Social Cognition
Self-Estimate Task
Also called: Dunning-Kruger effect task, Dunning-Kruger paradigm, unskilled and unaware task, self-assessment calibration task, overconfidence in self-estimates
Measure how well people judge their own results: participants answer 20 questions on logic, grammar and general knowledge, then estimate their score and their place among the study's participants.
6-10 minUntimed
Free to start: 10 free participants per account, then $0.99 per participant for academic research, or any yearly plan. Try it here first, with no account.
- Papers, 2021 to 2025
- 336
- Evidence grade
- C
- Registered clinical studies
- 0
What it measures
Measures how well people judge their own performance. Participants answer 20 four-answer questions on logic, grammar and general knowledge, then estimate how many they answered correctly and where their score falls among the other people in the study. The main score is the estimated number correct minus the actual number; when the study closes, each estimate of rank is compared with the person's actual rank among the study's participants.
Constructs
- Dunning-Kruger effect
- Metacognitive calibration
- Self-assessment
- Overconfidence
Research uses
- Metacognition research
- Self-assessment and calibration research
- Judgment and decision-making research
- Within-study group comparison
How the paradigm works
Here, participants answer 20 questions written for this task, each with four answers: 7 on logic, such as which box is lightest when the red box is heavier than the blue one and the blue heavier than the green, 7 on general knowledge, such as the capital of Australia, and 6 on grammar, written separately in each of the eight languages to test that language's own rules. Every answer cites a page that states the fact or the rule. The questions come in an order drawn from the session's seed, and so do the answers of each question. One practice question with feedback comes first; the quiz itself shows no answers. Afterwards, participants estimate how many of the 20 they answered correctly, from 0 to 20, and where their score falls among the other people taking part in the study, from 0 (at the very bottom) to 50 (exactly average) to 100 (at the very top), in an order drawn from the seed, as Kruger and Dunning counterbalanced theirs. There is no time limit.
More about the paradigm
The main score is the estimated number correct minus the actual number. When the study is completed or archived, each person's actual place is computed among the study's other participants in the same step, visit and language: the share of them who scored lower, with ties counted half. The estimate of place is compared with it, and each person is placed in a quarter of that comparison, so a study can set the estimates of each quarter beside its actual scores as Kruger and Dunning did. With fewer than three other people these study scores stay blank. The scores describe how people judge their own results on this quiz; they are research measures and say nothing about a person's ability or intelligence.
Scores
- Score Overestimatescore
- The estimated number of questions answered correctly minus the actual number, from -20 to 20, over a run with every question answered.
- Positive means the participant thought they answered more questions correctly than they did
- Correct Answerscount
- Questions answered correctly, out of 20.
- Informational
- Estimated Correct Answerscount
- The participant's estimate of how many of the 20 questions they answered correctly.
- Informational
- Estimated Place%
- Where the participant thought their score falls among the other people in the study, from 0 (at the very bottom) to 100 (at the very top).
- Informational
- Actual Place%
- The share of the study's other participants in the same step, visit and language who scored lower, with ties counted half. Computed when the study closes.
- Informational
- Place Overestimatescore
- The estimated place minus the actual place, in points from -100 to 100. Computed when the study closes.
- Positive means the participant placed their score higher among the study's participants than it was
- Score Quarterscore
- The quarter of the study's comparison the actual place falls in: 1 under 25, 2 under 50, 3 under 75 and 4 from 75. Computed when the study closes.
- Group the estimates by quarter, as Kruger and Dunning (1999) did
- People Comparedcount
- The other participants the actual place was computed among: each with their latest complete run of the same step, visit and language.
- Informational
- Logic Questions Correctcount
- Logic questions answered correctly, out of 7.
- Informational
- Knowledge Questions Correctcount
- General knowledge questions answered correctly, out of 7.
- Informational
- Grammar Questions Correctcount
- Grammar questions answered correctly, out of 6.
- Informational
- Median Answer Timems
- Median time from a question appearing to an answer being chosen, over answers at or after 100 ms.
- Informational
- Answers Under 100 mscount
- Answers chosen less than 100 ms after their question appeared.
