Académique Documents
Professionnel Documents
Culture Documents
Author(s): David Dunning, Kerri Johnson, Joyce Ehrlinger and Justin Kruger
Source: Current Directions in Psychological Science, Vol. 12, No. 3 (Jun., 2003), pp. 83-87
Published by: Sage Publications, Inc. on behalf of Association for Psychological Science
Stable URL: http://www.jstor.org/stable/20182845
Accessed: 17-05-2017 02:09 UTC
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
Sage Publications, Inc., Association for Psychological Science are collaborating with JSTOR to
digitize, preserve and extend access to Current Directions in Psychological Science
This content downloaded from 134.7.34.232 on Wed, 17 May 2017 02:09:58 UTC
All use subject to http://about.jstor.org/terms
CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 83
ogy, Tobin Hall, University of Massa Clifton, R., Rochat, P., Litovsky, R., & Perris, E.
their knowledge to regulate action, chusetts, Amherst, MA 01003. (1991). Object representation guides infants'
solve problems, and achieve goals. reaching in the dark. Journal of Experimental
In G. Bremner, A. Slater, & G. But Psychology: Human Perception and Performance,
17, 323-329.
terworth (Eds.), Infant development:
Recent advances (pp. 109-135). Hood, B., Carey, S., & Prasada, S. (2000). Predict
Hove, England: Psychology Press. ing the outcomes of physical events: Two
References year-olds fail to reveal knowledge of solidity
and support. Child Development, 71,1540-1554.
Baillargeon, R. (1993). The object concept revisited: Hood, B., Cole-Davies, V., & Dias, M. (2003). Look
Acknowledgments?This research was New directions in the investigation of infants'
supported by Grant HD27714 from the ing and search measures of object knowledge
physical knowledge. In CE. Granrud (Ed.), Vi
National Institutes of Health and Re sual perception and cognition in infancy (pp. 265
in pre-school children. Developmental Psychol
search Scientist Award MH00332 from ogy, 39, 61-70.
315). Hillsdale, NJ: Erlbaum.
the National Institute of Mental Health to Baillargeon, R., Graber, M., De Vos, J., & Black, J. Mash, C, Clifton, R.K., & Berthier, N.E. (2002,
Rachel K. Clifton (now Rachel Keen). I am (1990). Why do young infants fail to search for April). Two-year-olds' event reasoning and
grateful to Neil Berthier, my collaborator hidden objects? Cognition, 36, 225-284. object search. In L. Santos (Chair), Interpreting
in all of these studies, and to the other col Baillargeon, R., Spelke, E., & Wasserman, S. (1985). dissociations between infant looking and reaching:
laborators who contributed to various Object permanence in five-month-old infants. A comparative approach. Symposium conducted
Cognition, 20,191-208. at the meeting of the International Society on
phases of this work. Infant Studies, Toronto, Ontario, Canada.
Berthier, N.E., DeBlois, S., Poirier, C.R., Novak,
J.A., & Clifton, R.K. (2000). Where's the ball?
Mash, C, Keen, R., & Berthier, N.E. (in press). Vi
Two- and three-year-olds reason about unseen sual access and attention in two-year-olds'
Note events. Developmental Psychology, 36, 394-401.
event reasoning and object search. Infancy.
Butler, S.C, Berthier, N.E., & Clifton, R.K. (2002).
Two-year-olds' search strategies and visual Spelke, E.S., Breinlinger, K., Macomber, J., & Ja
1. Address correspondence to tracking in a hidden displacement task. Develj cobson, K. (1992). Origins of knowledge. Psy
Rachel Keen, Department of Psychol opmental Psychology, 38, 581-590. chological Review, 99, 605-632.
