Every article in this series ends with things a reader might do differently. This one is about a finding which says that a reader will apply those lessons to their colleagues and not to themselves, and that being told so will not change it.
Key Takeaway
From the founding paper's abstract: "Participants in one follow-up study who showed the better-than-average bias insisted that their self-assessments were accurate and objective even after reading a description of how they could have been affected by the relevant bias."[1] On our own arithmetic, someone who weights their own judgment at half in a group of ten gets the accuracy of a panel of 3.6, and no group size can lift that above 4.
The Verdict, Stated First
Five claims, in descending order of confidence.
One. The finding is robust and unusually well established. A 2002 paper with roughly 890 citations, whose full abstract we obtained from the lead author's own institutional portal, and a 2023 review by the same author reporting it generalises across bias types, cultures and age groups.
Two. Explanation does not remove it. Participants shown a description of how a bias could have affected them insisted their self-assessments were accurate anyway, which is stated in the founding abstract rather than inferred.
Three. The mechanism explains why. People assess their own bias by introspection, where bias does not appear, and others' by behaviour, where it does. That asymmetry is not correctable by trying harder to introspect.
Four. It is unrelated to how good your decisions are. A large study reportedly found people who scored better or worse on decision-making competence were no more or less susceptible, though it does predict susceptibility to other biases.
Five. The cost to a group is severe and arrives early. On our own arithmetic, weighting yourself at four times your fair share halves the accuracy a panel of ten could give you, and at half weight the panel caps at four effective judges however many people are in the room.
Our Grades For These Claims
Applying the scheme from the first article in this series.
Grade A for the 2002 findings, from a full abstract obtained from the lead author's own university research portal.
Grade A for the 2023 generalisation claims, from an abstract obtained verbatim from the publisher.
Grade B for the mechanism, quoted from a 2007 paper via a secondary source, with the paper itself not obtained.
Grade C for the decision-competence null, which reaches us through an encyclopedia summarising a 2015 paper we did not obtain.
Grade D for two specific percentages, which come from a sceptics' reference site and which we could not verify against any paper.
Our position: the core finding is among the best-sourced in this series and the numbers attached to it are among the worst.
A Note On Method
Everything here is verified to August 2026.
We obtained the 2002 paper's full abstract from the lead author's own university research portal[1], and its citation with DOI from three further independent sources[2]. We did not obtain the paper.
We obtained the 2023 review's abstract verbatim from the publisher[3] and not the review.
The mechanism is quoted from the 2007 paper inside a sceptics' reference site[4], flagged at every use, and we did not obtain that paper.
The decision-competence finding comes from an encyclopedia[2] summarising a 2015 paper we did not obtain.
Two percentages in circulation reach us only through a sceptics' site and we grade them D and use neither in any calculation.
All arithmetic is ours and rests on assumptions generous to the group, which we state.
This article discusses research on self-perception. It is not governance, management or decision-process advice.
The 2002 Paper
The source, with unusually good provenance.
Pronin, E., Lin, D. Y., and Ross, L. (2002), The Bias Blind Spot: Perceptions of Bias in Self Versus Others, Personality and Social Psychology Bulletin, 28(3), 369–381, March, DOI 10.1177/0146167202286008[1][2].
The third author is the same Lee Ross whose 1977 paper the previous article covered, which makes two consecutive articles resting partly on one researcher's work twenty-five years apart.
The record gives its citation count as 889 on one page of the same portal and 892 on another, which is a snapshot difference rather than a discrepancy[1]. The citation is confirmed independently, including its issue number, by three further reference lists[7].
Three observations, ours.
The term is named after the visual blind spot, which an encyclopedia records and which is a good analogy: a gap in perception that is not experienced as a gap[2].
The definition is a metabias, being a bias about biases, and that recursive quality is what makes it the appropriate seventy-first article in a series about biases.
And we obtained the abstract from the lead author's own institution, which is the best provenance available short of the publisher.
Three Studies
What the paper did, from its own abstract.
"Three studies suggest that individuals see the existence and operation of cognitive and motivational biases much more in others than in themselves. Study 1 provides evidence from three surveys that people rate themselves as less subject to various biases than the 'average American,' classmates in a seminar, and fellow airport travelers. Data from the third survey further suggest that such claims arise from the interplay among availability biases and self-enhancement motives."[1]
Four observations, ours.
