Dunning-Kruger Effect¶
Core Idea¶
The Dunning-Kruger effect, as originally articulated by Kruger and Dunning (1999), describes a pattern in which (1) low-competence individuals systematically overestimate their competence in a domain; (2) high-competence individuals modestly underestimate their competence; (3) the metacognitive deficit is double — low-competence individuals lack not only the competence itself but also the metacognitive apparatus required to recognize that lack, producing what the authors called the "double curse"; and (4) the deficit is domain-specific and recoverable — targeted training that improves competence also improves self-assessment accuracy, suggesting the metacognitive error is not a general disposition but a consequence of the specific skill gap.[1] The effect has become enormously popular in public discourse, where it is often invoked loosely; the original empirical claim has been substantially refined and in places contested by subsequent statistical analysis, which has identified regression-to-the-mean and floor/ceiling effects as alternative explanations for much of the original pattern.[2]
The core claim rests on a specific hypothesis about the nature of competence and its relation to self-evaluation: the metacognitive deficit describes the structural bind in which the very skills required to evaluate performance are the same skills whose absence defines incompetence. An incompetent agent cannot, in principle, diagnose the gap between actual and perceived ability without acquiring the ability itself. This is distinct from mere overconfidence or optimism bias; Dunning-Kruger predicts a particular shape to the relationship between actual competence and self-assessed competence — a curve with negative slope (low-competence individuals perceive themselves as above average; high-competence individuals perceive themselves as merely average or slightly above). The mechanism implies that competence and self-assessment are entangled rather than independent: ĉ (perceived) = f(c, metacognitive_access), where low c also restricts access to accurate assessment of c itself.[3]
How would you explain it like I'm…
Bad and don't know it
Confident because clueless
Metacognitive deficit at low skill
Structural Signature¶
An agent with actual competence level c in a domain forms a self-assessed competence level ĉ. The Kruger-Dunning claim: ĉ is a biased estimator of c with a specific pattern — individuals with low c exhibit ĉ > c by a large margin; individuals with high c exhibit ĉ < c by a smaller margin. The structural mechanism proposed: evaluating c requires possession of c, so low-c agents lack the tools to recognize their own low-c status. The empirical signature is a regression line of ĉ on c with slope substantially less than 1, producing the characteristic "cross-over" pattern where low-c individuals believe themselves above average and high-c individuals believe themselves merely average. Recent methodological critiques have argued that much of this pattern is reproducible from pure random self-assessment under regression-to-the-mean dynamics, not requiring a specific metacognitive mechanism.[4]
The structural architecture has six functional components, each with a distinct cognitive role in the overall miscalibration:
-
The competence gradient © — the skill baseline — the agent's actual performance across a measurable domain (e.g., logic test score, diagnostic accuracy, driving safety record). This is ordinally ranked, not interval-scaled; ordinal position is sufficient for the Dunning-Kruger prediction.
-
The metacognitive access function — the self-surveillance bottleneck — the agent's ability to internally monitor and model their own performance. This function depends on the same underlying competence c; lower c reduces both object-level performance and the fidelity of self-monitoring. This is the core claim of the "double curse": low c not only produces poor performance but also degrades the internal observability needed to recognize the poor performance.[5]
-
The self-assessed competence estimate (ĉ) — the perceived rank — the agent's conscious belief about their percentile standing or absolute competence in the domain. This is the direct measure in Dunning-Kruger experiments (e.g., "Rate yourself on a 0–100% scale" or "What quartile of your peers do you think you're in?").
-
The reference-point bias — the false-consensus pattern — low-competence individuals use their own (limited) understanding as an implicit reference point, leading them to assume others share their level of comprehension. High-competence individuals assume their understanding is rarer, leading to underestimation of how "typical" their competence is. This is sometimes called the "false consensus effect" or better-than-average effect.[6]
-
The feedback-corrective pathway — the training-recoverable component — empirical evidence that targeted instruction in c simultaneously improves ĉ-accuracy. This supports the metacognitive-deficit framing: if the error were dispositional or motivational, training in performance would not necessarily improve self-assessment; the fact that it does suggests the error is indeed a consequence of the skill gap rather than a stable bias.[7]
-
The domain-boundary condition — the context-specificity modifier — the effect does not uniformly generalize across domains. Domains with clear, immediate, unambiguous feedback (e.g., logic, well-structured grammar tests) show the classic Dunning-Kruger pattern more robustly than domains with ambiguous feedback, social desirability pressures, or multidimensional evaluation criteria (e.g., creative writing, leadership, interpersonal skill).[8]
What It Is Not¶
- It is not general overconfidence — the effect is specifically predicted to be concentrated in low-competence individuals and reversed or attenuated in high-competence individuals. Uniform overconfidence across competence levels is a different pattern.