- 0 expected
- Questions Answeredcount
- Questions answered, out of 20.
- Informational
Use and evidence
336 papers from 2021 to 2025 name this task in their title or abstract, up 107% compared with 2016 to 2020. That ranks it 94 of 711 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 101 |
| 2024 | 54 |
| 2023 | 75 |
| 2022 | 54 |
| 2021 | 52 |
| 2020 | 51 |
| 2019 | 41 |
| 2018 | 31 |
| 2017 | 25 |
| 2016 | 14 |
| 2015 | 15 |
| 2014 | 7 |
| 2013 | 14 |
| 2012 | 9 |
| 2011 | 6 |
| 2010 | 1 |
| 2009 | 3 |
| 2008 | 1 |
| 2007 | 1 |
| 2006 | 0 |
| 2005 | 2 |
| 2004 | 1 |
| 2003 | 0 |
| 2002 | 3 |
| 2001 | 1 |
| 2000 | 1 |
Use grade B
300 to 999 papers from 2021 to 2025.
Evidence grade C
At least 2 test-retest papers, 30 reliability or validity papers or 3 registered clinical studies.
- Papers on test-retest reliability
- 0
- Papers on reliability or validity
- 38
- Registered clinical studies using it as an outcome
- 0
Measured on ConductCognition
Pooled across the studies that feed our platform norms, by device group. A figure needs 50 or more people.
No pooled figure yet.
Note on the counts: many counted papers cite the effect without running a test.
How the grades work
Use: A is 1,000 or more papers from 2021 to 2025, B 300 to 999, C 100 to 299, D 30 to 99 and E fewer than 30. Evidence: A means at least 25 papers report its test-retest reliability and at least 25 registered clinical studies use it as an outcome; B, at least 8 test-retest papers, and either 5 registered clinical studies or 150 reliability or validity papers; C, at least 2 test-retest papers, 30 reliability or validity papers or 3 registered clinical studies; D, little published measurement evidence found. The counts measure use. They include every paper that names the task in its title or abstract. Grades describe the published research on this task, not ConductCognition's version of it.
Source: ConductCognition test census 0.58.0, counted 13 November 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Questions
- 20 written for this task, each with four answers: 7 on logic, such as which box is lightest from two comparisons, 7 on general knowledge, such as the capital of Australia, and 6 on grammar, written separately in each language to test its own rules. Every answer cites a page that states the fact or the rule. They are shown in the session's language.
- Order
- The questions, and the answers of each question, come in an order drawn from the session's seed. One practice question with feedback comes first; the quiz itself shows no answers, and there is no time limit.
- Estimates
- After the quiz, how many of the 20 the participant answered correctly, from 0 to 20, and where their score falls among the other people taking part in the study, from 0 (at the very bottom) to 50 (exactly average) to 100 (at the very top), in an order drawn from the seed.
- Scores
- The estimated number correct minus the actual number. When the study is completed or archived, each person's actual place among the study's other participants, the estimated place minus it and the quarter of the study the person falls in, so estimates can be set beside scores quarter by quarter. Also the number correct on each topic.
Kruger and Dunning (1999) found that the people who scored lowest overestimated their score and their place the most; this task runs their procedure online with questions of its own, as Jansen, Rafferty and Griffiths (2021) did with theirs. Part of the pattern can arise from regression to the mean and a general better-than-average tendency (Krueger and Mueller, 2002; Gignac and Zajenkowski, 2020). The comparison group is the study's own participants, never a norm, and the scores say nothing about a person's ability or intelligence.
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: self_estimate_answer from the scored run, one answer of four per question. Accuracy is not judged against chance, because low scores are part of what the test studies. Fewer than half the questions answered invalidates, and any other shortfall is flagged; a missing estimate is flagged. The same answer position on 90% of at least 12 answers is flagged. Answers under 200 ms on at least half of the answered questions invalidate. A hidden tab is flagged.
