Dweck - Praise For Intelligence

Copyright 1998 by the American Psychological Association, Inc.
0022-3514/98/S3.00
Journal of Personality and Social Psychology

1998, Vol. 75, No. 1, 33-52
Praise for Intelligence Can Undermine Children's

Motivation and Performance
Claudia M. Mueller and Carol S. Dweck
Columbia University
Praise for ability is commonly considered to have beneficial effects on motivation. Contrary to this
popular belief, six studies demonstrated that praise for intelligence had more negative consequences
for students' achievement motivation than praise for effort. Fifth graders praised for intelligence
were found to care more about performance goals relative to learning goals than children praised
for effort. After failure, they also displayed less task persistence, less task enjoyment, more lowability attributions, and worse task performance than children praised for effort. Finally, children
praised for intelligence described it as a fixed trait more than children praised for hard work,
who believed it to be subject to improvement. These findings have important implications for how
achievement is best encouraged, as well as for more theoretical issues, such as the potential cost of
performance goals and the socialization of contingent self-worth.
Praise for high ability is a common response to a job well

done. Whether it is on the sports field or in the classroom,
nothing seems more natural than to commemorate individuals'
achievements by applauding their abilities in some way. It is thus
unsurprising that this type of praise has been widely accepted as
a popular tool in the development and maintenance of individuals' academic achievement motivation, behaviors, and strategies
(Brophy, 1981; Koestner, Zuckerman, & Koestner, 1987;
Schunk, 1983, 1994; cf. Delin & Baumeister, 1994; Kanouse,
Gumpert, & Canavan-Gumpert, 1981).
Praise for intelligence, in particular, has been targeted as playing an important role in children's perceptions of their ability
and motivation to succeed. In one striking example, 85% of the
parents polled in a recent study believed that praising children's
ability (i.e., their intelligence) when they perform well on a
task is necessary to make them feel that they are smart
(Mueller & Dweck, 1996). Indeed, some child-care experts
claim that increasing children's beliefs that they "have the capacity" in this way will "turn on [their] 'go-power' " and help
motivate them to learn (Briggs, 1970).
In essence, one can identify a lay theory of achievement
motivation in which praise for intelligence makes children feel
smart and feeling smart, in turn, motivates learning. Thus, while
conventional wisdom for parenting may tell adults to criticize
the behavior but not the child, lest children learn to label themselves negatively (Briggs, 1970), the conventional wisdom for
praise is quite the opposite: The more we label children as
smart, the greater will be their enjoyment of and motivation for

achievement.
However, attributing children's good performance to intelligence may have an undesired impact on children's overall
achievement. Some interesting research has documented that
ability praise after success can have a variety of negative effects when it leads children to believe the praise to be insincere
(Meyer, 1992; Meyer, Mittag, & Engler, 1986) and when it
leads them to feel pressured to produce future good performance (Baumeister, Hutton, & Cairns, 1990; cf. Baumeister,
1984). Although both of these effects may describe ways in
which ability-related praise influences children, the primary
focus of this article is on the direct effects that this praise has
on children's goals and on their interpretations of subsequent
achievement.
Consequences of Praise for Ability Given

After Good Performance
Praise for ability may negatively affect children's responses
to achievement situations in two different ways. Having their
good performance linked to high intelligence may influence
children by changing their goals for achievement and by altering
the attributions that they make for their performance.
First, praise for ability or intelligence may lead children to
adopt a performance goal orientation toward their achievement
in which the documentation of high ability levels through successful performance becomes their primary motivational aim.
That is, telling children that they are smart when they perform
well may cause them to want to continue to prove that they are
intelligent by receiving high scores. Indeed, an emphasis on
grades and some types of verbal praise has been found to lead
children toward the assessment of their abilities through performance (Butler, 1987, 1988).
This focus on performance can have negative consequences
for children's affect, cognitions, and behavior (Butler, 1987;
Dweck & Leggett, 1988; Elliott & Dweck, 1988; cf. Nicholls,
Claudia M. Mueller and Carol S. Dweck, Department of Psychology,