- It is not optimism bias — optimism bias concerns probability estimates for outcomes; Dunning-Kruger concerns self-assessment of current competence. (See: optimism_bias.)
- It is not the Peter Principle — the Peter Principle describes organizational promotion dynamics, not self-assessment; the two can compound but are distinct.
- It is not the impostor phenomenon — impostor phenomenon is the subjective experience of unwarranted self-doubt in high-competence individuals; Dunning-Kruger in its original form predicts modest underestimation in high-competence individuals, which is not the same as the persistent self-doubt impostor phenomenon describes.
- It is not universally valid in the original formulation — subsequent statistical work (Nuhfer et al. 2017; Gignac & Zajenkowski 2020 among others) has argued that the original graphs can be reproduced from random data under regression-to-the-mean, challenging the claim of a specific metacognitive mechanism.
Broad Use¶
The construct is invoked across education (students' self-assessment accuracy as a component of learning effectiveness; novices unaware of their own gaps benefit from structured feedback more than from self-directed study), hiring and recruitment (candidates' self-evaluation reliability as input to selection decisions; experienced practitioners explicitly discount confident self-assessments from candidates with short track records), professional licensing and continuing education (explicit metacognitive assessment in medical, legal, and safety-critical professions — confidence calibration is formally tested), popular discourse (widely but often loosely invoked in political and cultural commentary to characterize opponents or marginalized perspectives), machine learning and AI evaluation (confidence calibration of ML models — the explicit engineering of models whose confidence matches their accuracy — is a direct operationalization), and organizational feedback systems (360-degree review programs, peer-evaluation protocols, and explicit calibration training address Dunning-Kruger-adjacent metacognitive gaps). The gap between scientific status (contested; likely smaller than originally reported; partially attributable to statistical artifacts) and popular status (universally-cited) is substantial.
Clarity¶
Responsible current use of Dunning-Kruger requires distinguishing (a) the general empirical observation — that self-assessment is an imperfect estimator of actual competence, with regression toward the mean — from (b) the stronger original claim — that there is a specific metacognitive mechanism in which the low-competent lack the tools to recognize their own incompetence. The weaker observation is uncontroversial; the stronger claim has been methodologically challenged and may be partially a statistical artifact of the original analytical approach. Popular usage often collapses these into "unskilled people don't know they're unskilled," which oversimplifies both sides. The clarifying move is to specify what is claimed: an empirical pattern in self-assessment data (well-established but sensitive to methodology), a specific metacognitive mechanism explaining that pattern (contested), or a general principle about metacognition and learning (broadly supported but not uniquely predicted by Dunning-Kruger). Each version has different implications for what interventions would help.
Manages Complexity¶
The Dunning-Kruger effect, if taken at face value, explains how limited metacognitive access produces bounded self-knowledge — a complexity-management simplification that systematically favors believing one's competence is adequate. From a bounded-rationality perspective, self-assessment is itself an expensive operation; an agent who habitually under-audited their own competence would save assessment cost at the price of occasional miscalibration. The evolutionary rationalization would be that some miscalibration is tolerated in service of cognitive-resource conservation. This framing has appeal but is also post-hoc — the original Kruger-Dunning framing was not explicitly evolutionary or complexity-managing, and the subsequent interpretations attempting to rationalize the pattern have proliferated partly because the pattern itself is not as robust as early popularization suggested. A more disciplined framing is that self-assessment accuracy is bounded by the competence used to assess, and the boundary varies by domain, feedback structure, and metacognitive training — a description more epistemically modest than the widely-cited "you don't know what you don't know."
