In the field of psychology, the Dunning–Kruger effect is a cognitive bias in which people assess their cognitive ability as greater than it is. It is related to the cognitive bias of illusory superiority and comes from the inability of people to recognize their lack of ability. Without the self-awareness of metacognition, people cannot objectively evaluate their competence or incompetence.
As described by social psychologists David Dunning and Justin Kruger, the cognitive bias of illusory superiority results from an internal illusion in people of low ability and from an external misperception in people of high ability; that is, "the miscalibration of the incompetent stems from an error about the self, whereas the miscalibration of the highly competent stems from an error about others."
As described by social psychologists David Dunning and Justin Kruger, the cognitive bias of illusory superiority results from an internal illusion in people of low ability and from an external misperception in people of high ability; that is, "the miscalibration of the incompetent stems from an error about the self, whereas the miscalibration of the highly competent stems from an error about others."
Original study
The
psychological phenomenon of illusory superiority was identified as a
form of cognitive bias in Kruger and Dunning's 1999 study, "Unskilled
and Unaware of It: How Difficulties in Recognizing One's Own
Incompetence Lead to Inflated Self-Assessments".
The identification derived from the cognitive bias evident in the
criminal case of McArthur Wheeler, who robbed banks while his face was
covered with lemon
juice, which he believed would make it invisible to the surveillance
cameras. This belief was based on his misunderstanding of the chemical
properties of lemon juice as an invisible ink.
Other investigations of the phenomenon, such as "Why People Fail
to Recognize Their Own Incompetence" (2003), indicate that much
incorrect self-assessment of competence derives from the person's
ignorance of a given activity's standards of performance.
Dunning and Kruger's research also indicates that training in a task,
such as solving a logic puzzle, increases people's ability to accurately
evaluate how good they are at it.
In Self-insight: Roadblocks and Detours on the Path to Knowing Thyself (2005), Dunning described the Dunning–Kruger effect as "the anosognosia
of everyday life", referring to a neurological condition in which a
disabled person either denies or seems unaware of his or her disability.
He stated: "If you're incompetent, you can't know you're
incompetent ... The skills you need to produce a right answer are
exactly the skills you need to recognize what a right answer is."
In 2011, David Dunning wrote about his observations that people
with substantial, measurable deficits in their knowledge or expertise
lack the ability to recognize those deficits and, therefore, despite
potentially making error after error, tend to think they are performing
competently when they are not: "In short, those who are incompetent, for
lack of a better term, should have little insight into their
incompetence—an assertion that has come to be known as the
Dunning–Kruger effect".
In 2014, Dunning and Helzer described how the Dunning–Kruger effect
"suggests that poor performers are not in a position to recognize the
shortcomings in their performance".
Later studies
Dunning and Kruger tested the hypotheses of the cognitive bias of illusory superiority
on undergraduate students of introductory courses in psychology by
examining the students' self-assessments of their intellectual skills in
logical reasoning (inductive, deductive, abductive),
English grammar, and personal sense of humor. After learning their
self-assessment scores, the students were asked to estimate their ranks
in the psychology class. The competent students underestimated their
class rank, and the incompetent students overestimated theirs, but the
incompetent students did not estimate their class rank as higher than
the ranks estimated by the competent group. Across four studies, the
research indicated that the study participants who scored in the bottom
quartile on tests of their sense of humor, knowledge of grammar, and
logical reasoning, overestimated their test performance and their
abilities; despite test scores that placed them in the 12th percentile,
the participants estimated they ranked in the 62nd percentile.
Moreover, competent students tended to underestimate their own
competence, because they erroneously presumed that tasks easy for them
to perform were also easy for other people to perform. Incompetent
students improved their ability to estimate their class rank correctly
after receiving minimal tutoring in the skills they previously lacked,
regardless of any objective improvement gained in said skills of
perception.
The study "Mind-Reading and Metacognition: Narcissism, not Actual
Competence, Predicts Self-estimated Ability" (2004) extended the
cognitive-bias premise of illusory superiority to test subjects' emotional sensitivity toward other people and their perceptions of other people.
The study "How Chronic Self-Views Influence (and Potentially
Mislead) Estimates of Performance" (2003) indicated a shift in the
participants' view of themselves when influenced by external cues. The
participants' knowledge of geography was tested; some tests were
intended to affect the participants' self-view positively and some were
intended to affect it negatively. The participants then were asked to
rate their performances; the participants given tests with a positive
intent reported better performance than did the participants given tests
with a negative intent.
