Unconscious bias examples: 16 at work and how to fix each
16 biases grouped by where they show up at work, each with a scenario and the process change that reduces it. Plus the best evidence and a printable checklist.
What is unconscious bias?
Unconscious bias, also called implicit bias, is a mental shortcut that shapes how we judge people without our noticing. These unconscious bias examples are habits of judgment everyone has, not a verdict on anyone's character. At work they matter most at a few moments: hiring, meetings, reviews and who gets the next opportunity.
Conscious bias is a preference someone knows they hold. Unconscious bias works below awareness, so good intentions don't switch it off. What does help is a process that makes the fair call the easy one. Every fix below is a change to how a decision gets made.
16 unconscious bias examples at work
The types of unconscious bias you'll meet most at work, grouped by moment so you can lift one group straight into a briefing. Every scenario is made up but typical.
In hiring
1. Affinity bias: Warming to people who remind us of ourselves.
Example: An interviewer rates a candidate higher after they discover they went to the same college and both run marathons.
Fix: Score each answer against a rubric written before the first interview, and leave small talk out of the score.
What it isn't: Valuing a skill the job genuinely needs because you happen to have it too.
2. Halo effect: Letting one strong trait color the whole judgment.
Example: A candidate from a famous company gets top marks on teamwork, though nobody asked a single teamwork question.
Fix: Rate each competency on its own line, and leave a line blank when you have no evidence for it.
What it isn't: Giving credit for a strength that was actually demonstrated in the interview.
3. Horns effect: The halo in reverse: one weak point drags every rating down.
Example: A candidate stumbles on the first question, and the panel marks them down on everything that follows.
Fix: Ask questions in a fixed order, score each answer straight after it, and only then move on.
What it isn't: Scoring a real gap low when the role depends on that skill.
4. Name bias: Judging an application by the name at the top.
Example: Two near-identical resumes get different callbacks, and the only real difference is the applicant's name.
Fix: Hide names and photos at the first screen, and screen against written must-haves.
What it isn't: Checking the name to schedule an interview or pronounce it correctly.
5. Contrast effect: Judging someone against whoever came just before, not against the standard.
Example: A solid candidate seems weak because the person interviewed right before them was outstanding.
Fix: Compare every candidate to the scorecard, never to each other, until the debrief.
What it isn't: Ranking finalists against each other at the very end, once each has been scored on their own.
For set questions and a printable scorecard, see our interview questions to ask candidates.
In meetings
6. Conformity bias: Going along with the group view to avoid standing out.
Example: Five people nod at a launch plan; two had doubts and stayed quiet because nobody else objected.
Fix: Collect each person's view in writing or a quick anonymous poll before anyone discusses it.
What it isn't: Agreeing because the evidence has actually changed your mind.
7. Authority bias: Giving extra weight to an idea because of who said it.
Example: A director's offhand suggestion becomes the plan, while an analyst's better-tested idea gets one line in the notes.
Fix: The most senior person speaks last, and ideas go on the board without names.
What it isn't: Deferring to someone's real expertise on the topic in front of you.
8. Anchoring bias: Sticking close to the first number or idea we hear.
Example: The first person to estimate a project says six weeks, and every estimate after lands between five and seven.
Fix: Everyone writes an estimate privately and shares at the same moment, then you discuss the spread.
What it isn't: Using last year's real data as a starting point and adjusting from it.
9. Status quo bias: Preferring the current way just because it is the current way.
Example: A team keeps a weekly status meeting nobody finds useful because "we've always had it".
Fix: Give every recurring process a review date and an owner who must argue for keeping it.
What it isn't: Keeping something that works after you've weighed the alternatives.
In reviews and promotion
10. Recency bias: Rating a whole year on the last few weeks.
Example: A strong year ends with one missed deadline in November, and the review reads like an average year.
Fix: Keep a short running log of wins and misses for each person, and review it before writing anything.
What it isn't: Weighting a recent change in performance that has lasted and still matters.
11. Confirmation bias: Noticing what fits what we already believe and missing the rest.
Example: A manager who decided early that a hire was slow remembers every late reply and none of the early deliveries.
Fix: Before calibration, write one piece of evidence that cuts against your rating.
What it isn't: Holding a view that the full record supports.
12. Gender bias: Judging the same behavior or record differently depending on gender.
Example: Two engineers push back on a deadline in the same words; one is called decisive and the other difficult.
Fix: Use the same behavior-based wording for everyone, and compare ratings by group in aggregate before they're final.
What it isn't: A difference in ratings that the documented work explains.
13. Leniency bias: Rating everyone high to avoid hard conversations.
Example: Every person on a team is rated "exceeds", so the strongest work can't be told apart from the rest.
Fix: Define each rating level with examples, and have calibration compare spreads across managers.
What it isn't: A genuinely strong team earning strong ratings against a shared standard.
Need behavior-based wording? Our performance review phrases help.
Every day
14. Attribution bias: Explaining others' mistakes by character and our own by circumstance.
Example: When a colleague misses a deadline they're disorganized; when we miss one, the brief changed.
Fix: Ask what got in the way before deciding why it happened, and write the answer into the post-mortem.
What it isn't: Addressing a pattern that keeps repeating after the obstacles have been cleared.