Columbia University.
We thank David Krantz, Geraldine Downey, Walter Mischel, Diane
Ruble, Harvey Hornstein, Jeremy Goodridge, and James Shah for helpful
comments on earlier versions of this article.
Correspondence concerning this article should be addressed to Carol
S. Dweck, Department of Psychology, Columbia University, 1190 Amsterdam Avenue, New %rk, New \brk 10027. Electronic mail may be
sent to [email protected].
33
34
MUELLER AND DWECK
1984). For example, Dweck and her associates have demonstrated that children who hold performance goals are likely to
sacrifice potentially valuable learning opportunities if these opportunities hold the risk of making errors and do not ensure
immediate good performance (Elliott & Dweck, 1988). That is,
' 'being challenged'' and ' 'learning a lot" are rejected in favor of
"seeming smart" by children who subscribe to a performance
orientation (Mueller & Dweck, 1997). Furthermore, an emphasis on performance goals has been linked to vulnerability to a
maladaptive helpless response to achievement setbacks
(Dweck & Leggett, 1988; Elliott & Dweck, 1988), which is
characterized by negative affect, negative self-cognitions, and
performance impairment in the face of failure (Diener & Dweck,
1978, 1980).
Second, praise for intelligence after good performance may
also directly contribute to children's responses through the development of stable ability attributions for failure. Specifically,
evaluators who praise for ability may teach children that intelligence is a stable trait that is reflected in and can be easily read
from performance. If children carry away this lesson, they may
read low intelligence from poor performance and thus make
ability attributions not only for their successes but also for their
failures.
Whereas individuals who attribute their successes to internal
abilities or traits have been classified as being high in achievement efficacy (cf. Schunk, 1994, 1996) and motivation (cf.
Weiner, 1972, 1985; Weineretal., 1971), individuals who attribute their failures to ability have not been credited with the same
positive strivings. On the contrary, the negative motivational
consequences of attributions that ascribe failures to ability have
been well documented by researchers interested in achievement
(Bell, McCallum, Bryles, & Driesler, 1994; Covington & Omelich, 1984; Dweck, 1975; Jagacinski & Nicholls, 1984; Weiner,
Russell, & Lerman, 1979). In particular, Dweck and her associates have linked ability attributions for failure to helpless responding in the face of failure (Diener & Dweck, 1978, 1980;
Dweck & Leggett, 1988; Elliott & Dweck, 1988). In this study,
we propose that praise for their intelligence, even when it follows
a genuine success, teaches children that they can measure how
smart they are from how well they do. Therefore, if they subsequently do poorly, children may remeasure their ability from
this low performance.
process of their work and the possibilities for learning and improvement that hard work may offer. Because of this emphasis
on their efforts, children may feel able to focus on the development of their skills through the mastery of new material. In
other words, they may orient toward learning goals, which have
been associated with high achievement motivation (cf. Nicholls,
1984) as well as continued displays of persistence, enjoyment,
and good performance in the face of setbacks (Dweck & Leggett, 1988; Elliott & Dweck, 1988).
Second, children praised for their hard work may learn to
attribute their performance to effort, which can vary in amount,
rather than to a stable ability. Thus, they will interpret subsequent poor performance as indicating a temporary lapse in effort
rather than as a deficit in intelligence. Attributions that emphasize effort have been correlated with achievement motivation
(Powers, Douglas, Cool, & Gose, 1985) and positive postfailure
striving (Diener & Dweck, 1980; Dweck, 1975; Nicholls, 1976).
Thus, praise for hard work may lead children to display more
adaptive achievement behaviors after failure than children
praised for intelligence.
Previous Comparisons of Effort Feedback

and Ability Feedback
Findings from previous work that compared the effects of
effort praise and ability praise do not, at first glance, appear to
follow the proposals outlined above (Miller, Brickman, & Bolen, 1975; cf. Schunk, 1996). For instance, Schunk found that
praise for ability sometimes had a more beneficial effect on
children's achievement motivation, in particular their self-efficacy, than praise for effort. In addition, Miller et al. found that
children told that they were "very good" and had "excellent
ability'' in mathematics improved their performance more than
children told that they had worked hard.
However, these comparisons of effort praise and intelligence
praise focused mainly on the feedback's effects under conditions
of success. Whether the praise may lead to differences in children's responses to a specific failure has remained largely unexamined. In addition, previous researchers did not clearly examine the effects that praise for effort versus praise for intelligence
may have on children's achievement goals and performance
attributions.
Consequences of Praise for Effort Given