The three comparison groups are chosen well and the choice is the point. The average American, classmates in a seminar, and fellow airport travellers range from an abstraction to a group the participant is physically standing among.
That range forecloses an obvious objection. You might rate yourself above an abstraction and not above the person beside you, and the finding holds against both.
The proposed sources are named as two, not one: availability biases and self-enhancement motives. The first is cognitive and the second is motivational, and the abstract says they interact.
And that pairing recurs through the whole literature, including a dispute that is still open in 2025, as a section below reports.
The Follow-Up That Matters Most
The sentence a reader of this series should sit with.
"Participants in one follow-up study who showed the better-than-average bias insisted that their self-assessments were accurate and objective even after reading a description of how they could have been affected by the relevant bias."[1]
Four observations, ours.
This is the exact intervention this publication performs. Reading a description of how a bias could have affected you is what an article in this series is.
The result is not that the intervention worked partially. Participants "insisted", which is the abstract's own word, and insisting is what people do when they have understood the description and rejected its application.
Note what is not claimed. The abstract does not say the description made things worse, only that self-assessments were maintained, and we will not upgrade that to a stronger claim.
And a sceptics' reference site reports a figure for how many insisted, which we could not verify against the paper and therefore do not use[4]. The abstract's qualitative statement is enough and is properly sourced.
And The Study After That
The cleanest demonstration in the paper.
"Participants in a final study reported their peer's self-serving attributions regarding test performance to be biased but their own similarly self-serving attributions to be free of bias."[1]
Four observations, ours.
The word doing the work is "similarly." The two sets of attributions were the same kind, and only the ownership differed.
That design removes the most reasonable defence available. You might genuinely be less biased than your peers, and this study holds the behaviour constant so that possibility cannot explain the gap.
It is also the version closest to business life. Explaining a good quarter by your judgment and a bad one by conditions is a self-serving attribution, and the finding is that you will spot it in a colleague and not in yourself.
And the abstract closes by naming the consequence: "The relevance of these phenomena to naïve realism and to conflict, misunderstanding, and dispute resolution is discussed."[1]
Twenty-One Years Later
The lead author's own review, and it is more expansive than the original.
Pronin, E., and Hazel, L. (2023), Humans' Bias Blind Spot and Its Societal Significance, Current Directions in Psychological Science[3].
Its abstract: "Human beings have a bias blind spot. We see bias all around us but sometimes not in ourselves. This asymmetry hinders self-knowledge and fuels interpersonal misunderstanding and conflict. It is rooted in cognitive mechanics differentiating self- and social perception as well as in self-esteem motives. It generalizes across social, cognitive, and behavioral biases; begins in childhood; and appears across cultures."[3]
Four observations, ours.
The hedge in the second sentence is worth keeping. "Sometimes not in ourselves" is weaker than the popular version and is the author's own wording twenty-one years on.
The generalisation claims are three and they are strong: across bias types, from childhood, and across cultures. Each would be a separate research programme.
The mechanism is again given as two things, cognitive mechanics and self-esteem motives, matching the 2002 abstract's pairing.
And we obtained the abstract and not the review, so we cannot tell you what evidence supports the generalisation claims or how strong it is.
Where It Shows Up
The commercial sentence, from the same abstract.
"People show a bias blind spot in high-stakes contexts, including investing, medicine, human resources, and law. Strategies for addressing the problem are described."[3]
Four observations, ours.
Investing and human resources are named explicitly, which spares us the transfer argument this series usually has to make.
The four domains share a property. All are fields where practitioners hold formal credentials and believe their training protects them, which is exactly the belief the finding attacks.
The closing sentence promises strategies and we did not obtain them, which is the most useful thing in the review and the part we cannot report.
A separate 2025 paper's introduction lists further domains from the citing literature, including criminal and legal procedures, financial and organizational practice, and conflict eruption and resolution[5].
The Introspection Illusion
Why the blind spot exists, and the account is elegant enough to be worth stating carefully.