To test Dunning and Kruger's hypotheses, "that people, at all
performance levels, are equally poor at estimating their relative
performance", the study "Skilled or Unskilled, but Still Unaware of It:
How Perceptions of Difficulty Drive Miscalibration in Relative
Comparisons" (2006) investigated three studies that manipulated the
"perceived difficulty of the tasks, and, hence, [the] participants'
beliefs about their relative standing". The investigation indicated that
when the experimental subjects were presented with moderately difficult
tasks, there was little variation among the best performers and the
worst performers in their ability to predict their performance
accurately. With more difficult tasks, the best performers were less
accurate in predicting their performance than were the worst performers.
Therefore, judges at all levels of skill are subject to similar degrees
of error in the performance of tasks.
In testing alternative explanations for the cognitive bias of
illusory superiority, the study "Why the Unskilled are Unaware: Further
Explorations of (Absent) Self-insight Among the Incompetent" (2008),
reached the same conclusions as previous studies of the Dunning–Kruger
effect: that, in contrast to high performers, "poor performers do not
learn from feedback suggesting a need to improve".
Individuals of relatively high social class are more overconfident than lower-class individuals.
Mathematical critique
The
Dunning-Kruger Effect is a statement about a particular disposition of
human behavior, but the Effect also makes quantitative assertions that
rest on mathematical arguments. In this section, we focus on the
mathematical arguments.
The Effect relies on the quantifying of paired measures
consisting of: (a) the measure of the competence that people can
demonstrate when put to the test (actual competence) and (b) the measure
of competence that people believe that they have (self-assessed
competence). Researchers express the measures either as percentages that
range from 0% to 100% or as percentile scores that are scaled either
from 0 to 1 or from 0 to 100. By convention, researchers express the
differences between the two measures as (self-assessed competence -
actual competence). In such a convention, negative numbers signify
erring toward underconfidence, positive numbers signify erring toward
overconfidence, and a value of zero expresses the value of perfectly
accurate self-assessment with no error.
Ehrlinger et al.
summarized the major assertions of the Effect that had first appeared
in the 1999 seminal paper and that continued to be supported by many
studies after 9 years of research.
"People are typically overly optimistic when evaluating the
quality of their performance on social and intellectual tasks. In
particular, poor performers grossly overestimate their performances…."
(Ehrlinger et al. 2008, p. 98).
The Effect asserts that most people are overconfident about their
actual abilities and being accurate or underconfident is atypical.
Further, the Effect asserts that the least competent people are the most
overconfident. Support for both assertions rests upon interpreting the
patterns produced from graphing the paired measures.
The most common graphical convention is the Kruger-Dunning type graph typified by Figure 1 in the seminal paper.
That paper depicted college students' accuracy in self-assessing their
competencies in humor, logical reasoning, and grammar. Later
researchers adopted that convention in the subsequent papers that
studied the Effect. Additional graphs used by other researchers who
argued for the legitimacy of the Effect include (y-x) versus (x) cross
plots (Figure 5 of Pazicni and Bauer, 2015), bar charts (Figure 3 of Bell and Volckman, 2011) and histograms (Figure 1 of Stinson and Xiaofeng, 2008).
The first two of these studies depicted college students' accuracy in
self-assessing their competence in introductory chemistry, and the third
listed depicted college students' accuracy in self-assessing their
competence in business classes.
Recent researchers who focused on the mathematical reasoning behind the Effect studied 1154 participants' accuracy to self assess
their competence in understanding the nature of science. These
researchers graphed their data in all of the varied conventions of the
earlier papers, and explained how the numerical reasoning used to argue
for the Effect are similar in all. When graphed in these established
conventions, the researchers' own data also supported the assertions of
the Effect. Had the researchers ended their study at this point, their
results would have added to the established consensus that validated the
Effect. However, their deeper analyses led them to conclude that the
numerical procedures used repeatedly in all prior work were the likely
source of misleading conclusions.
To expose the source of the misleading conclusions, the
researchers employed their own real data set of paired measures from
1154 participants, and they created a second simulated data set that
employed random numbers to simulate random guessing by an equal number
of simulated participants. The simulated data set contained only random
noise that was meaningless nonsense, without any measures of human
behavior.