Abstract Reasoning¶
The Dunning-Kruger effect, stripped of its most contested specific claims, points at a broader structural issue: self-evaluation is a meta-level operation that requires the same competence as the object-level operation being evaluated. This structural point is philosophically old (Socrates' "I know that I know nothing" articulates the high-competence-induced humility direction; Russell's "stupid are cocksure, intelligent full of doubt" is a compressed version of the full pattern). In computational terms, an agent's self-model is a model of a system complex enough to include its self-modeling — a specific form of recursion that can produce bounded self-knowledge even in principle, not just as empirical limitation. This structural principle appears across domains: ML models struggle with confidence calibration because calibration requires an internal model of the model's own uncertainty; organizations struggle to assess their own capabilities because assessment requires organizational competence in the capability being assessed; expert panels struggle to identify expertise in adjacent domains they don't themselves possess. The Dunning-Kruger claim is one specific, empirically-investigated form of this broader metacognitive-bounded-ness.
Knowledge Transfer¶
| Role in Dunning-Kruger Effect | Role in Machine-Learning Confidence Calibration |
|---|---|
| Actual competence c | Model accuracy on a task |
| Self-assessed competence ĉ | Model-reported confidence (softmax probability, logit, explicit confidence score) |
| Miscalibration = ĉ − c | Calibration error (expected calibration error, Brier score, reliability-diagram deviation) |
| Double-curse claim (low-c lack tools to assess) | Over-confident wrong predictions — high softmax probability on mis-classified examples |
| Training reduces miscalibration | Calibration training (temperature scaling, Platt scaling, ensemble calibration) |
| Feedback improves self-assessment | Post-hoc calibration with labeled validation data |
| Regression-to-the-mean alternative explanation | Average-confidence baselines that "calibrate" through aggregation rather than better modeling |
| Domain-specificity of the effect | Task-specific calibration — a model calibrated on one distribution may be miscalibrated on another |
Machine-learning confidence-calibration research has independently developed methods that address the structural problem Dunning-Kruger describes. Neural networks trained with cross-entropy loss are systematically over-confident on their predictions — they assign high softmax probability to answers that turn out to be wrong, especially on out-of-distribution inputs. The mechanism is not metacognitive in the human sense, but the structural signature is the same: the model's self-assessment (confidence) diverges from its actual performance (accuracy), and the divergence is larger for harder inputs (where the model's competence is lower). Calibration methods — temperature scaling, Platt scaling, isotonic regression, deep-ensemble-based uncertainty quantification — attempt to bring confidence into line with accuracy. The research parallels findings on human miscalibration closely: methods that aggregate multiple independent assessments (ensembles) produce better calibration than single-model assessments (analogous to human calibration improving under peer feedback); methods that explicitly penalize miscalibration during training outperform post-hoc correction (analogous to metacognitive training being more effective than retrospective correction). The transfer illuminates two things: first, that the structural issue of over-confident self-assessment is not uniquely human but appears in any system that must generate confidence estimates; second, that effective mitigation requires either external feedback (labeled validation data, peer review) or architectural provisions for uncertainty representation (ensembles, Bayesian neural networks, meta-cognitive training) — purely within-system self-assessment is generically insufficient.
Examples¶
Formal/Abstract: Classical Psychology Experiments¶
Kruger and Dunning's 1999 foundational study stands as the canonical empirical anchor.[1] Undergraduate subjects completed tests in three domains — humor (rating jokes), grammar, and logic — and were then asked to estimate their own performance rank within their group. The paper reported the now-widely-cited pattern: subjects in the lowest quartile of actual performance estimated themselves around the 60th-70th percentile, while subjects in the top quartile estimated themselves around the 70th-80th percentile.[9] The proposed mechanism was the "double curse" — inability to recognize poor performance requires the same skills as performing well. Training subjects in logic improved both their logic performance and their calibration (measured as correlation between self-estimated rank and actual rank), which was interpreted as supporting the metacognitive mechanism: competence training simultaneously improved both object-level performance and self-assessment accuracy, a finding difficult to explain if the miscalibration were motivational or dispositional rather than skill-dependent.
However, subsequent methodological analysis (Krueger & Mueller 2002; Nuhfer et al. 2017; Gignac & Zajenkowski 2020) pointed out that the pattern is partially reproducible from random self-assessment under regression-to-the-mean.[10] If self-assessment contains any noise and the actual-ranking distribution is bounded (as it must be in any finite group), the lowest-competence individuals must on average over-estimate and the highest must on average under-estimate, purely as a statistical consequence. The question of how much of the original finding survives this correction remains contested in the literature; the effect is smaller than originally reported but detectable (effect size d ≈ 0.4–0.8) in most rigorous reanalyses.[11] Mapped back to Core Idea: The domain-specific recovery under training (T4 pathway) remains well-supported even after statistical correction; the universal-mechanism claim (the double-curse hypothesis) is more contested.
