related criteria first, use the same questions, ask for past examples, take notes, score each answer before you form an overall view, hide scores from other interviewers until yours are filed, use a diverse panel, plan adjustments, watch the first minutes, separate small talk from scoring, debrief on evidence, and check outcomes over time. No control removes bias alone, but together they make decisions fairer and easier to explain.
Key takeaways
- Bias is normal human thinking. The aim is to make it harder for bias to change the decision.
- The strongest controls are structural: same questions, written criteria, scoring before discussion, and records.
- Interviewer habits matter: do not decide in the first five minutes, and keep small talk neutral.
- Offer adjustments to everyone, and check the process for barriers.
- Measure outcomes over time, where the law allows, and fix patterns you cannot justify by job-related reasons.
How bias shows up in interviews
| Bias | What it looks like in an interview | Control that helps |
|---|---|---|
| First impression | Deciding in the first minutes, then looking for facts to support it | Scoring after each answer; set questions; delay overall judgement |
| Similarity | Preferring people with the same school, hobby, accent or style | Criteria first; evidence scoring; diverse panel |
| Halo and horn | One strong or weak trait colours everything | Score each competency separately |
| Confirmation | Asking leading questions that confirm a first view | Same questions for all |
| Contrast | Comparing a candidate with the previous one rather than with the criteria | Score against the scale, not against others |
| Stereotyping | Assuming skills or limits from age, sex, race, accent or disability | Job-related questions only; notes on evidence |
| Affect or likeability | Choosing the person you enjoy talking to | Evidence-based scorecard |
| Groupthink | The most senior voice sets the view in the debrief | Written scores first; speak in order |
These tendencies affect everyone. They do not mean that interviewers are bad people. They mean that the process should not depend on perfect self-control. The law looks at what you did and at the effect, not only at your intent. See disparate treatment vs disparate impact.
The twelve controls
1. Write job-related criteria first
Decide the must-haves from the role brief before the first interview. Each criterion should link to a task. See the role brief and must-haves vs nice-to-haves.
2. Use the same questions in the same order
Differences in questions create differences in scores that have nothing to do with candidates. Use a structured guide. See structured interviews. Allow follow-up probes to clarify, but give each candidate a similar chance.
3. Ask for past examples and realistic tasks
Behaviour in the past and in realistic tasks gives more evidence than opinions about oneself. See behavioural interview questions.
4. Take notes of facts, not feelings
Write what the candidate said or did. Avoid words such as “good vibe”. Do not write about protected traits or appearance. Tell the candidate that you are taking notes.
5. Score each answer on a defined scale
Use a 1 to 4 scale with written anchors. Score after each answer or each block, not at the end. See the interview scorecard.
6. File your scores before you talk to others
Ask each interviewer to submit scores and evidence within two hours, before any discussion. This protects independent judgement and avoids anchoring on the loudest view.
7. Use a diverse panel where you can
Different interviewers notice different things. Include people from different teams, levels and backgrounds. Do not ask one person to be the “representative” of a group. Train all panel members equally.
8. Plan adjustments and access
Offer adjustments in every invitation. Make the room, the platform and the timing accessible. Use the same length and questions for all, with changes as needed. See the UK Equality Act guide and the EU equal treatment overview.
9. Do not decide in the first five minutes
Be aware of your own first impression, and treat it as a hypothesis to test. Write it down if it helps, then look for evidence that could prove it wrong. Start with the first question as if you knew nothing.
10. Keep small talk neutral
Small talk often drifts into family, age, home town and hobbies. Choose neutral topics, keep it short and do not score it. If the candidate shares personal information, do not follow up and do not use it. See what you can and cannot ask.
11. Debrief on evidence
In the debrief, read the scores first and discuss differences. Ask for examples. Speak in order from the most junior. See how to run a debrief meeting.
12. Check outcomes over time
Where the law allows, compare pass rates by stage and source. Look for patterns you cannot explain. Review what happened to hires at six months. See the four-fifths rule explained for a US method and note that other countries use different tests.
A one-page interviewer checklist
| Before | During | After |
|---|---|---|
| Read the criteria and questions | Start with the first question as if no view yet | Score within 15 minutes |
| Read the CV for facts, not impressions | Ask the same questions; use probes | Write the evidence for each score |
| Offer adjustments | Take notes of facts | File scores before any discussion |
| Plan neutral small talk | Keep to job-related topics | Remove any notes on protected traits |
| Block time for notes | Let the candidate finish | Join the debrief with evidence ready |
Training interviewers in 45 minutes
- 10 minutes: Explain the biases in the table above with a short example each.
- 10 minutes: Practise the notes sheet and the scale with a recorded or written example answer. Everyone scores it alone, then compare.
- 10 minutes: Practise rewriting five risky questions into safe ones.
- 10 minutes: Role-play the first five minutes and the small talk.
- 5 minutes: Agree the rules for the debrief.
Repeat once a year. Training alone may not change results. It tends to work best when the structure is in place: the guide, the scale and the scoring habits.