A sceptics' reference site quotes a 2007 paper arguing that the blind spot "involves the value that people place, and believe they should place, on introspective information (relative to behavioral information) when assessing bias in themselves versus others. Participants considered introspective information more than behavioral information for assessing bias in themselves, but not others."[4]
The source is Pronin, E., and Kugler, M. (2007), Valuing thoughts, ignoring behavior: The introspection illusion as a source of the bias blind spot, Journal of Experimental Social Psychology, 43[6]. This is a sceptics' reference site, not an academic source, flagged at every use, and we did not obtain the paper.
Four observations, ours.
The account has two halves and both are needed. You judge your own bias by looking inward. You judge others' by watching what they do.
Those two methods have different power, and the asymmetry is fatal. Bias does not announce itself in introspection, because a biased judgment feels exactly like an unbiased one from inside.
So the person is not being unreasonable. They look, honestly, and find nothing, because there is nothing to find by looking.
And the clause "and believe they should place" is the sharpest part. It is not merely that people use introspection; they think introspection is the right instrument, which makes the error normative rather than accidental.
Why That Makes It Hard
The consequence for every remedy in this series. Ours.
Four observations.
Most advice about bias amounts to think about whether you are biased, which on this account is asking someone to use the instrument that cannot detect the thing.
That is why the 2002 follow-up result is what it is. Reading a description of a bias improves your model of the bias and does nothing to your access to your own state, so you check, find nothing, and conclude the description does not apply.
The remedy the account implies is specific and unappealing. Judge yourself by your behaviour, as you judge others, which means records rather than reflection.
And that is why every practical recommendation in this series has converged on the same instrument. Write the number down beforehand, because a written prior is behavioural evidence about yourself of the kind you already accept about other people.
Not Related To Decision-Making Ability
A finding that complicates the picture, reported at the strength our sourcing allows.
An encyclopedia records: "Initial evidence suggests that the bias blind spot is not related to actual decision-making ability. Participants who scored better or poorer on various tasks associated with decision making competence were no more or less likely to be higher or lower in their susceptibility to bias blind spot."[2]
It attributes this to Scopelliti, I., Morewedge, C. K., McCormick, E., Min, H. L., Lebrecht, S., and Kassam, K. S. (2015), Bias blind spot: Structure, measurement, and consequences, Management Science[3]. This is an encyclopedia, not an academic source, flagged at every use, and we did not obtain the paper.
Four observations, ours.
If the description holds, being good at decisions does not protect you, and the two are separate traits rather than one.
That is a genuinely surprising null. The natural expectation is that better decision-makers would be better calibrated about their own judgment, and this reports no relationship in either direction.
It also means the blind spot is not a proxy for competence. You cannot infer someone is a poor thinker from the fact that they think they are unbiased, which is a courtesy the finding extends to everyone equally.
And the paper appeared in Management Science, which is a business journal, and its title promises a measurement instrument. We did not obtain it, which is a real loss because a measurable version would be usable.
But It Predicts Other Biases
The clause that follows immediately, and it changes the reading.
The same encyclopedia entry continues: "Bias blind spot does, however, appear to increase susceptibility to related biases."[2]
Three observations, ours.
So the picture is a specific pair of findings. The blind spot does not track how good your decisions are, and it does track how vulnerable you are to particular errors.
Those are compatible and the combination is coherent. General ability and specific vulnerability are different quantities, and a strong reasoner with a large blind spot would be exactly the person to worry about.
And we obtained no magnitude for either half, so we report a direction and a null without knowing how large either is.
A Motivational Account, Still Open
Where the argument stands in 2025, from a paper's introduction.
It records that the blind spot "is believed to emerge due to people's overreliance on introspection for assessing their biases (which is unlikely to turn up evidence of bias), while bias in other people is ascribed based on their behaviors." Then it raises the alternative: "Many biases, however, are perceived as negative and thus the bias blind spot may reflect the desire to see oneself in a positive light. Moreover, not all biases are necessarily undesirable, and thus people may be motivated to admit to biases that are considered desirable."[5]
We obtained the introduction and not the findings.
Four observations, ours.
The test proposed is elegant and decisive in principle. If the blind spot is motivational, it should shrink or reverse for biases people would be pleased to have.
If it is purely the introspection illusion, desirability should not matter at all, because the instrument is equally blind either way.
That the question is being tested in 2025, twenty-three years after the founding paper, is the eighth time this series has reported a mechanism still open decades on.
