> ## Documentation Index
> Fetch the complete documentation index at: https://meta.niceshare.site/llms.txt
> Use this file to discover all available pages before exploring further.
# Illusory Superiority
> Illusory Superiority is a self-rank above the peer median on an easy trait. See the 1981 driver percentiles and why a hard skill reverses the rank.
**Category**: Effects
**Type**: Judgment bias
**Origin**: Ola Svenson, *Acta Psychologica* (1981); the easy-hard split in Justin Kruger, *Journal of Personality and Social Psychology* (1999)
**Also known as**: Better-than-average effect; above-average effect; Lake Wobegon effect
**Quick Answer** — Illusory Superiority is a comparative rank that puts the self above the typical peer on an easy or desirable trait, inside a group that cannot all sit above its own median. Svenson, in 1981, had licensed students mark a percentile in the room. Kruger, in 1999, showed that the same kind of rank falls below the midpoint when the skill is rare.
## What is Illusory Superiority?
Illusory Superiority is a self-placement above the typical member of a stated group, on a trait or a skill, when that placement cannot hold for the whole group at once.
> When people compare themselves with their peers, they focus egocentrically on their own skills and insufficiently take into account the skills of the comparison group.
That sentence is from Justin Kruger, *Journal of Personality and Social Psychology*, 1999. The above-average effect is the upward rank on a skill most people in the group can do. The below-average effect is the downward rank on a skill most of them cannot do. Illusory Superiority, on this page, is the upward case. The downward case is the limit that keeps the name from covering every self-rank.
A room of licensed drivers makes the upward case concrete. Each person marks a band from the least safe person in the room to the most safe. If the marks were a true ranking of that room, about half would sit above the middle. [Overconfidence Effect](/effects/overconfidence-effect) asks three different questions: a score, a rank against one other person, and a range. On that page a hard quiz pushes the claimed score up while the rank goes down. The rank here is a percentile written before any test is scored.
### Illusory Superiority in 3 Depths
* **Beginner**: The cue is a sense of being above the usual person. Name the group, and ask whether most people in that group can do the thing.
* **Practitioner**: On a review, write a percentile from 0 to 100 against a named peer group, and write whether the skill is common or rare. Later, score that percentile against a record.
* **Advanced**: A common skill pulls the written percentile up. A rare skill pulls it down. The pull is a light weight on the other people, rather than a measured test of the skill. [Dunning-Kruger Effect](/effects/dunning-kruger-effect) pairs a test score with the judgment of that score. This page does not re-read that test.
## Origin
Illusory Superiority, as a counted rank inside a defined group, is Svenson's 1981 driving study. The easy-hard split that limits the name is Kruger's 1999 paper.
Ola Svenson, at the Department of Psychology, University of Stockholm, published "Are we all less risky and more skillful than our fellow drivers?" in *Acta Psychologica*, volume 47, pages 143–148 (1981). The paper was accepted in March 1980. The Swedish Council for Social Science Research sponsored the work. Svenson thanks Decision Research, a branch of Perceptronics, in Eugene, Oregon.
He starts from an older assertion. Studies summarized by Näätänen and Summala (1975) had asked how safely people drove compared with a vaguely defined average driver. Around 70 to 80 percent were reported to place themselves in the safer half. Svenson writes that those reports were poorly documented and, by then, nearly impossible to obtain. His experiment replaces "drivers in general" with the people in the room.
One hundred sixty-one licensed participants took part. Eighty-one were students who answered an advertisement in the University of Oregon student paper. Their median age was 22. Forty-one judged driving skill, and forty judged safety. The two questions went to different people. In Sweden, eighty psychology students at the University of Stockholm took part. Their median age was 33. Forty-five judged skill, and thirty-five judged safety. The US questions sat among other judgment tasks. The Swedish session contained only this task.
> By definition, there is a least safe and a most safe driver in this room.