The researchers
then used the simulated data set and the graphical conventions of the
behavioral scientists to produce patterns like those described as
validating the Dunning-Kruger Effect. They traced the origin of the
patterns, not to the dominant literature's claimed psychological
disposition of humans, but instead to the nature of graphing data
bounded by limits of 0 and 100 and to the process of ordering and
grouping the paired measures to create the graphs. These patterns are
mathematical artifacts that random noise devoid of any human influence
can produce. Further, they showed that the graphs used to establish the
Effect in three of the four case examples presented in the seminal
paper are the patterns characteristic of purely random noise. These
patterns are numerical artifacts that behavioral scientists and
educators seem to have interpreted as evidence for a human psychological
disposition toward overconfidence.
However, the graphic presented on the case study on humor in the seminal paper and the Numeracy researchers' real data (Figure 4 of Nuhfer et al.)
were not the patterns of purely random noise. Although the data was
noisy, that human-derived data exhibited some order that could not be
credited to random noise. The researchers credited that order to human
influence and called it the "self-assessment signal."
The researchers went on to characterize the signal and worked to
determine what human disposition it revealed. To do so, the Numeracy
researchers employed different kinds of graphics that suppress or
eliminate the noise responsible for most of the artifacts and
distortions. The authors discovered that the different graphics refuted
the assertions made for the Effect. Instead, they showed that most
people are reasonably accurate in their self-assessments. About half of
the 1154 participants in their studies accurately estimated their
performance within ±10 percentage points (ppts). Two-thirds of these
participants self-assessed their competency scores within ±15 ppts
accuracy. Only about 6% of participants studied displayed wild
overconfidence and were unable to accurately self-assess their abilities
within 30 ppts. All groups of people studied displayed that they
overestimated and underestimated their actual ability with equal
frequency. No marked tendency toward overconfidence, as predicted by the
Effect, occurs, even in the most novice groups. In 2020, with an
updated database of over 5000 participants, this relationship still held
true. The revised mathematical interpretation of data confirmed that
people typically have no pronounced tendency to overestimate their
actual proficiency.
The mean self-assessments of groups of people prove more than an
order of magnitude more accurate than do isolated individuals. In
randomly selected groups of fifty participants, 81% of individual
groups' self-assessed mean scores were within 3 ppts of that group's
actual measured mean competency score. The discovery that groups of
people are accurate in their self-assessments opens an entirely new way
to study groups of people with respect to paired measures of cognitive
competence and affective self-assessed competence. A third Numeracy
paper by these researchers
reports from a database of over 3000 participants to illuminate the
effects of privilege on different ethnic and gender groups of college
students. The paper confirms that minority groups are, on average, less
privileged and score lower in the cognitive test scores and
self-assessed confidence ratings on the instruments used in this
research. They verified that women, on average, self-assessed more
accurately than men, and they did so across all ethnic groups that had
sufficient representation in the researchers' database.
Cultural differences in self-perception
Studies
of the Dunning–Kruger effect usually have been of North Americans, but
studies of Japanese people suggest that cultural forces have a role in
the occurrence of the effect.
The study "Divergent Consequences of Success and Failure in Japan and
North America: An Investigation of Self-improving Motivations and
Malleable Selves" (2001) indicated that Japanese people tended to
underestimate their abilities, and tended to see underachievement
(failure) as an opportunity to improve their abilities at a given task,
thereby increasing their value to the social group.
Popular recognition
In 2000, Kruger and Dunning were awarded an Ig Nobel Prize in recognition of the scientific work recorded in "their modest report".
"The Dunning–Kruger Song" is part of The Incompetence Opera, a mini-opera that premiered at the Ig Nobel Prize ceremony in 2017. The mini-opera is billed as "a musical encounter with the Peter principle and the Dunning–Kruger Effect".
Journalists have often cited the Dunning–Kruger effect in
discussions of political incompetence. In 2018, the British Brexit
withdrawal deal was described by Bonnie Greer
as 'the supreme example of the Dunning–Kruger effect....Dunning-Kruger
implies that we may be in the midst of an epidemic of incompetence.' At the same time, Martie Sirois wrote that President Donald Trump was 'the Dunning Kruger effect personified.' In 2020, Otto English discussed Priti Patel's
defence of the death penalty on a TV debate in which she appeared
'completely unfazed by the inherent contradictions in her
responses....Here is the very essence of the Dunning Kruger effect in
motion.'