And we cannot tell you what they found, which is the most conspicuous gap in this article and sits exactly where the dispute would be settled.
What It Costs A Group
Turning the finding into a number, because the individual version is hard to price and the group version is not. Ours.
Three observations.
The blind spot's operational form in a business is a weighting decision. You believe your own judgment is less contaminated than the others in the room, so you weight it more heavily.
That is not a belief anyone states aloud, and it does not need to be. It shows up in whose view prevails when the room disagrees, which is observable.
And the cost of it is computable from the arithmetic of averaging, which this publication covered separately. Combining independent judgments reduces error, and the reduction depends entirely on the weights.
The Self-Weighting Arithmetic
Our own calculation. Assume every person in a group holds an independent estimate with the same error variance, which is generous to the group and therefore understates nothing.
Equal weighting across N people cuts error variance to one over N. If instead you place weight w on yourself and split the remainder equally among the others, the resulting accuracy is equivalent to a smaller equally-weighted panel. We call that the effective number of judges.
In a group of ten, where a fair share would be 10 percent:
At 10 percent self-weight: 10.00 effective judges. At 30 percent: 6.92. At 40 percent: 5.00. At 50 percent: 3.60. At 60 percent: 2.65. At 75 percent: 1.76. At 100 percent: 1.00.
Four observations.
The decline is steep and early. Moving from a fair share to 30 percent, which is a modest amount of self-favouring, already costs three of the ten judges.
At half weight the group of ten is worth 3.6, meaning nearly two thirds of the available accuracy has gone.
The figures for groups of five and twenty are close to these at every level of self-weighting, which is the result the next-but-one section explains.
And every number here rests on independence between judgments, which real meetings violate badly. Correlated judgments are worth less than this table assumes, so the group side of the comparison is flattered.
Four Times Your Fair Share
The threshold, which is the usable form. Our own calculation.
In a group of ten, the self-weight that halves the value of the panel is 40.0 percent, against a fair share of 10.0 percent.
Four observations.
Weighting yourself at four times your fair share costs half the accuracy the group could have given you.
Four times sounds like a lot and is not. In a room of ten, taking 40 percent of the weight is what happens when the most senior person's view carries, which describes a great many meetings.
And the framing matters for how the finding should be used. This is not a claim that you should defer; it is a claim about what a specific weighting costs, and a reader who genuinely is more accurate than the room should weight themselves more.
Which is precisely where the blind spot bites. The question is whether you are more accurate or whether you only feel less biased, and the literature says introspection cannot tell you which.
The Cap
A result that falls out of the algebra and which we think is the most useful thing here. Our own derivation.
As the group grows, the effective number of judges approaches one divided by the square of your self-weight, and stops there.
At a self-weight of 50 percent, the effective panel across group sizes of 3, 5, 10, 20, 50 and 100 runs 2.67, 3.20, 3.60, 3.80, 3.92 and 3.96, approaching 4.
Four observations.
At half weight the panel caps at four effective judges, however many people are in the room. Going from ten people to a hundred buys you 0.36 of a judge.
The general rule is exact and easy to carry. Your self-weight sets a ceiling of one over its square. A quarter weight caps at 16, a third caps at 9, a half caps at 4.
Which means the size of the group is almost irrelevant once meaningful self-favouring is present. Adding advisers, consultants or board members buys nothing if the weighting does not change.
And that is the practical form of the whole finding. The blind spot does not reduce the value of consultation; it caps it, and the cap is set by a number nobody in the room states.
Two properties of that cap are worth spelling out, because they are counterintuitive in opposite directions. It is unforgiving at the top: the difference between weighting yourself at a quarter and at a half is the difference between sixteen effective judges and four, so small changes near the middle of the range cost enormously.
And it is forgiving at the bottom. Moving from a fair tenth to a fifth in a group of ten still leaves 9.00 effective judges, and a quarter still leaves 8.00, so mild self-favouring is nearly free. The damage is concentrated in a band most people would not describe as arrogant.
Two things follow that we would put to anyone designing a decision process. The useful intervention is not to eliminate self-favouring but to keep it under about a quarter, which is a far more achievable target than impartiality and captures most of the available accuracy.