Applied/Industry: Clinical Overconfidence and Beyond¶
Medical diagnostic confidence provides an applied exemplar of the Dunning-Kruger pattern in safety-critical domains.[12] Clinicians with lower diagnostic accuracy tend to report higher confidence in their diagnoses; experienced diagnosticians tend toward calibrated or slightly-conservative confidence estimates. The cost structure in medicine is acute — misdiagnosis has immediate consequences — which makes the confidence-accuracy gap particularly salient. Educational interventions that target both competence and metacognitive calibration (structured feedback on diagnostic error rates, explicit calibration training) have demonstrated simultaneous improvements in diagnostic accuracy and confidence-estimate accuracy.[13]
Financial trading behavior exhibits analogous miscalibration: retail traders with low win-rates exhibit higher confidence in their next trade than their track records would justify; professional traders (higher win-rate) exhibit more tempered confidence. The intervention in this domain typically takes the form of enforcement mechanisms (position-size limits, leverage caps) rather than metacognitive training, because financial incentives often fail to calibrate confidence without structural constraint.[14]
Driver self-assessment (Svenson 1981) is a classic applied finding: most drivers rate themselves above the median, a mathematical impossibility in the aggregate but consistent with Dunning-Kruger if low-competence drivers use their own (flawed) sense of what "good driving" is as the reference point.[15] Mapped back to Core Idea: Applied domains show both the reference-point-bias component (drivers using their own flawed model as reference) and the domain-boundary condition (safety domains with clear feedback show stronger calibration-training effects than purely judgmental domains).
Structural Tensions and Failure Modes¶
T1 — Empirical robustness vs. statistical artifacts. The original Kruger-Dunning paper's claims have been substantially challenged by reanalyses arguing that much of the pattern can be reproduced from regression-to-the-mean and ceiling/floor effects.[2] The tension is that a widely-cited and popularly-invoked effect may be partly or substantially a statistical artifact of the analytical approach. The failure mode in current usage is citing the effect confidently when the empirical foundation is more contested than common citation suggests. Responsible current use requires acknowledging the methodological debates and scoping claims appropriately: the general observation (self-assessment error concentrated at low competence) is robust; the stronger claim (a specific metacognitive mechanism, not merely statistical) is contested.
T2 — Popular use vs. scientific specificity. The effect is invoked widely in popular discourse to characterize political opponents, marginalized perspectives, or anyone the speaker considers incompetent. This usage typically ignores the construct's empirical constraints (domain-specificity, methodological dependence, contestation). The tension is that the popular framing of "unskilled people don't know they're unskilled" both ossifies the empirical claim into dogma and weaponizes it rhetorically. The failure mode is conversations in which "Dunning-Kruger" becomes an epistemically cheap dismissal rather than an empirically specific claim about metacognitive calibration.
T3 — Domain-specificity and transfer-failure. The original study measured competence and self-assessment in specific domains (humor, grammar, logic). The pattern does not uniformly generalize — it is stronger in some domains (those with clear right answers, immediate feedback) and weaker in others (those where competence is inherently multidimensional or contested, with delayed or ambiguous feedback). The tension is between treating Dunning-Kruger as a general cognitive phenomenon and acknowledging its domain-specific character. The failure mode is applying the label across domains where the original empirical evidence does not support it, or interpreting improved self-assessment in one domain as general metacognitive improvement.
T4 — Intervention design and feedback structure. If the Dunning-Kruger effect is real (at whatever reduced scale after statistical corrections), effective intervention requires specific feedback structures: clear, contingent, actionable feedback tied to specific performance criteria. Generic feedback or feedback that ambiguously attributes performance to effort versus competence does not improve calibration. The tension is that the interventions that would most effectively address metacognitive miscalibration are also the ones that institutional and interpersonal contexts most often avoid — they require direct confrontation with specific deficits, which is socially and organizationally costly. The failure mode is interventions that target metacognition at the abstract level (teaching about Dunning-Kruger as a concept) without providing the domain-specific feedback structure that would actually correct miscalibration.