After Good Performance
The Current Research
If intelligence praise has unwanted consequences for children's achievement after failure, what type of praise might result
in resiliency to setbacks? In this research, we compared the
goals and achievement behaviors of children praised for intelligence with those of children praised for effort (i.e., hard work).
Praise for hard work was chosen as a comparison for praise for
intelligence primarily because effort is one of the fundamental
causal ascriptions for achievement outcomes (Weiner, 1972,
1985) and its attributional message was one that was predicted
to enhance achievement motivation. Praise for effort is proposed
to affect children in terms of both their goals and their
attributions.
First, effort-related praise may lead children to focus on the
The six studies described in this article were designed to

distinguish between the effects of praise for ability (i.e., intelligence) and praise for effort (i.e., hard work) on a variety of
measures under conditions of failure as well as success.
Specifically, we expected children praised for intelligence to
make more ability attributions for their failures than children
praised for hard work, whom we expected to prefer effort attributions (Studies 1, 3, 5, and 6). We also expected children
praised for intelligence to show that they had begun to see their
performance as a reflection of their ability by choosing to work
on tasks that would ensure good performance (Studies 1 to 4 ) ,
exhibiting performance-oriented behaviors such as misrepresenting their actual scores to others (Study 3), and seeking
PRAISE CAN UNDERMINE MOTIVATION
information about the scores of others over strategy information

(Studies 3 and 4 ) .
In addition, we expected children praised for intelligence to
show more negative responses, such as less persistence, less
enjoyment, and worse performance, after setbacks than children
praised for effort (Studies 1, 3, 5, and 6). Support for these
hypotheses was provided by the recent finding that kindergartenage children who received person- or trait-related feedback did
display more negative responses after setbacks (e.g., self-blame,
negative affect, and low persistence) than children who received
strategy-related feedback (Kamins & Dweck, 1997).
Further, we hypothesized that praise for intelligence may influence children's beliefs about and definitions of intelligence.
We predicted that praise that portrayed intelligence as being
measured from performance would lead children to define intelligence in terms of a fixed, internal entity, whereas we expected
that praise that emphasized effort would lead children to focus
on its malleable, motivational components (Studies 4 and 6 ) .
Indeed, the pattern of performance goals, ability attributions,
and helpless postfailure responses described above has been
associated with an entity view of intelligence as a fixed trait,
whereas learning goals, effort attributions, and mastery responses have been linked to an incremental view of intelligence
as subject to improvement (Dweck & Leggett, 1988; Mueller &
Dweck, 1997).
Several of the studies also were designed to investigate possible alternate explanations for the effects of ability and effort
praise. In Studies 2 and 4 we examined whether children's
divergent responses might be linked to their different expectations for future performance, as opposed to differences in the
meanings that they assign to performance. In Study 5 we investigated whether children's responses might be caused by differences in their expectations of evaluator judgments, as proposed
in previous work (Baumeister, 1984; Baumeister et al., 1990).
Finally, in Study 6 we examined whether children's divergent
responses might be specific to their beliefs about the nature of
the experimental task.
Study 1
As noted earlier, we hypothesized that praise directed at an
ability (in these experiments, intelligence) and praise directed
at effort or hard work, when administered after success, would
lead children to hold different goals for their achievement and
to have different responses when confronted with failure or
challenge.
We expected praise for intelligence to foster a desire in children to pursue a performance goal, leading children toward
documentation of their intelligence at the expense of learning.
On the other hand, we expected praise for effort to lead children
to prefer a learning goal that emphasizes the mastery of new
and challenging material.
In addition, we hypothesized that praise for effort and praise
for intelligence would lead children to have different responses
when confronted with difficult situations. As described earlier,
a performance goal has been linked to a helpless reaction characterized by negative self-cognitions, negative affect, challenge
avoidance after failure, and impaired performance, whereas a
learning goal has been associated with a mastery orientation
35
characterized by positive self-cognitions, positive affect, challenge-seeking behaviors, and enhanced performance (Elliott &
Dweck, 1988). Therefore, we hypothesized that children receiving intelligence praise would show less task enjoyment, less
task persistence, and worse task performance after failure than
children praised for effort.
Furthermore, the attributions that children made for their failures were expected to be strongly influenced by praise. In this
study, we expected children praised for their intelligence after
success to explain subsequent failures in terms of this ability.
On the other hand, we expected children praised for hard work
to attribute failure to a lack of effort.
This study also included a control group of children who
received praise without any attributional component. On the
basis of previous goal research (Elliott & Dweck, 1988), we
expected these children to hold learning and performance goals
in approximately equivalent numbers. Further, we expected their
postfailure reactions to fall between those of the two experimental groups.
Method
Participants. A total of 128 fifth graders (70 girls and 58 boys)
participated in this study. Forty-nine percent were from one public elementary school in a small midwestern town, and 51% were from two
public elementary schools in a large northeastern city. Children ranged
in age from 10 to 12 years; their mean age was 10.7 years (SD =
0.6). Fifty percent of the children were Caucasian, 19% were African
American, and 31% were Hispanic. Informed consent for the participation of all children in this and subsequent studies was given by parents,
teachers, and school principals.
Measures. All children were asked to work on three sets of problems, each containing 10 Standard Progressive Matrices (Raven, 1976).
Scores were based on the number of problems solved in each set.
Children's achievement goals were measured after they had worked
on the first set of (success) matrices and received feedback. The measure
was designed to contain a choice of tasks that embodied different goals
and has been used successfully in previous studies (see Dweck & Leggett, 1988).
Three of the choices represented variations of a performance goal in
that they focused on the display of ability: "problems that aren't too
hard, so I don't get many wrong," "problems that are pretty easy, so
I'll do well," and "problems that I'm pretty good at, so I can show that
I'm smart." The fourth choice, "problems that I'll learn a lot from,
even if I won't look so smart," represented a learning goal in that it
emphasized the development of ability over the display of high performance. Three performance goal selections were used to offset the potential social desirability of a learning goal (Leggett, 1986).
After they had made their selections, children were told that their
choices would be granted if there was extra time at the end of the
session, but first they were to work on the experimental tasks that had
been decided before the study began. This explanation was given so that
children selecting different options would not differ in their expectations
of the nature and difficulty of their subsequent tasks. Experimenters
remained blind to children's goal choices because children were asked
to place their response sheets, as well as all subsequent response sheets,
in an unmarked envelope without showing them to the experimenters.
After a second, difficult trial, children were asked to respond to a
series of questions that probed their desire to persist on the problems,
their enjoyment of the problems, their perceptions of the quality of their
performance, and their attributions for poor performance. Children rated
their task persistence, task enjoyment, and performance quality on a
scale from 1 (not at all) to 6 (very much). Task persistence was indexed
36
MUELLER AND DWECK
by children's responses to the question "How much would you like to