This content downloaded from 134.7.34.232 on Wed, 17 May 2017 02:09:58 UTC
All use subject to http://about.jstor.org/terms
84 VOLUME 12, NUMBER 3, JUNE 2003
This content downloaded from 134.7.34.232 on Wed, 17 May 2017 02:09:58 UTC
All use subject to http://about.jstor.org/terms
CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 85
cial domains, the skills needed to have the skills needed to avoid poor good sense of how well they per
produce correct responses are vir performance in the first place. They form in absolute terms, such as
tually identical to those needed to would no longer be incompetent. their raw score on a test (see Fig. 2).
evaluate the accuracy of one's re Despite this paradox, we de Where they err is in their estimates
sponses. The skills needed to pro cided to put this hypothesis to the of other people?consistently over
duce logically sound arguments, for test (Kruger & Dunning, 1999, estimating how well other people
instance, are the same skills that Study 4). In a first phase of the are doing on the same test (Fussell
are necessary to recognize when a study, participants were tested on & Krauss, 1992). As a result, they
logically sound argument has been their ability to solve a certain type tend to underestimate how their
made. Thus, if people lack the of logic problem. Not surprisingly, performance compares with that of
skills to produce correct answers, poor performers grossly overesti others. One can disabuse top per
they are also cursed with an inabil mated their performance on the formers of this misperception by
ity to know when their answers, or test. Then, in a second phase, we showing them the responses of
anyone else's, are right or wrong. gave roughly half of the partici other people. They then tend to re
They cannot recognize their re pants a mini-lecture about how to alize how unique and distinctive
sponses as mistaken, or other peo solve this type of logic problem, their performances are, providing
ple's responses as superior to their giving them the skills needed to more positive and accurate self-eval
own. In short, incompetence means distinguish accurate from inaccu uations. For example, asking people
that people cannot successfully rate answers. When given their who are particularly proficient in
complete the task of metacogni original test to look over, the par grammar to evaluate the grammar
tion, which, among its many mean ticipants who received the lecture, of others causes them to appropri
ings, refers to the ability to evalu and particularly those who were ately raise their perceptions of their
ate responses as correct or incorrect. poor performers, provided much own relative grammar skill. This
A good deal of research demon more accurate self-ratings than exercise has no effect on the self
strates that poor performers have they had originally. They judged impressions of poor performers
more difficulty with metacognitive their performance quite harshly? (Kruger & Dunning, 1999, Study 3;
judgments than their more compe and even lowered their confidence see also Hodges et al., 2001, for sim
tent peers do. Relative to students in their own general logical reason ilar findings involving interviewing
who are doing well, students doing ing ability, even though, if any skills among medical residents).2
poorly on a college exam do not as thing, the mini-lecture had strength
successfully distinguish which in ened that ability, not weakened it.
dividual questions they are getting
right from which they are getting
wrong (Sinkavich, 1995). Poor read
WHERE PERCEPTIONS OF
ers are less accurate than more able COMPETENCE COME FROM
readers in judging what they com
THE UNDUE MODESTY OF
TOP PERFORMERS The work we have summarized
prehend from a passage of text
(Maki & Berry, 1984). In our own leaves open an important mystery.
research, students unskilled in Top performers also suffer a It explains what does not happen
grammar provided less accurate burden, albeit one that differs from (i.e., people recognizing their in
"grades" of the grammatical perfor that of their less skilled counter competence), but it does not explain
mances of others than did their parts in that they tend to underesti what does. How do people arrive at
more skilled counterparts (Kruger mate their percentile rank relative the impressions, sometimes nega
& Dunning, 1999, Study 3). to the people with whom they tive but usually positive, that they
This double-curse explanation compare themselves. Their under hold of their performances?
also suggests a crucial hypothesis: If estimation is usually statistically sig In recent research, we have
poor performers are given the nificant (Ehrlinger et al., 2003; Haun identified one important source of
skills necessary to distinguish cor et al., 2000; Hodges et al., 2001; people's performance evaluations,
rect from incorrect answers, then Kruger & Dunning, 1999), al and shown that it can be a potential
they would be in a position to rec though in the case of Figure 1 it ap source of error in those evalua
ognize their own incompetence. Of pears quite small. tions. At first blush, one might
course, this hypothesis comes with This underestimation has a dif think that people judge how well
a paradox: If poor performers had ferent source than the overestima they are doing on a test by moni
the skills needed to distinguish ac tion of poor performers. Top per toring their experience with it. Are
curacy from error, they would then formers tend to have a relatively they taking a long time to provide
This content downloaded from 134.7.34.232 on Wed, 17 May 2017 02:09:58 UTC
All use subject to http://about.jstor.org/terms
86 VOLUME 12, NUMBER 3, JUNE 2003
the answers? Are they sure there selves quite negatively). Although Study 4). Perception of performance,
are no competitors to the answers participants in the two conditions not reality, influenced decisions
they give? Such an approach would took the same test, and achieved on about future activities.