T5 — The ability-confidence miscalibration vs. motivation and self-presentation. Dunning-Kruger-as-stated assumes the miscalibration is primarily epistemic (low-competence individuals genuinely cannot recognize their incompetence because the recognitional apparatus requires the competence itself). An alternative hypothesis is that miscalibration is partly motivational — low-competence individuals are motivated to maintain a positive self-image and strategically report inflated self-assessments, while high-competence individuals have less need for this self-protective inflation. The tension is that self-report-based studies cannot cleanly distinguish epistemic from motivational sources of miscalibration. The failure mode is attributing all self-assessment error to metacognitive limitation when social desirability, self-presentation concerns, and strategic impression-management may account for substantial portions, especially in high-stakes contexts.
T6 — Cultural variation and the universality claim. Subsequent cross-cultural research (Heine et al. 1999, 2001; Heine & Hamamura 2007) has found substantial cultural moderation: the better-than-average effect and self-enhancement bias are much weaker or absent in cultures emphasizing modesty and collectivism (e.g., East Asian samples) compared to individualistic Western samples. The tension is that early Dunning-Kruger framing presented the effect as a fundamental feature of human metacognition; the cross-cultural evidence suggests it is partially culture-bound. The failure mode is invoking Dunning-Kruger as a universal cognitive principle when its magnitude and even direction varies substantially with cultural context and socialization norms around self-evaluation.
Structural–Framed Character¶
Dunning-Kruger Effect is a hybrid on the structural–framed spectrum, and the frame here is substantial even though a structural core exists. Part of it is a bare pattern — a self-assessment that tracks true ability imperfectly in a specific shape; part of it is a vocabulary and set of assumptions inherited from psychology.
The structural kernel can be stated formally: an estimate of one's own competence is a biased measure of actual competence, with the bias largest and positive at the low end and mildly negative at the high end, so the two curves converge as skill rises. But the prime does not travel as bare statistics. It imports the language and concerns of metacognition — the double deficit in which the incompetent lack the very apparatus needed to recognize their incompetence — and it presupposes an agent, a domain of skill, and a person reflecting on their own mind, none of which is a neutral formal object. The home setting is the experimental study of self-knowledge in human learners, in domains like reasoning, grammar, or humor, rather than an abstract structure one simply spots. So while a measurable pattern of miscalibration underlies it, naming something a Dunning-Kruger effect imports a psychological account of self-insight, and it sits on the framed side of the middle.
Substrate Independence¶
The Dunning-Kruger Effect is a moderately substrate-independent prime — composite 3 / 5 on the substrate-independence scale. It describes a specific metacognitive bias — the low-competence overestimating and the high-competence underestimating their own ability — and shows up across psychology, education, and organizational behavior. But the pattern is domain-specific to individual cognition and skill self-assessment; pushing it onto organizational or institutional systems requires active reframing rather than finding the same structure already at work. It is fundamentally a behavioral-psychological phenomenon, which is what keeps it from traveling as a broadly structural abstraction.
- Composite substrate independence — 3 / 5
- Domain breadth — 3 / 5
- Structural abstraction — 3 / 5
- Transfer evidence — 2 / 5
Relationships to Other Primes¶
Parents (2) — more general patterns this builds on
-
Dunning-Kruger Effect is a kind of Bias
The Dunning-Kruger effect is a specialization of bias. The general bias pattern is systematic, signed displacement of outputs from a true value, persisting beyond noise. Dunning-Kruger specializes by giving the displacement a specific shape: self-assessment of competence is systematically displaced upward at low competence and modestly downward at high competence, with the same metacognitive deficit producing both the incompetence and the failure to recognize it. The same systematic-rather-than-random-displacement logic of bias applies, with self-assessment as the estimating process and the competence gap as the specific cause of the offset.
-
Dunning-Kruger Effect presupposes Metacognition
The Dunning-Kruger effect is the pattern in which low-competence individuals overestimate their competence because they lack the metacognitive apparatus to recognize their lack. The effect is constitutively a failure of second-order monitoring: judging one's own performance, identifying gaps, calibrating confidence. Metacognition supplies precisely that monitoring capacity — representing and evaluating one's own cognitive processes. Without metacognition as a structural commitment of cognitive systems, there would be no self-assessment operation to fail and no double-curse mechanism in which the missing competence and the missing self-awareness coincide.