take these problems home to work on?" Task enjoyment was indexed
by children's responses to the questions "How much did you like working on the first/second set of problems?" and "How much fun were
the problems?" Finally, children's judgments about the quality of their
task performance were indexed by their responses to the question "How
well did you do on the problems overall?"
After children responded to the measures described above, their attributions for their poor performance on the second set of problems were
assessed through the use of a disk device described in previous research
on children's achievement (Diener & Dweck, 1980; Leggett, 1986; Nicholls, 1975). Specifically, the disk consisted of four superimposed circles of colored paper that could be adjusted to reveal different amounts
of each color. Each colored paper contained an attributional statement.
The four statements were chosen to represent three possible explanations
for poor performance on the progressive matrices: lack of effort ( " I
didn't work hard enough"), lack of ability ( " I ' m not good enough at
the problems" and "I'm not smart enough"), and lack of time ( " I
didn't have enough time"). Two ability attributions were included to
increase the perceived acceptability of this choice. Children were asked
to explain why they "had some trouble" and "made some mistakes"
on the problems. By exposing different amounts of each color, they were
able to choose how much weight, if any, they desired to assign to each
attributional statement. The circles were divided into 36 equal segments,
and attributions were scored by noting the proportion of the 36 segments
that was assigned to each attribution.
Finally, children were asked to assign weights to the importance of
their smartness and their hard work for their performance by coloring
in portions of a circle. This measure served as a manipulation check
that allowed us to determine whether or not children had understood
(and believed) the feedback given to them by the experimenters. Thus,
this measure indicates the believability of the experimental feedback.
Procedure. Children were seen individually by one of four female
experimenters. After being escorted from their usual classroom to an
empty classroom, they were introduced to the task, given a brief tutorial
in one strategy for problem solving, and asked to work on the first set
of 10 progressive matrices, chosen to be of moderate difficulty. They
were told that there would be a time limit on their work, and after 4
min, they were asked to stop working on the problems. At this point,
the experimenter scored their solutions, and children were given one of
three types of feedback that constituted the experimental manipulation.
All children were told that they had performed well on this problem
set: "Wow, you did very well on these problems. Ifou got [number of
problems] right. That's a really high score." No matter what their actual
score, all children were told that they had solved at least 80% of the
problems that they answered.
Some children (n = 41) were praised for their ability after the initial
positive feedback: " \ o u must be smart at these problems." Some children (n = 41) were praised for their effort after the initial positive
feedback: "You must have worked hard at these problems." The remaining children (n = 46) were in the control condition and received
no additional feedback.
After children were praised, they were asked whether they preferred
to pursue performance or learning goals as described above. Next, they
were given 4 min to work on a more difficult set of 10 progressive
matrices. After 4 min or the completion of all 10 problems, they were
informed that they had performed poorly ( " a lot worse") on them.
Children in all three groups were told that they had solved no more
than 50% of the problems that they answered.
After receiving this negative feedback, children were asked to rate
their desire to persist on the problems, their enjoyment of the problems,
the quality of their performance, and the failure attributions, as described
above. They subsequently were given 4 min to work on a third set of
progressive matrices equal to the first set in level of difficulty. This