And the steepness above that point explains a pattern any owner will recognise. Two firms can consult identically wide and get completely different value from it, because the visible input, being who was asked, is the same and the invisible one, being how much each answer moved the decision, is not.
Our own draft of this section said a fifth self-weight leaves "about eight" effective judges. The computation gives 9.00, and a quarter gives 8.00, so we had shifted the figure by one row. We correct it here rather than quietly, because an article about failing to see one's own errors is a poor place to hide one.
What Actually Survives
Our reading, stated directly.
Five statements.
People see bias more in others than in themselves, robustly. Three studies in 2002, roughly 890 citations, and a 2023 review reporting generalisation across bias types, cultures and childhood.
Reading about a bias does not remove the blind spot. Participants shown a description insisted their self-assessments were accurate anyway, which is in the founding abstract.
The mechanism is an asymmetry of instruments. Introspection for yourself, behaviour for others, and only the second detects bias.
It does not track decision-making ability and does track vulnerability to other biases, on an encyclopedia's account of a paper we did not obtain.
And whether it is cognitive or motivational was still being tested in 2025.
The Objection We Would Raise
The strongest argument against reading too much into this, which none of our sources makes. Ours.
Four observations.
Rating yourself as less biased than average could simply be true for roughly half the population. Somebody is below average on any trait, and the studies establish an asymmetry rather than an error in any individual case.
The 2002 paper's final study is the answer to that, and it is why it exists. Holding the behaviour constant, participants called a peer's self-serving attribution biased and their own identical one unbiased, which cannot be explained by anyone actually being less biased.
But the objection survives in weakened form and matters commercially. The finding is about a population tendency, not a diagnosis of any reader, and this article cannot tell you whether your own self-weighting is justified.
What it can tell you is where to look. If your confidence in your own judgment rests on introspection, the literature says that evidence is worthless in either direction, and if it rests on a record of past calls then the question is already settled by the record.
This Publication's Own Problem
The section this article exists for. Ours.
Four observations.
This series has published seventy articles describing failures of judgment, each ending with things a reader might do differently. The finding here says a reader will apply them to their colleagues.
Worse, the 2002 follow-up describes our exact intervention and reports it failing. Reading a description of how a bias could have affected you left self-assessments intact.
We are not going to resolve that, and pretending otherwise would be the error the article describes. We have no evidence that reading these articles changes anyone's self-assessment, and the one relevant study says it does not.
What we can say is narrower and we think it is enough. Almost every recommendation this series has made has been procedural rather than introspective, and procedures do not require anyone to detect their own bias in order to work.
What We Can Honestly Claim
Setting out what the series is and is not doing, given the finding. Ours.
Four observations.
Writing an estimate down before the outcome works whether or not you believe you are biased, because it produces behavioural evidence about yourself of the kind you already trust about others.
Counting all four cells of a contingency table works the same way. It does not require you to accept that your impression was wrong; it just computes the number.
Scoring factors separately and adding them, as the sixty-seventh article described, removes a judgment step rather than improving it.
And that is the honest claim for this whole series. Not that reading it will make you less biased, which the evidence does not support, but that it supplies procedures which do not depend on your believing you need them.
The Recursion, And Why It Is Not A Trap
The obvious response to this article, which we want to handle before the applications. Ours.
A reader can reasonably say: if nobody can see their own bias, then neither can the people who wrote this, and the finding defeats itself. That is true and it is less damaging than it sounds.
Four observations.
The finding is not self-undermining in the way a paradox is. It says introspection is a poor instrument for detecting bias, which is a claim supported by behavioural evidence rather than by anyone's introspection.
Nor does it imply that all judgments are equally suspect. It implies that a specific kind of evidence, being your own sense of your own objectivity, carries no information, and other kinds still do.
The trap only closes if you try to escape it by thinking harder. Someone who reads this and concludes they must be especially vigilant has done exactly what the 2002 follow-up describes, and someone who starts keeping a record has not.
And that is why we would not treat the recursion as a reason for despair. The finding does not say judgment is hopeless. It says one particular test of it is, and it names the replacement.
Your Meetings
The first application. Ours, untested, and not governance or management advice.
Four points.
The measurable version of the blind spot in a firm is whose view prevails when the room disagrees, and that is recordable without anyone discussing psychology.