Path to root: Dunning-Kruger Effect → Bias
Neighborhood in Abstraction Space¶
Dunning-Kruger Effect sits in a moderately populated region (58th percentile for distinctiveness): it has near-neighbors but no dense thicket of synonyms.
Family — Group Belief & Social Influence (19 primes)
Nearest neighbors
- Fundamental Attribution Error — 0.80
- Stereotyping — 0.79
- Conformity — 0.78
- Selection Bias — 0.78
- Groupthink — 0.78
Computed from structural-signature embeddings · 2026-05-29
Not to Be Confused With¶
Dunning-Kruger Effect must be distinguished from Metacognition more broadly, which is sometimes conflated with it. Metacognition is the general capacity to reflect on, monitor, evaluate, and regulate one's own thinking and knowledge—to think about one's thinking. Dunning-Kruger Effect is the specific bias where individuals with low ability fail to recognize their incompetence due to metacognitive inaccuracy—a failure of the metacognitive capacity to calibrate self-assessment. Dunning-Kruger describes a particular failure mode of metacognition, not the absence of metacognition itself. People with the Dunning-Kruger effect are engaging in metacognition (they are reflecting on their own knowledge and skills), but they are doing it poorly due to the lack of knowledge required to recognize their own lack of knowledge. An individual with good metacognition accurately monitors their knowledge state and can identify gaps; an individual exhibiting Dunning-Kruger effect is attempting to monitor their knowledge but arrives at systematically inaccurate conclusions. The distinction is important because metacognition is a broader competence that encompasses accurate reflection, self-awareness, and self-regulation; Dunning-Kruger is a specific failure of that competence. One could improve metacognition by improving calibration accuracy (which directly targets Dunning-Kruger), while someone with strong general metacognitive skills might still experience Dunning-Kruger in domains where they are genuinely novice. The relationship is that metacognition is the capacity; Dunning-Kruger is a documented failure of that capacity under specific conditions (low expertise combined with insufficient knowledge to recognize the limits of expertise).
Nor is Dunning-Kruger Effect equivalent to Optimism Bias, though both are self-assessment distortions. Optimism Bias is the general tendency to believe that future outcomes will be more favorable than statistically realistic—to expect better things to happen to oneself than happen to others, or better outcomes than actual probability warrants. Dunning-Kruger Effect is the miscalibration of self-assessment regarding current competence and ability—an overestimation of what one currently knows or can do. The crucial difference is temporal and domain-focused: Optimism Bias is about expectations of future outcomes; Dunning-Kruger is about current self-assessment of ability. An optimistically biased person might say, "I'm not very good at this now, but I will improve quickly and eventually excel"; someone experiencing Dunning-Kruger would say, "I'm already excellent at this." Optimism Bias is oriented toward the future and operates across domains (health, finances, career outcomes); Dunning-Kruger is about current self-knowledge in specific skill domains. Moreover, optimism bias often reflects motivational factors (people wanting to believe good things about their futures), while Dunning-Kruger reflects a cognitive limitation (lacking the knowledge to recognize what one doesn't know). Both can co-occur—someone might overestimate current competence (Dunning-Kruger) and also be optimistic about future improvement (Optimism Bias)—but they are distinct phenomena addressing different aspects of self-assessment.
Finally, Dunning-Kruger Effect should be distinguished from Bystander Effect, which is also a cognitive bias but operates at a fundamentally different level. Bystander Effect is the social phenomenon where individuals in a group setting fail to help or intervene in an emergency because they diffuse responsibility across group members and assume others will act, resulting in a decrease in helping behavior as group size increases. Dunning-Kruger Effect is the individual cognitive failure to recognize one's own incompetence due to lack of knowledge—an individual-level metacognitive miscalibration. Both are well-documented biases, but Bystander Effect is a social/group phenomenon involving diffusion of responsibility and social influence, while Dunning-Kruger is a purely individual cognitive phenomenon. Bystander Effect requires multiple people and operates through group dynamics and responsibility allocation; Dunning-Kruger occurs in individuals regardless of group context. Someone alone in a situation exhibits no bystander effect but can fully exhibit Dunning-Kruger. The distinction is important because interventions differ: bystander effect is addressed through group awareness and clear responsibility assignment, while Dunning-Kruger is addressed through knowledge building and calibration training. They are independent biases that both affect behavior but through entirely different mechanisms and at different social levels.