Limits and honest cautions
No set of controls can promise a bias-free interview. Controls reduce risk, and they can add work. Some, such as panels, add scheduling time. Some, such as strict scripts, can feel less natural. Balance structure with warmth: a structured interview can still be friendly. Also remember that the interview is only one stage. Bias can enter in the ad, the screening and the offer. See how to reduce bias in resume screening and the overview of blind CV screening.
A worked example (fictional)
This example is invented. A team notices that candidates who studied at a few well-known universities are more often chosen, though the role does not require a particular degree. The hiring manager adds controls: written criteria that remove “degree from a top university”, a question list, a scale with anchors, scores filed before discussion and a second reviewer on 20 percent of applications. After two searches, the team compares scores. They find that interviewers still give slightly higher scores to candidates who share their background in the first question, which is about motivation. They replace it with a question about a specific work example and score it with an anchor. The change is recorded and applied to all candidates in the next search.
Example question rewrites
| Biased or risky question | Why it is risky | Better question |
|---|---|---|
| “Tell me about yourself.” | Open; invites personal details and rewards confident talkers | “Tell me about your most relevant work for this role. What did you do yourself?” |
| “Where do you see yourself in five years?” | Rewards rehearsed answers; may disadvantage people with caring duties | “What skills do you want to build in this role, and how would you go about it?” |
| “Are you a culture fit?” | Vague; rewards similarity | “Tell me about a time you worked in a team with different views. What happened?” |
| “Do you have any commitments outside work?” | May reveal family, health or religion | “The role needs these hours. Can you work them?” |
| “How old is your experience?” | Age proxy | “What recent projects show the skill?” |
Rewrite your own question list this way. Ask a colleague who did not write the questions to review them. Keep the revised list in a shared place and version it with a date.
Handling bias you see in colleagues
If you notice a colleague asking a risky question or using biased language in a debrief, act early and calmly. Use a neutral prompt: “Which criterion does that relate to?” If it continues, speak to the person after the meeting and, if needed, raise it with HR. Keep a short note of what happened. In most cases the person did not mean harm and welcomes a quiet correction. Where a comment shows intent to discriminate, escalate to HR and take advice. Create a culture where anyone can say, “Let us check this against the criteria.”
Remote and phone interviews
Remote interviews bring their own risks. Sound quality, accents, lighting and home settings can colour the judgement. Decide in advance that you will ignore the background, the camera quality and the accent unless the job needs specific speech skills, and score only the content of the answers against the criteria. Give candidates the same time and the same tools. Allow captions. Ask whether a candidate needs a different format, such as a phone call. Offer a back-up plan for connection failures so that technical problems do not become a hidden filter.
Measure what happens
| Question | Data to review | Action if the answer is “no” |
|---|---|---|
| Do interviewers use the same questions? | Question logs or checklists | Retrain; use a shared guide |
| Do scores differ widely between interviewers? | Score grid | Clarify anchors; calibrate on a sample |
| Do certain steps lose more people from one group? | Pass rates by stage, where lawful | Review the criteria and the step |
| Do hires succeed at six months? | Manager rating, probation outcome | Review criteria and questions |
| Do candidates report a fair process? | Short survey | Fix the issues they name |
Calibration: practise scoring together
Before a big search, run a 30-minute calibration. Give all interviewers the same written example answer to one question. Ask each person to score it alone with the anchors. Compare the scores. If they differ by two points or more, discuss why and adjust the anchor wording. Repeat with a second example. After this, interviewers use the scale in a more similar way, and the scores in the debrief mean the same thing to everyone. Keep the examples for new interviewers. Calibration costs little and improves the quality of every later decision.
Finally, share a one-page summary of these controls with every new interviewer and ask them to confirm they have read it before their first interview. A short, signed-off page makes the rules visible, sets a shared standard and gives you a simple record that interviewers were briefed.
Keep the list of controls visible in the interview room or in your calendar invitation, so that the habits are in front of you when you need them. Habits become easier with repetition, and the second search will feel faster than the first.
Official sources for this topic: EEOC: prohibited employment policies and practices, Acas: recruitment. These pages are maintained by the authorities and may change, so check them for the current text.
Frequently asked questions
Can I remove bias completely from interviews?
No. You can reduce it with structure, evidence and checks, and you can make decisions easier to explain.
Does unconscious bias training work?
Training raises awareness, but results usually change more when the process changes: same questions, written criteria and scoring habits.
Should every interview have a panel?
Not always. A panel adds views, but it adds cost. Use at least two perspectives on important decisions, and share independent scores.
Is small talk a problem?
It can be if it drifts into protected topics or influences the score. Keep it short and neutral and do not score it.
How do I know if my interviews are fair?
Check consistency of questions and scoring, review outcomes where the law allows, and ask candidates for feedback on the process.
Related guides
- Hiring guides for hiring managers
- Structured interviews: what they are and how to run one
- How to reduce bias in resume screening: 15 practical steps
- How to run a debrief meeting after interviews
- All hiring guides for employers
- Sex, gender and pregnancy: rules for interview questions
- How to hire for culture without hiring clones
- Career gaps, job hopping and short stints: how to read them fairly
- Fair hiring for older workers and career returners
This is general information, not legal advice.