process yielded a measure of postfailure performance.
During the debriefing given at the end of the experimental session,
all children were informed that the second problem set contained problems of increased difficulty, which were considered to be appropriate
for older, seventh-grade students. In fact, they were told that answering
even one of these difficult problems was quite an achievement for students in their grade level. Thus, they were assured of the overall high
quality of their task performance. Extensive precautions were taken to
ensure that all children left the experimental setting proud of their
performance.
Results and Discussion

In general, a series of two-way analyses of variance
(ANOV\s) were conducted to examine the effects of different
experimenters, schools, gender, and ethnicity on children's responses to the dependent measures. Only a few sporadic and
inconsistent effects were found for these variables; none of them
affected the interpretation of the study findings. These variables
were therefore not examined further and will not be discussed
in greater detail.
Five participants were excluded from analysis in Study 1
because they were able to solve only one or fewer of the first
set of problems. Children's average actual score on the first set
of problems was 5.2 (SD = 1.8) out of 10 total problems. The
average number of problems attempted was 7.9 (SD = 2.0).
Preliminary analyses were conducted in order to examine the
effect of children's task ability on their responses. Participants
were divided into three groups according to their actual scores
on the first set of problems, a rough index of their ability on
the task. Two-way ANOV\s (praise by initial task performance)
revealed no significant interactions between these factors on any
of the dependent measures. This result indicates that children's
ability on the progressive matrices did not influence or moderate
the effects of praise for intelligence versus praise for effort on
their responses. Thus, children who in fact had high ability on
the task were affected by the intelligence praise in the same
way as children who were less skilled.
Finally, preliminary analyses indicated that all children appeared to accept experimenters' explanations for their performance. A one-way ANOV\ revealed that the weights that they
assigned to smartness and hard work differed according to the
type of feedback that they were given, F(2, 116) = 15.90, p <
.001. Children praised for their intelligence (M = 8.89) considered their smartness to be significantly more important to their
performance than did children praised for their effort (M =
5.68), t(73) = 6.88,p < .001. Again, two-way ANOVAs (praise
by actual ability) revealed no significant interactions between
these factors.
Goal choice. As shown in Figure la, goal choice was
clearly affected by the content of the praise. A chi-square analysis revealed a significant difference in children's choice of
achievement goals after praise, x 2 ( 2 , N = 123) = 29.04, p <
.001. Most children who received intelligence feedback chose
performance goals (67%), whereas few who received effort
feedback preferred this type of goal (8%). Instead, those who
received effort feedback chose learning goals (92%). Children
in the control condition were divided equally between performance and learning goals. Thus, effort praise led children to
37

Study 1
Intelligence Control
Effort
Type of Praise Given
Study 2
Effort
Study 3
Effort
Figure 1. Proportion of children who selected performance (rather than learning) goals plotted as a
function of the type of praise given.
want to learn new things, while intelligence praise led children