On our own arithmetic, a founder taking 40 percent of the weight in a group of ten halves the accuracy the group could supply.
The cap result gives the sharpest management implication. If the weighting does not change, hiring more advisers buys nothing, because the ceiling is set by the self-weight and not by the headcount.
And the cheap procedure is the one this series keeps arriving at. Collect written estimates before the discussion, which both records the weights and removes the anchoring the discussion would create.
Your Advisors
The second application, and it is uncomfortable for a firm that sells advice. Ours.
Four points.
The 2023 abstract names investing, medicine, human resources, and law as high-stakes contexts where the blind spot appears, which is a list of advisory professions.
So the finding applies to the adviser as much as the client, and this publication is written by an accounting and advisory firm. We are not exempt and we would not claim to be.
The useful consequence is about what you should ask an adviser for. A recommendation is a judgment; a recorded forecast is behavioural evidence, and the second can be checked and the first cannot.
And the same test works in both directions. An adviser who cannot show you what they predicted last time is asking to be judged by introspection, which is the instrument the finding says fails.
One qualification on that test, since it is easy to apply unfairly. Most advisory work does not produce checkable forecasts, and an adviser who structures a transaction or drafts a policy has not predicted anything that a later date will score.
So the test applies where it applies. Ask for the record where a record could exist, being cash flow projections, valuations, timelines and recommendations with a stated expected outcome, and do not treat its absence elsewhere as evasion.
The Remedy That Follows
What the mechanism implies, stated as plainly as we can. Ours.
Four observations.
If the blind spot comes from using introspection on yourself and behaviour on others, the remedy is to use behaviour on yourself too.
That is not a matter of effort or humility. It is a change of instrument, and it works for the same reason your judgments about colleagues are better calibrated than your judgments about yourself.
In practice it means keeping the kind of record you would want from someone else. What you predicted, when, and what happened, which is a file rather than a disposition.
And it explains why the recommendation in this series is always the same instrument in different clothes. There is only one way to get behavioural evidence about your own judgment, and it is to write things down before you know the answer.
Three Variants In Two Citations
The bibliographic entry, since this series keeps a running count.
The 2002 paper's volume appears as 28 in every academic source we checked and as 23 in one 2025 paper's own abstract[5].
The 2007 introspection paper's page range appears in one publisher's reference list as 565–678[6], which would be a 114-page article in a journal whose papers run to a dozen or so pages.
And that same paper is dated 2006 by a sceptics' reference site[4] and 2007 by two publisher reference lists[3][6].
Three observations, ours.
The volume error is the consequential one, because a reader searching volume 23 of that journal will not find the paper.
The page range is almost certainly a transposition of 578, and it is worth flagging because it appears in a publisher's own reference list rather than a blog.
That brings the running count of bibliographic variants across this series to twenty-three, and we note without comment that they appear in an article about failing to see one's own errors.
What To Do
Do not expect reading about biases to change your self-assessment. The founding paper reports participants shown a description of a relevant bias insisting their judgments were accurate anyway.
Judge yourself by records, not reflection. The mechanism is that you use introspection on yourself and behaviour on others, and only the second detects bias.
Notice whose view prevails when your room disagrees. That is the observable form of the weighting decision, and it does not require anyone to discuss psychology.
Learn the cap. On our own arithmetic, your self-weight sets a ceiling of one over its square: a half caps at four effective judges, a third at nine, a quarter at sixteen, however large the group.
Do not add advisers without changing the weighting. Going from ten people to a hundred at half self-weight buys 0.36 of a judge.
Collect written estimates before discussion. This records the weights and removes the anchoring the discussion would otherwise create.
Ask an adviser what they predicted last time. A recommendation is a judgment and a recorded forecast is behavioural evidence, and only the second can be checked.
Prefer procedures that do not require you to believe you need them. That is the only kind this finding leaves standing.