15. Proximity bias: Favoring the people we see in person most often.
Example: The stretch project goes to whoever was at the desk nearby when it came up, and remote staff never hear of it.
Fix: Post stretch work in a shared channel with a deadline to raise a hand, and pick against written criteria.
What it isn't: Choosing someone because the work truly needs a person on site.
16. Age bias: Assuming what someone wants or can do from their age, older or younger.
Example: A team picks pilot users for a new tool and skips its most experienced member, assuming they'd rather keep the old one.
Fix: Open pilots, training and stretch work to everyone, and let each person opt in or out themselves.
What it isn't: Respecting a preference the person has told you about themselves.
The best evidence on bias at work
Many lists quote numbers with no study behind them. These four are real, widely cited and described here with their limits.
Are Emily and Greg More Employable Than Lakisha and Jamal?
Bertrand and Mullainathan, American Economic Review, 2004
The researchers answered help-wanted ads in Boston and Chicago with made-up resumes. Resumes with White-sounding names got about 50 percent more interview callbacks than similar resumes with African-American-sounding names.
Limits: Callbacks, not hires, from newspaper ads in two cities in the early 2000s.
Systemic Discrimination Among Large U.S. Employers
Kline, Rose and Walters, Quarterly Journal of Economics, 2022
More than 83,000 made-up applications went to jobs at 108 of the largest US employers. Distinctively Black names cut the chance of an employer contact by 2.1 percentage points, and the gap was concentrated: the most discriminating fifth of firms accounted for nearly half of it.
Limits: Measures first contact, not hiring. Firms with centralized recruiting tended to have smaller gaps, a hint that process matters.
Science faculty's subtle gender biases favor male students
Moss-Racusin and colleagues, PNAS, 2012
127 science professors rated the same application for a lab manager job, named either John or Jennifer. They rated John more competent and more hireable, offered him a higher salary and more mentoring. Female professors did this as much as male professors.
Limits: Biology, chemistry and physics faculty only, one fictional student application, rated on paper.
A meta-analysis of procedures to change implicit measures
Forscher and colleagues, Journal of Personality and Social Psychology, 2019
Across 492 studies with 87,418 participants, interventions could shift scores on implicit bias tests, though usually only a little, and those shifts did not reliably carry over into behavior.
Limits: Mostly short lab studies. It's the main reason to change processes rather than rely on awareness alone.
One caution: the Implicit Association Test is a useful teaching tool, but Project Implicit itself says it should not be used to diagnose an individual or to make hiring decisions. Don't use a score to judge a colleague.
Does unconscious bias training work?
On its own, not much. The 2019 meta-analysis above found that shifting implicit bias scores didn't change behavior. In Why Diversity Programs Fail, sociologists Frank Dobbin and Alexandra Kalev reported that mandatory diversity training often failed to help and could backfire, while structural changes did better. Process beats awareness. Four changes do the most:
- Decide the criteria first. Write what good looks like before you meet a candidate or open a review.
- Judge alone, then together. Independent scores before any discussion keep the first voice from setting the rest.
- Make people explain decisions. Managers who know they'll walk a panel through their reasons rely on evidence more.
- Look at the pattern. Compare outcomes by team and group in aggregate each cycle; one decision rarely shows bias, a year of them can.
A session still has a job: it gives a team shared words for what they're trying to catch, so the new process makes sense. Keep it voluntary where you can, practical and tied to the decisions your team makes. For small daily habits, our DEI tip of the day has 120.
A printable bias checklist for panels and calibration
12 checks to run before a hire or a rating is final. Print it for the panel, or paste it into your hiring and review templates.
Before interviews
- The must-haves for the role are written down and shared with the panel.
- Each interviewer owns different competencies, with set questions for each.
- The first screen hides names, photos and addresses.
During and after interviews
- Each answer is scored against the rubric, right after it is given.
- Every score has a note of what the candidate said or did.
- Candidates are compared to the scorecard, not to each other, until the debrief.
Before calibration
- Each manager has reviewed the full year's log, not just the last month.
- Each rating has one piece of evidence for it and one that cuts against it.
- Rating levels are defined with examples everyone has read.
In the calibration meeting
- The most senior person gives their view last.
- Spreads are compared across managers, and very high or very low spreads are questioned.
- Ratings and stretch assignments are compared by group in aggregate before they're final.
Run a session with a facilitator
A 60-minute live workshop gives a team the shared language before you change the process. Each runs with a facilitator on your video call.
- Diversity, Equity & Inclusion Foundations covers the core principles, with interactive exercises and real-life examples on recognizing biases.
- Equity vs. Equality explains the difference between the two and how to apply both at work, with evidence-based information.
- Allyship in the Workplace teaches how to support underrepresented colleagues and take action for inclusion.
- They run on Zoom by default, or Teams, Google Meet or Webex on request; book about 2 weeks ahead. Compare them on our DEI workshops page.
Three 60-minute DEI workshops, with today's price. Get a free quote with your date and headcount.
To see whether people feel fairly treated, KraftyLab Intelligence sends short pulse surveys by email that people answer in the browser with no account, with AI summaries of the answers. Only compare teams where enough people answered for the numbers to mean something. More on the culture side in our guides to an inclusive workplace and employee resource groups, and words for a slide in our diversity and inclusion quotes.