to wish to continue looking smart.
Postfailure dependent measures. As shown in Table 1, significant differences were found in the attributions that children
made for their low performance on the second set of problems.
As described above, children's failure attributions were measured by asking them to apportion 36 segments of a wheel
among four attributional statements. One of the statements attributed poor performance to low effort, two statements attributed it to low ability, and one statement attributed it to lack of
time. The two ability statements were averaged to create the
measure of children's low-ability attributions.
One-way ANOVAs indicated two differences in the attributions that the groups made for their poor performance. Specifically, children differed in their endorsements of low effort, F(2,
120) = 8.64, p < .001, and low ability, F(2, 120) = 4.63, p
< .05, as causes of their failure. No significant group differences
were expected or found for the attribution that focused on lack
of time as an explanation for failure, and it will not be discussed
further.
Follow-up t tests revealed that children praised for effort
assigned greater weight to low effort (Af = 11.96, SD = 8.15)
than did children praised for ability (Af = 4.94, SD = 7.04),
f(75) = -4.05, p < .001. Children in the control (Af = 10.58,
SD = 8.43) and effort conditions did not differ in their loweffort attributions, r(82) = 0.76, ns. In addition, children
praised for intelligence after success attributed relatively more
of their failure to a lack of ability (Af = 16.49, SD = 11.04)
than did those praised for effort (A/ = 9.78, SD = 9.00), r(75)
= 2.92, p = .005. Children in the control (Af = 13.88, SD =
9.18) and ability conditions did not differ, f(83) = 1.19, ns.
These differences in children's use of ability and effort attributions to account for their poor performance provided support
for the contention that praise can have differential effects on
the meanings that children assign to their performance. Children
praised for intelligence appeared to learn that their performance
reflected their ability and thus attributed low performance to

low ability. Children praised for hard work, on the other hand,
did not show such a marked tendency to measure their intelligence from how well they did on the problems.
In addition, the different forms of praise were found to lead
children to display divergent responses after failure on a number
of other measures. Four one-way ANOVAs were performed on
the dependent measures of task persistence, task enjoyment, task
performance, and performance judgments. As shown in Table
2, significant differences were found between the groups for
children's ratings of the first two of these dependent measures.
First, a one-way ANO\^\ disclosed a significant difference in
postfailure task persistence, F(2,120) = 11.14,/? < .001. Children praised for intelligence were less likely to want to persist
on the problems (Af = 3.25, SD = 1.41) than children praised
for effort (Af = 4.53, SD = 1.03) and children in the control
condition (Af = 4.30, SD = 1.33). Follow-up t tests indicated
that children praised for intelligence differed in their desire to
persist from those praised for effort, f(75) = -4.50, p < .001,
as well as those in the control condition, f(83) = -3.52, p <
.005. No significant differences were noted for the persistence
of children in the effort and control conditions, r(82) = 0.84,
ns.
Another one-way ANOV\ revealed a significant difference in
children's postfailure task enjoyment, F(2, 120) = 7.73, p <
.005. Children praised for intelligence (Af = 4.11, SD = 1.02)
enjoyed the tasks less than did children praised for effort (Af
= 4.89, SD = 0.72); again, children in the control condition
(Af = 4.52, SD = .81) fell in between the other two groups.
Follow-up t tests showed that children praised for intelligence
were significantly less likely to enjoy the problems than were
children in the effort, f(81) = - 3 . 8 1 , p < .001, and control,
r(83) = -2.03, p < .05, conditions. Further, children in the
control condition were less likely to enjoy the problems than
were those praised for effort, f(82) = 2.16, p < .05.
A third one-way ANOV\ revealed significant differences be-
38
MUELLER AND DWECK
Table 1
Failure Attributions Made After Poor Performance as a
Function of Type of Praise Given
Dependent measure
Type of praise
Study 1
Low ability*
M
SD
Low effort****
M
SD
Intelligence
(n = 41)
Effort
(it = 41)
Control
(n = 46)
16.49,
11.04
9.78b
9.00
13.88,
9.18
4.94,
7.04
11-96,,
8.15
10.58b
8.43
Effort
(it = 30)
Control
(n = 29)
Study 3
Intelligence
(n = 29)
Low ability****
M
SD
Low effort****
M
SD
19.79,
7.18
7.70b
6.20
12.28C
7.43
4.07,
3.43
14.83b
7.70
7.97C
4.87
Study 5
Effort
in = 15)
Control
(it = 15)
20.94,
7.17
1.1%
9.50
12.06,,
8.06
7.13,
5.52
20.06b
11.32
10.06.
6.79
Dependent measure
Type of praise
Intelligence
(n = 16)
Effort
(it = 16)
Control
(n = 16)
16.94,
9.74
7.13k
6.48
13.31,
8.67
7.25,
5.34
20.8U
9.42
5.75.
4.92
Task persistence****
M
SD
Task enjoyment***
M
SD
Intelligence
(it = 41)
Effort
(n = 41)
Control
(n = 46)
3.25.
1.41
4.53 b
1.03
4.3C
1.33
4.11.
1.02
4.89b
0.72
4.52C
0.81
Intelligence
(n = 29)
Effort
(it = 30)
Control
(n = 29)
3.24,
0.83
5.20b
1.00
4.28C
1.29
3.86,
1.01
4.99b
0.55
4.49C
0.94
Intelligence
(n = 16)
Effort
(n = 15)
Control
(n = 15)
3.44,
1.59
4.62b
1.63
4.56b
1.26
3.92,
0.95
5.19,,
0.82
4.90b
0.95
Intelligence
(it = 16)
Effort
(n = 16)
Control
(n = 16)
3.75
1.18
4.63
1.20
4.00
1.03
3.84,
0.74
0.88
4.41 b
0.80
Study 3
Study 6
Low ability**
M
SD
Low effort****
M
SD
Table 2
Ratings of Task Persistence and Enjoyment Made After Poor
Performance as a Function of Type of Praise Given
Study 1
Intelligence
in = 16)
Low ability****
M
SD
Low effort****
M
SD
Fbllow-up t tests indicated that children praised for intelligence