The Limits Of This Analysis
Several caveats matter. This article discusses research on self-perception and is not governance, management or decision-process advice; the applications are our own reasoning and untested. Everything is verified to August 2026. We did not obtain the 2002 paper, only its full abstract from the lead author's own university research portal, so we report no sample sizes, no effect sizes and no detail of the three surveys or two follow-up studies. We did not obtain the 2023 review, only its abstract, which means its three generalisation claims and its promised strategies for addressing the problem are reported without any of the evidence behind them; the strategies are the most useful thing in it and we have none of them. We did not obtain the 2007 paper that supplies the mechanism, and quote it through a sceptics' reference site, flagged at every use. We did not obtain the 2015 Management Science paper, and its two findings, that the blind spot is unrelated to decision-making ability and does increase susceptibility to related biases, reach us through an encyclopedia; we have no magnitude for either. We did not obtain the 2025 paper's findings, only its introduction, so the cognitive-versus-motivational question is reported as open without its answer. Two specific percentages in circulation, concerning how many participants insisted on their objectivity and how many managers rate themselves above average, reach us only through a sceptics' reference site; we grade them D, use neither in any calculation, and note that the founding abstract supports the first claim qualitatively without supplying the number. All arithmetic is ours. It assumes judgments that are independent, equally accurate and unbiased, which real meetings violate badly; correlated judgments are worth less than the table assumes, so the figures flatter the group side of the comparison and the true cost of self-favouring may be smaller than shown. And nothing in this article establishes that any reader's self-weighting is wrong: a person genuinely more accurate than their room should weight themselves more, and the finding is that introspection cannot tell you which case you are in.
Frequently Asked Questions
What is the bias blind spot?
Does learning about biases help?
Why is it so hard to fix?
Are better decision-makers protected?
What does it cost in a meeting?
What is the cap?
Does this undermine the rest of the series?
References
- Pronin, E., Lin, D. Y., & Ross, L. (2002). The bias blind spot: Perceptions of bias in self versus others. Personality and Social Psychology Bulletin, 28(3), 369–381, March. Record on the lead author's own university research portal, reproducing the abstract in full: on three studies suggesting that individuals see the existence and operation of cognitive and motivational biases much more in others than in themselves; on Study 1 providing evidence from three surveys that people rate themselves as less subject to various biases than the average American, classmates in a seminar, and fellow airport travelers; on data from the third survey suggesting such claims arise from the interplay among availability biases and self-enhancement motives; on participants in one follow-up study who showed the better-than-average bias insisting that their self-assessments were accurate and objective even after reading a description of how they could have been affected by the relevant bias; on participants in a final study reporting their peer's self-serving attributions regarding test performance to be biased but their own similarly self-serving attributions to be free of bias; and on the relevance of these phenomena to naive realism and to conflict, misunderstanding, and dispute resolution being discussed. The same portal gives the citation count as 889 on one page and 892 on another. Note: the lead author's own university research portal, which is the best provenance short of the publisher. We obtained the abstract in full and not the paper, so we report no sample sizes or effect sizes. collaborate.princeton.edu
- Encyclopedia entry on the bias blind spot, recording that it is the cognitive bias of recognizing the impact of biases on the judgment of others while failing to see the impact of biases on one's own judgment; that the term was created by Emily Pronin of Princeton University with colleagues Daniel Lin and Lee Ross; that it is named after the visual blind spot; that most people appear to exhibit it; that initial evidence suggests it is not related to actual decision-making ability, with participants who scored better or poorer on various tasks associated with decision-making competence being no more or less likely to be higher or lower in their susceptibility; that it does, however, appear to increase susceptibility to related biases, citing Scopelliti and colleagues (2015); and confirming the 2002 citation as Personality and Social Psychology Bulletin, 28(3), 369–381, DOI 10.1177/0146167202286008. Note: an encyclopedia, not an academic source, flagged at every use. Our only source for the decision-competence null and for the claim that the blind spot increases susceptibility to related biases; we obtained the 2015 paper it cites for neither. en.wikipedia.org