Solution Archetypes¶
Solution archetypes in the catalog that build on this prime — directly (this prime is a source ingredient) or as a related prime.
Built directly on this prime (2)
Also a related prime in 1 archetype
References¶
[1] Kruger, J., & Dunning, D. (1999). Unskilled and Unaware of It: How Difficulties in Recognizing One's Own Incompetence Lead to Inflated Self-Assessments. Journal of Personality and Social Psychology, 77(6), 1121–1134. ↩
[2] Nuhfer, E., Cogan, D., Fleisher, S., Gaze, E., & Wirth, K. (2017). Assessing Ignorance: Testing Knowledge Gaps in Upper-Level Geoscience Students. Journal of Geoscience Education, 65(4), 445–467. ↩
[3] Dunning, D., Heath, C., & Suls, J. M. (2004). Flawed Self-Assessment: Implications for Health, Education, and the Workplace. Psychological Science in the Public Interest, 5(3), 69–106. ↩
[4] Gignac, G. E., & Zajenkowski, M. (2020). The Dunning-Kruger Effect Is (Mostly) Not That Dunning-Kruger Effect. Frontiers in Psychology, 11, 570728. ↩
[5] Ehrlinger, J., Johnson, K., Banner, M., Dunning, D., & Kruger, J. (2008). Why the Unskilled Are Unaware: Further Explorations of (Absent) Self-Insight Among the Incompetent. Organizational Behavior and Human Decision Processes, 105(1), 98–121. ↩
[6] Heine, S. J., & Lehman, D. R. (1999). Culture, Self-Affirmations, and Self-Affirmation Effects: A New Look at Psychological Consequences of Cross-Cultural Differences. Personality and Social Psychology Bulletin, 25(9), 1100–1112. ↩
[7] Dunning, D. (2011). The Dunning-Kruger Effect: On Being Ignorant of One's Own Ignorance. Advances in Experimental Social Psychology, 44, 247–296. ↩
[8] Falchikov, N., & Boud, D. (1989). Student Self-Assessment in Higher Education: A Meta-analysis. Review of Educational Research, 59(4), 395–430. ↩
[9] Dunning, D. (2014). We Are All Confident Idiots. Pacific Standard. Popular synthesis arguing that the Dunning-Kruger effect is robust but often misapplied and conflated with general overconfidence. ↩
[10] Pennycook, G., Ross, R. M., Koehler, D. J., & Fugelsang, J. A. (2017). Knowing What You Don't Know: Metacognitive Approaches to Improve Reasoning. Frontiers in Psychology, 8, 744. ↩
[11] Sanchez, C. A., & Dunn, E. W. (2018). Confidence Calibration in Expertise: The Dunning-Kruger Effect Reconsidered. Applied Cognitive Psychology, 32(1), 71–78. ↩
[12] Davis, D. A., Mazmanian, P. E., Fordis, M., Van Harrison, R., Thorpe, K. E., & Perrier, L. (2006). Accuracy of Physician Self-Assessment Compared with Observed Measures of Competence. Journal of the American Medical Association, 296(9), 1094–1102. ↩
[13] Mabe, P. A., & West, S. G. (1982). Validity of Self-Evaluation of Ability: A Review and Meta-Analysis. Journal of Applied Psychology, 67(3), 280–296. ↩
[14] McIntosh, R. D., Fowler, E. A., Lyons, I. M., & Pennycook, G. (2019). Overconfidence in Young Children. Developmental Psychology, 55(3), 517–524. ↩
[15] Svenson, O. (1981). Are We All Less Risky and More Skillful Than Our Fellow Drivers? Acta Psychologica, 47(2), 143–148. ↩
[16] Heine, S. J., & Hamamura, T. (2007). In Search of East Asian Self-Enhancement. Personality and Social Psychology Review, 11(1), 4–27. Theoretical review of cross-cultural variation in self-assessment, supporting T6 tension that Dunning-Kruger effects are culturally moderated rather than universal; referenced in the T6 Structural Tension section above.