be termed bottom-up, as it refers to average essentially the same score,
the specific experiences people those who thought they had taken
have with the test. an abstract reasoning test estimated
However, we have found that that they had achieved higher scores
CONCLUDING REMARKS
people's estimates of their perfor than did those who thought they had
mance arise, at least in part, from a taken a computer programming test
top-down approach. People start (Ehrlinger & Dunning, 2003). This research, combined with pre
with their preconceived beliefs The top-down nature of perfor vious work (for a review, see Fal
about their skill (e.g., "I am good at mance estimates can have important chikov & Boud, 1989), calls into
logical reasoning") and use those behavioral consequences. Women, question the ability people have to
beliefs to estimate how well they for example, tend to disproportion form accurate views of their skills
are doing on any specific test. This ately leave science careers along and expertise. But more than that,
strategy at first seems to be a good every step of the educational and it calls into question whether peo
one?people who believe they professional ladder (Seymour, ple are, or ever can be, in a position
have logical reasoning skill should 1992). We began to wonder if top to form accurate self-impressions.
have some basis for that claim?ex down influences on performance es If incompetent individuals do not
cept for one fly in the ointment. timates might contribute to this pat have the skills necessary to achieve
People's impressions of their intel tern. Starting in adolescence, women insight into their plight, how can
lectual and social skills often corre tend to rate themselves as less scien they be expected to achieve accu
late only modestly, and sometimes tifically talented than men rate rate self-views? How can anybody
not at all, with measures of their themselves (Eccles, 1987). Because be sure that he or she is not in the
actual performance (Falchikov & of this, women might start to think same position?
Boud, 1989). Indeed, and perhaps they are doing less well on specific This research also potentially ex
more important, people just tend scientific tasks than men tend to plains, in part, a mystery that peo
to hold overinflated views of their think, even when there is no gender ple regularly confront in their ev
skills that cannot be justified by difference in performance. Thinking eryday dealings. Everyone knows
their objective performance (Dun they are doing less well, women people who just seem to accept their
ning, Meyerowitz, & Holzberg, might become less enthusiastic about deficiencies, failing to work to im
1989; Weinstein, 1980). Therefore, participating in scientific activities. prove upon them. Perhaps these in
preconceived notions of skill can We put these notions to a test by dividuals "accept" their deficiencies
lead people to err in their perfor giving male and female college stu because they are unaware that they
mance estimates. dents a pop quiz on scientific rea have them. As Alfred North White
In several studies, we have shown soning. Before the quiz, the students head once observed, it is not igno
that people tend to make top-down were asked to rate themselves on rance, but ignorance of ignorance,
performance estimates that have their scientific skills, and the women that is the death of knowledge.
little to do with objective perfor rated themselves more negatively Our work suggests many differ
mance. In one study, we measured than the men did. The students' es ent avenues of follow-up, but one
participants' views of their "ab timates of their performance on the particularly important future ave
stract reasoning ability" before quiz showed the same pattern, with nue would focus on how, or whether,
they sat down and took a quiz on the women thinking that they had people can become aware of their in
logic. Later, their estimates of how done less well than the men thought, tellectual and social deficiencies.
well they had done on the test (e.g., even though there was no gender What are the domains in which peo
how many items they got right) difference in actual performance. ple naturally intuit their deficits, and
correlated just as highly with their Later, when asked if they would how do those domains differ from
preexisting self-views as with their like to participate in a science compe the ones we have studied? Are there
actual performance. In another tition for fun and prizes, the women rules of thumb that people can fol
study, we gave participants a test were more likely than the men to de low to ferret out their areas of in
that was purported to assess abstract cline the invitation. This reluctance competence? Can people ever be
reasoning ability (on which partici correlated significantly with their expected to uncover their pockets
pants rated themselves highly) or perceptions of performance on the of incompetence on their own, or
computer programming skills (on quiz, but not at all with actual perfor is outside intervention always nec
which participants rated them mance (Ehrlinger & Dunning, 2003, essary? Removing barriers to self
This content downloaded from 134.7.34.232 on Wed, 17 May 2017 02:09:58 UTC
All use subject to http://about.jstor.org/terms
CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 87
improvement may rest on answers imperfect, and that any measurement assessment in higher education: A meta-analy
sis. Review of Educational Research, 59, 395-430.
to these questions. flaw would lead perceptions of perfor
Fussell, S.R., & Krauss, R.M. (1992). Coordination
mance to correlate less than perfectly of knowledge in communication: Effects of
with objective performance. This im speakers' assumptions about what others
Recommended Reading perfect correlation would then cause know. Journal of Personality and Social Psychol
the perceptions of poor performers to ogy, 62,378-391.