- Pronin, E., & Hazel, L. (2023). Humans' Bias Blind Spot and Its Societal Significance. Current Directions in Psychological Science. Publisher record reproducing the abstract in full: on human beings having a bias blind spot, seeing bias all around but sometimes not in themselves; on this asymmetry hindering self-knowledge and fuelling interpersonal misunderstanding and conflict; on it being rooted in cognitive mechanics differentiating self- and social perception as well as in self-esteem motives; on it generalizing across social, cognitive, and behavioral biases, beginning in childhood, and appearing across cultures; on people showing a bias blind spot in high-stakes contexts including investing, medicine, human resources, and law; and on strategies for addressing the problem being described. The same record carries a reference list confirming Pronin, Lin and Ross (2002), 28(3), 369–381; Pronin and Kugler (2007); and Scopelliti, Morewedge, McCormick, Min, Lebrecht and Kassam (2015), Bias blind spot: Structure, measurement, and consequences. Note: the publisher's record for a review by the founding paper's lead author. We obtained the abstract only, so the promised strategies for addressing the problem, which are the most useful part, are not reported here. journals.sagepub.com
- Sceptics' reference site entry on the bias blind spot, recording that it was described by Emily Pronin and colleagues in 2002 as the tendency to perceive cognitive and motivational biases much more in others than in oneself; that it is a metabias, referring to a pattern of inaccurate judgment in reasoning about cognitive biases; quoting Pronin and Kugler as arguing that the blind spot involves the value that people place, and believe they should place, on introspective information relative to behavioral information when assessing bias in themselves versus others, with participants considering introspective information more than behavioral information for assessing bias in themselves but not others; and reporting two specific percentages concerning participants insisting on their objectivity and managers rating themselves above average at managing. Note: a sceptics' reference site, not an academic source, flagged at every use. Our source for the mechanism quotation. The two percentages it reports are graded D and used in no calculation in this article; it also dates the Pronin and Kugler paper to 2006 where two publisher reference lists give 2007. skepdic.com
- Publisher record for a 2025 paper on the bias blind spot and bias desirability, reproducing its abstract and introduction: on people readily claiming that others fall prey to several biases while being less likely to recognize those biases in themselves; on the blind spot being believed to emerge due to people's overreliance on introspection for assessing their biases, which is unlikely to turn up evidence of bias, while bias in others is ascribed based on their behaviors; on many biases being perceived as negative so that the blind spot may reflect a desire to see oneself in a positive light; on not all biases being necessarily undesirable, so that people may be motivated to admit to biases considered desirable; on the authors exploring this motivational account by manipulating bias desirability within and across biases; and on the blind spot being robust and emerging for common social and cognitive biases alike, across different cultures and from childhood, and being detrimental across real-world domains including criminal and legal procedures, financial and organizational practice, and conflict eruption and resolution. Note: the publisher's record. We obtained the abstract and introduction and not the findings, so the cognitive-versus-motivational question is reported as open without its answer. This source gives the 2002 paper's volume as 23 where every other source gives 28. link.springer.com
- Publisher record for a 2014 paper on claimed objectivity, carrying a reference list which confirms Pronin, E. (2009), The introspection illusion, Advances in Experimental Social Psychology, 41, 1–67; Pronin, E., Gilovich, T., and Ross, L. (2004), Objectivity in the eye of the beholder: Divergent perceptions of bias in self versus others, Psychological Review, 111, 781–799; Pronin, E., and Kugler, M. (2007), Valuing thoughts, ignoring behavior: The introspection illusion as a source of the bias blind spot, Journal of Experimental Social Psychology, 43, 565–678; and Pronin, E., Lin, D. Y., and Ross, L. (2002), Personality and Social Psychology Bulletin, 28, 369–381. Note: a publisher's reference list; citations only. Recorded as the source of a bibliographic variant: the 2007 paper's page range is given as 565 to 678, which would be a 114-page article in a journal whose papers run to about a dozen pages. journals.sagepub.com
- Academic preprint reference list confirming Pronin, E., Lin, D. Y., and Ross, L. (2002), The bias blind spot: Perceptions of bias in self versus others, Personality and Social Psychology Bulletin, 28(3), 369–381. Note: a reference list, used as a further independent confirmation of the citation including issue number. arxiv.org
This article discusses research on self-perception and is not governance, management or decision-process advice. None of the underlying papers was obtained: the 2002 paper and 2023 review are reported from abstracts, the 2007 mechanism paper through a sceptics' reference site, the 2015 measurement paper through an encyclopedia, and the 2025 paper from its introduction only. Two percentages in wide circulation are graded D and used in no calculation. All arithmetic is the authors' own and assumes independent, equally accurate judgments, which real meetings violate; the figures therefore flatter the group side of the comparison. Nothing here establishes that any reader's self-weighting is wrong.