The prompt then asked for a position in the experimental group. It told the US students to leave Eugene and the United States out of the answer, and it told the Swedish students to leave Stockholm and Sweden out. The response was one of ten bands, each ten percentiles wide. Higher bands meant less risky and more skillful driving.
In the US group, 88 percent placed themselves safer than the median driver, and 93 percent placed themselves more skillful than that median. In the Swedish group the shares were 77 percent and 69 percent. The median of the US safety marks falls in the 81–90 band. The median of the Swedish safety marks falls in the 71–80 band. Of the US skill sample, 46.3 percent placed themselves in the most skillful 20 percent. In the Swedish skill sample that share was 15.5 percent. The abstract says "the average driver." The counted line in the results is the median of the room.
Mark D. Alicke, in the *Journal of Personality and Social Psychology*, volume 49, pages 1621–1630 (1985), asked about traits rather than a skill with a road record. Two samples rated how desirable each adjective was, and how far a person could control it. From those ratings he kept 154 adjectives, four levels of desirability crossed with two levels of controllability. A third sample rated the self and the average college student. Self ratings, set against that average student, grew more positive as the traits grew more desirable. On highly desirable traits the gap was larger when the trait was controllable. On undesirable traits the pattern reversed. This page does not print those cell means.
Kruger's paper is "Lake Wobegon be gone! The 'below-average effect' and the egocentric nature of comparative ability judgments," volume 77, pages 221–232 (1999). He points to Garrison Keillor's 1985 Lake Wobegon, a fictional town whose children are above average. The eight-ability study used 37 Cornell undergraduates in an introductory psychology course, 8 men and 29 women, for course credit. A separate pretest of 39 people rated their own skill on each ability from 1 (very unskilled) to 10 (very skilled). Those ratings were reverse-scored, and four abilities fell on the easy side of the split. Four fell on the difficult side. The pretest and the class of 37 are separate samples. The larger repeat is Max Korbmacher, Ching (Isabelle) Kwan, and Gilad Feldman, *Judgment and Decision Making*, volume 17, pages 449–486 (January 2022). Korbmacher was at the Western Norway University of Applied Sciences in Bergen. Kwan and Feldman were in psychology at the University of Hong Kong. They recruited 756 US adults on MTurk in 2020. Sixty-five failed the preregistered inclusion criteria and were excluded, leaving 691. Of those, 240 received the original wording, 225 the easier wording, and 226 the harder wording.
## Key Points
The upward rank is a comparison that puts too little weight on the other people. [Statistical Thinking](/thinking/statistical-thinking) would score a percentile against a real ordering. The steps below stay with Kruger's abilities and with the larger repeat of that questionnaire.
Svenson's prompt pointed at the other licensed people in the session. A mark above the middle of that room is a claim about those people. The 93 percent is the share of the Oregon skill sample, 41 students. The 69 percent is the share of the Stockholm skill sample, 45 students. Each share belongs to its own session.
Kruger's easy set was using a computer mouse, driving, riding a bicycle, and saving money. The mean percentiles in the class of 37 were 58.8, 65.4, 64.0, and 61.5. Each sits above the midpoint of 50. The difficult set was telling jokes, playing chess, juggling, and computer programming. Those means were 46.5, 27.8, 26.5, and 24.8. Telling jokes was not reliably under 50. The other three difficult means sat under 50.
Kruger's account is that people anchor on their own absolute skill and adjust too little for the group. In the 2022 repeat of the original wording, 240 US adults rated each ability from 1 (very unskilled) to 10 (very skilled), for themselves and for their peers. A mixed model with a random intercept for each person put the self-rating coefficient at 7.18 percentile points per scale point, interval 6.86 to 7.50. The peer-rating coefficient was −0.42, interval −0.84 to −0.01.