Why these rules
The test compares people's estimates of their own results with those results, and the effect it studies concerns the people who score lowest (Kruger and Dunning 1999), so a low score never invalidates and accuracy is not judged against chance. A run with fewer than half the questions answered cannot be compared with a full one. The answers appear in an order drawn for each question, so one position for nearly every answer is flagged for review, never invalidated, since a person may guess on hard questions. An answer under 200 ms is faster than a simple visual reaction, so it cannot reflect the item. Following response-time effort screening for untimed tests (Wise and Kong 2005), the session is invalid only when at least half the answers are that fast. Another tab can be used to look up an answer, so a hidden tab is flagged.
No percentile. Scores and quality flags only.
Running this test
Settings you can change
This test runs with fixed settings.
Need a different version of this test?
Devices and languages
- Untimed test: runs on desktops, laptops, tablets and phones. Participant devices
- Languages: English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese.
- Available on: Every plan, including the free plan.
- No percentile. Scores and quality flags only. How norms work
- Recruiting online? See the setup guides for Prolific, Sona Systems, CloudResearch Connect and Qualtrics.
Cite and describe
Participants completed the Self-Estimate Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Kruger J, Dunning D, 1999; Jansen RA, Rafferty AN, Griffiths TL, 2021). The main score was score overestimate (the estimated number of questions answered correctly minus the actual number, from -20 to 20, over a run with every question answered).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- 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. https://doi.org/10.1037/0022-3514.77.6.1121The procedure: 20-item tests, then estimates of one's score and of one's percentile among classmates, set beside the actual results quarter by quarter.
- Jansen, R. A., Rafferty, A. N., & Griffiths, T. L. (2021). A rational model of the Dunning-Kruger effect supports insensitivity to evidence in low performers. Nature Human Behaviour, 5(6), 756-763. https://doi.org/10.1038/s41562-021-01057-0The online version, with about 3,500 people for each test and the place estimated among the study's own participants.
- Krueger, J., & Mueller, R. A. (2002). Unskilled, unaware, or both? The better-than-average heuristic and statistical regression predict errors in estimates of own performance. Journal of Personality and Social Psychology, 82(2), 180-188. https://doi.org/10.1037/0022-3514.82.2.180Statistical regression and the better-than-average heuristic as an account of the pattern.
- Gignac, G. E., & Zajenkowski, M. (2020). The Dunning-Kruger effect is (mostly) a statistical artefact: Valid approaches to testing the hypothesis with individual differences data. Intelligence, 80, 101449. https://doi.org/10.1016/j.intell.2020.101449Tests of the hypothesis on individual data from 929 adults.
Source papers
- Jansen, Rafferty and Griffiths (2021): online, 3,515 adults answered 10.17 of 20 grammar questions correctly on average and estimated 12.49, and 3,543 adults answered 9.45 of 20 logic questions and estimated 10.86.Paradigm source, Adults
Questions
How long does the Self-Estimate Task take?
About 6 to 10 minutes, as listed in the test library.
What device does a participant need?
Untimed test: runs on desktops, laptops, tablets and phones. The browser must be Chrome, Edge, Firefox or Safari.
What scores does it produce?
Score Overestimate, Correct Answers, Estimated Correct Answers, Estimated Place, Actual Place, Place Overestimate, Score Quarter, People Compared, Logic Questions Correct, Knowledge Questions Correct, Grammar Questions Correct, Median Answer Time, Answers Under 100 ms and Questions Answered. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Self-Estimate Task used in research?
336 papers from 2021 to 2025 name it in their title or abstract, up 107% compared with 2016 to 2020. That ranks it 94 of 711 test families in the ConductCognition test census (use grade B). Source: ConductCognition test census 0.58.0, counted 13 November 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Self-Estimate Task have?
In the published research, 0 papers report its test-retest reliability and 0 registered clinical studies use it as an outcome (evidence grade C in the ConductCognition test census).
Does it report a percentile?
No. Scores and quality flags only.
Which languages can participants use?
Participants can take this test in English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese, with its words in the same language. English is on every plan, and the other languages come with yearly plans. Have a native speaker read the machine-translated text before launch.
Is this a clinical test?
No. ConductCognition is for research use only. It runs, scores and exports the test; clinical interpretation is outside the platform.
Every test: the full test list.
Run this test in your next study
Start free with 10 free participants on any test, then use participant credits or a yearly plan.