differed significantly from those in the effort, f(75) = -5.93, p
< .001, and control, f(83) = -3.12, p < .005, conditions. In
addition, children praised for effort differed significantly from
those in the control condition, r(82) = 3.08, p < .005.
Children did not differ in their performance on the second,
failure trial, F(2, 120) = 0.18, ns. Their average actual score
was 1.6 (SD = 1 . 3 ) out of 10 problems, with an average of 5.8
(SD = 2.6) problems attempted.
Interestingly, a fourth one-way ANO\& revealed no significant differences in how the three groups rated their task performance after two problem sets, F ( 2 , 120) = 0.79, ns. This
Note. Within rows, means with different subscripts differ significantly.

Asterisks refer to the overall analysis of variance.
*p<.05.
**p < .01. ****/><.001.
tween the groups in their change in performance from the first

to the third problem set, F(2, 120) = 17.62, p < .001, even
though a separate analysis showed no differences in their performance on the first set of problems, F(2, 120) = 1.82, ns. As
shown in Figure 2a, scores for children receiving intelligence
feedback dropped an average of 0.92 (SD = 1.53) after failure.
Children in the effort condition, however, improved their prefailure scores by 1.21 (SD = 1.63) problems. Children in the control
condition improved their scores only very slightly (M = 0.13,
SD = 1.57). These results are particularly striking because they
demonstrate that the scores of children praised for intelligence
decreased after failure even though their increased familiarity
with the tasks should have bolstered, not weakened, their skills.
Task persistence****
M
SD
Task enjoyment****
M
SD
Study 5
Task persistence*
M
SD
Task enjoyment***
M
SD
Study 6
Task persistence
M
SD
Task enjoyment***
M
SD
Note. Within rows, means with different subscripts differ significantly.

Asterisks refer to the overall analysis of variance.
*p < .05. ***p < .005. ****p < .001.
39
Trial 1
Trial 3
Trial 1
Study 1
Trial 3
Study 3
INTELLIGENCE PRAISE
- - - -
CONTROL PRAISE
EFFORT PRAISE
6.5
Trial 1
Trial 3
Study 5
Trial 1
Trial 3
Study 6
Figure 2. Number of problems children solved before failure (Trial 1) and after failure (Trial 3). Bar
data from Study 6, different tasks were used on THals 1 and 3.
finding indicates that the effects of praise for intelligence were

not attributable to this group's harsh judgment of their performance. Instead, it lends further support to our contention that
the same failure experience took on different meanings for the
different groups.
Overall, the findings of Study 1 support our hypothesis that
children who are praised for intelligence when they succeed are
the ones least likely to attribute their performance to low effort,
a factor over which they have some amount of control. Instead,
they show the most marked preference for ability over effort
explanations among the three groups. Thus, praise for intelligence does not appear to teach children that they are smart;
rather, such praise appears to teach them to make inferences
about their ability versus their effort from how well they
perform.
Particularly interesting is the fact that no significant differ-
ences were noted between two very different samples of participants. That is, both children from a midwestern school, whose
ethnic makeup was almost entirely Caucasian American, and
children from two northeastern schools, whose ethnic makeup
was predominantly African American and Hispanic, were affected by praise for effort and praise for ability in the same
way. The similar responses of these two divergent populations
provide an internal replication for the findings of Study 1.
Study 2
The results of Study 1 supported our initial predictions that
praise for effort versus praise for intelligence after success leads
children to hold different achievement goals and to display different postfailure responses. However, Study 1 did not fully
investigate the different consequences of these types of praise
40
MUELLER AND DWECK
for children's motivation before their poor performance.