Ehrlinger,J., & Dunning, D. (2003). be more positive than their objective per Haun, D.E., Zeringue, A., Leach, A., & Foley, A.
(See References) formance (see Krueger & Mueller, 2002).
(2000). Assessing the competence of specimen
processing personnel. Laboratory Medicine, 31,
Falchikov, N., & Boud, D. (1989). (See However, across several studies, we 633-637.
References) have found that statistically estimating Hodges, B., Regehr, G., & Martin, D. (2001). Diffi
Kruger, J., ? Dunning, D. (1999). (See and then correcting for imperfections culties in recognizing one's own incompetence:
References) in our measures leaves our original pat Novice physicians who are unskilled and un
aware of it. Academic Medicine, 76, S87-S89.
Metcalfe, J. (1998). Cognitive opti tern of misperception almost wholly in Krueger, J., & Mueller, R.A. (2002). Unskilled, un
mism: Self-deception or memory tact (Ehrlinger et al., 2003; Kruger & aware, or both? The contribution of social-per
based processing heuristics. Person Dunning, 2002). ceptual skills and statistical regression to self
ality and Social Psychology Review, 2, enhancement biases. Journal of Personality and
1(XM10. Social Psychology, 82,180-188.
Kruger, J., & Dunning, D. (1999). Unskilled and
unaware of it: How difficulties in recognizing
one's own incompetence lead to inflated self
Acknowledgments?The research de References assessments. Journal of Personality and Social
scribed in this report was supported fi Psychology, 77,1121-1134.
nancially by National Institute of Mental Dunning, D., Meyerowitz, J.A., & Holzberg, A.D. Kruger, J., & Dunning, D. (2002). Unskilled and
Health Grant RO? 56072, awarded to (1989). Ambiguity and self-evaluation: The unaware?but why? A reply to Krueger and
David Dunning. role of idiosyncratic trait definitions in self Mueller. Journal of Personality and Social Psy
serving assessments of ability. Journal of Per chology, 82,189-192.
sonality and Social Psychology, 57,1082-1090.
Maki, R.H., & Berry, S.L. (1984). Metacomprehen
Eccles, J.S. (1987). Gender roles and women's sion of text material. Journal of Experimental
Notes achievement-related decisions. Psychology of
Women Quarterly, 11,135-172.
Psychology: Learning, Memory, and Cognition,
10, 663-679.
Ehrlinger, J., & Dunning, D. (2003). How chronic Seymour, E. (1992). Undergraduate problems with
1. Address correspondence to David self-views influence (and potentially mislead) teaching and advising in SME majors: Explain
Dunning, Department of Psychology, estimates of performance. Journal of Personality ing gender differences in attrition rates. Journal
Uris Hall, Cornell University, Ithaca, NY and Social Psychology, 84, 5-17. of College Science Teaching, 21,284-292.
14853-7601; e-mail: dad6@cornell.edu. Ehrlinger, J., Johnson, K., Banner, M, Dunning, D., Sinkavich, F.J. (1995). Performance and metamem
& Kruger, D. (2003). Why the unskilled are un ory: Do students know what they don't know?
2. In other analyses, we have ruled aware: Further explorations of (absent) self-insight Instructional Psychology, 22, 77-87.
out another explanation for our find among the incompetent. Unpublished manu Weinstein, ND. (1980). Unrealistic optimism about
ings. It is almost a given that our mea script, Cornell University, Ithaca, NY. future life events. Journal of Personality and So
sures of objective performance were Falchikov, N., & Boud, D. (1989). Student self cial Psychology, 39, 806-820.
This content downloaded from 134.7.34.232 on Wed, 17 May 2017 02:09:58 UTC
All use subject to http://about.jstor.org/terms