The same 2022 paper kept the eight ability names and changed the sentences. Driving an automatic car in one's home country drew a mean percentile of 66.32, standard deviation 22.74, in a group of 225. Driving a manual car, in a foreign country, on the opposite side of the road, drew 40.63, standard deviation 29.28, in a group of 226. Saving 20 percent of income each month stayed above the midpoint, at 62.68, standard deviation 25.77.
## Applications
Use the percentile when a self-rank will be treated as a fact. Keep the sample numbers as the picture of the pattern. Leave them in the sample when you leave the sample.
For a class, a license, or a team review, name the people on the other side of the line before you write a number. In Oregon, 93 percent of 41 students placed their skill above the median of that session. The prompt had already said the group was the room.
A familiar skill follows the high pattern in the Cornell class: mouse 58.8, driving 65.4, bicycle 64.0, saving money 61.5. A rare skill follows the low pattern: chess 27.8, juggling 26.5, programming 24.8. Jokes, at 46.5, did not reliably fall under 50. These are written percentiles from 37 students. They are a self-rank, and the class took no performance test of those eight skills.
The 2022 driving pair shows what a vague sentence hides. Automatic, at home, 225 adults: mean 66.32. Manual, abroad, opposite side of the road, 226 adults: mean 40.63. The standard deviations are 22.74 and 29.28. If a review asks only how good someone is, add the version of the task a peer would actually face.
Svenson notes that a general road-safety message is easy to set aside when the reader believes it addresses other drivers. The Oregon and Stockholm marks are not a test of any campaign. Ask for one recent near-miss of the reader's own, then a percentile against the people in the room. [Self-Serving Bias](/effects/self-serving-bias) is a story about who caused an outcome. The rank here is the number written before that story.
## Case Study
The counted room is Svenson's 1981 session. One hundred sixty-one licensed people took part, and skill and safety were separate questions.
In Oregon, students answered an advertisement in the University of Oregon paper. The median age was 22. Forty judged safety and forty-one judged skill, among other tasks in the same session. In Stockholm, psychology students at the university did only this task. The median age was 33. Thirty-five judged safety and forty-five judged skill. The prompt said there is a least safe and a most safe driver in the room, and it ruled a national comparison out. Each person marked one of ten percentile bands.
On safety, 88 percent of the Oregon group and 77 percent of the Stockholm group placed themselves above the median driver. The median Oregon safety mark falls in the 81–90 band. The median Stockholm safety mark falls in the 71–80 band. On skill, 93 percent of the Oregon group and 69 percent of the Stockholm group placed themselves above the median. Among the Oregon skill marks, 46.3 percent sit in the top 20 percent. Among the Stockholm skill marks, 15.5 percent do. The two cities are separate samples. The two questions are separate samples.
Svenson sets that pattern next to Preston and Harris, *Journal of Applied Psychology*, volume 49, pages 284–288 (1965). They compared 50 drivers whose crashes were serious enough to require a hospital stay with 50 drivers who had no accident history and were matched on the variables the study treated as relevant. On a nine-point scale of skill, the two groups gave almost identical means, and both means sat on the side of being more skillful than the average. Police records listed 34 of the 50 hospitalized drivers as responsible. That group also had a higher frequency of previous traffic violations. Svenson does not print the two means.
The lesson is the rank inside a defined group. A room of licensed students can place itself above its own median, on safety and on skill, with no one having watched them drive. A group with a serious crash record can still give the same kind of skill rating as a matched group without that record. The 88 percent is the Oregon safety sample of 40. The 93 percent is the Oregon skill sample of 41. The median ages were 22 and 33, the US task was one task among others, and the Swedish task stood alone. A national crash rate is a different sentence from a percentile in a room.
## Boundaries and Failure Modes
Illusory Superiority is the wrong name when the skill is rare, when a percentile in a room is carried out as a national rate, and when a neighbor's statistic is moved onto this page.
The first boundary is the hard set. In the class of 37, chess, juggling, and programming sat at 27.8, 26.5, and 24.8. Telling jokes sat at 46.5 and was not a reliable below-average effect. The 2022 original-wording condition, 240 US adults recruited on MTurk in 2020, put jokes at 52.4, standard deviation 22.63, again without a reliable gap from 50, and put programming at 40.7, standard deviation 29.22. Saving 20 percent of income, in the hard-wording group of 226, stayed at 62.68. The upward name covers a common skill. A rare skill has its own direction.
The second boundary is the median, the mean, and the export. A skewed count, such as crashes, can put most people on the safe side of the mean. Svenson asked for a percentile, and the shares he prints are people above the median of the room. The same town's name was later attached to a debate about school-test reports, discussed in *American Educator* in 1988, after a booklet by John Jacob Cannell. Those reports are about published scores and norms. They are a different measurement from a student marking a driving percentile. This page does not carry the state figures.
The third boundary is the neighbors. [Optimism Bias](/effects/optimism-bias) is a belief about one's own future set against other people's futures. [Illusion of Control](/effects/illusion-of-control) is how far a person thinks they can steer a chance outcome. [Illusion of Validity](/effects/illusion-of-validity) is staying sure of cues that predict poorly. The quiz points 0.79, −0.22, 0.48, and −1.36, and the 73.1 percent hit rate, stay on Overconfidence Effect. Kruger and Dunning's 1999 tests of humor, grammar, and logic stay on Dunning-Kruger Effect. Langer's cents and the engineer-file medians stay on their own pages.
## Common Misconceptions
Three confusions travel with the driving story. A majority above the middle is treated as a logical joke. The lift is treated as true of rare skills. A classroom percentile is treated as a rate on the road.
A majority can sit above the mean when the distribution is skewed. Crashes are that kind of count. Svenson's scale was a percentile in one room, where the middle is the median. On that scale, 93 percent of 41 Oregon students placed their skill above the median of the session. That share cannot be a true ranking of those 41. It can still be the belief they marked. The ranking and the belief are different claims.
The Cornell means split by the skill. Mouse, driving, bicycle, and saving money sit above 50, at 58.8, 65.4, 64.0, and 61.5. Chess, juggling, and programming sit at 27.8, 26.5, and 24.8. Jokes, at 46.5, do not reliably join the low set. In 2022, rewriting driving from an automatic car at home to a manual car abroad moved the mean from 66.32 to 40.63. Saving 20 percent of income stayed at 62.68. The difficulty of the sentence is part of the rank.
The 88, the 77, the 93, and the 69 are shares of student samples who marked a band in a session. No one scored their driving. The 58.8 and the 65.4 are mean written percentiles from 37 Cornell students. The 66.32 and the 40.63 are means from two MTurk wordings, with 225 and 226 people, collected in 2020. The standard deviations on that pair are 22.74 and 29.28. None of these figures is a share of crashes, a national share of unsafe drivers, or a hiring error.
## Related Concepts
Illusory Superiority sits next to a three-part confidence split, a test-score pattern, a belief about one's future, and a story about causes. The links are neighbors. Each one keeps the numbers from its own study.
A score, a rank against one other person, and a range. A hard quiz pushes the claimed score up and the rank down. The percentile on this page is written before that kind of test.
A test score paired with the judgment of that score. Kruger and Dunning, 1999, used humor, grammar, and logic. Those tests stay on that page.
A belief that one's own future compares well with other people's futures. Placement against a peer in the room is a present skill rank. The future-self case is a different page.
Taking credit for success and handing failure to the situation. That is a story about causes. Illusory Superiority is the percentile written against a named group.
How far a person thinks they can steer a chance outcome. A high percentile on an easy skill is a rank. It is a separate claim from steering the outcome.
Staying sure of cues that predict poorly. Confidence that a pattern fits is a different miss from placing oneself above the peer median.
## One-Line Takeaway
Name the group and whether the skill is common, then treat the percentile as a claim you can check.
This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.