Whereas children's achievement goals were measured before
they encountered setbacks, their task enjoyment, task persistence, and performance judgments were not examined until after
they had experienced a failure.
Thus, the following question remains: Does the vulnerability
to failure caused by an emphasis on ability become evident
before children face setbacks? One aim of Study 2 was to investigate whether praise for intelligence, in the absence of failure,
has more negative motivational consequences for enjoyment or
persistence than does praise for effort.
In Study 1, we also proposed that praise for intelligence
and praise for effort can result in distinct postfailure responses
because they send children different messages about the meaning
of their performance. That is, praise for intelligence leads children to believe that performance is a direct measure of this
ability in a way that praise for hard work does not. Study 2
was designed to investigate two possible alternate explanations
for the findings of Study 1.
Given the findings of some previous research (Schunk, 1983),
children praised for their intelligence might show higher expectations for their future performance (as well as greater enjoyment and persistence) than might children praised for effort. If
they did, then their low scores on the second problem set might
lead them to be more disappointed and thus to display less
task enjoyment, less persistence, and worse performance than
children praised for effort.
Alternatively, the negative responses to challenge shown by
children praised for ability might result from their judgments
of their initial, successful, performance. Children praised for
intelligence might judge their performance more highly than
might children praised for hard work, which might lead to
greater disappointment and the more negative responses to setback described above. This explanation is not supported by the
results of Study 1, because no differences were found between
the groups' performance judgments after the second problem
set. However, it cannot be entirely discounted by the findings
of Study 1.
Method
Participants. A total of 51 fifth graders (26 girls and 25 boys) from
a public elementary school in a large northeastern city participated in
this study. Children ranged in age from 9 to 11 years; their mean age
was 10.5 years (SD = 0.5). T\vo percent of the children were Caucasian,
76% were African American, and 22% were Hispanic.
Measures and procedure. Children were randomly assigned to one
of three categories of praiseintelligence, effort, and controlso that
17 children made up each group. All children worked on the first set of
Standard Progressive Matrices (Raven, 1976) used in Study 1. Achievement goals were again measured after children had received positive
feedback.
Children were also asked to respond to questions from Study 1 that
probed their desire to persist on the problems, their enjoyment of the
problems, and their perception of their performance. However, these
measures were taken after their success on the first problem set. In
addition, expectations of future performance on the problems were measured by responses to the question "How well do you expect to do on
another set of problems like these?"

As shown in Figure lb, this study replicated the finding that
children praised for effort and those praised for intelligence
differed in the achievement goals that they chose to pursue,
X 2 (2, N = 51) = 11.3, p < .01. A total of 69% of children
praised for intelligence preferred performance to learning goals,
compared to only 12% of children praised for effort. That is,
children praised for hard work preferred learning goals (88%).
Children in the control condition endorsed performance (47%)
and learning (53%) goals in approximately equal measures.
These results are remarkably similar to those obtained in Study
1 and underscore the powerful effects that effort praise and
ability praise have on the goals that children have for their
achievement.
One-way ANO\As were conducted on children's task persistence and task enjoyment. No significant differences were found
between the intelligence, effort, and control conditions for measures of children's task persistence, F(2, 48) = 1.06, ns (M =
5.00, M = 4.76, and M = 5.29, respectively), and task enjoyment, F(2, 48) = 0.17, ns (M = 4.97, M = 5.00, and M =
4.82, respectively).
These findings are interesting in part because they do not
correspond to the findings of other researchers, who have suggested that ability praise may lead to benefits that effort praise
does not (cf. Schunk, 1996). However, because the studies varied in both the praise statements and the dependent measures
used, it is not altogether surprising that Study 2 failed to reveal
these effects.
The findings of Study 2 do, however, support the contention
of earlier work (Elliott & Dweck, 1988) that goals do not lead
to different achievement behaviors in the absence of challenge.
Praise for intelligence does not appear to lead to obvious motivational deficits immediately after a successful and well-received
task performance. Therefore, the differences in postfailure task
enjoyment, performance, and task persistence observed for children in Study 1 cannot be attributed to prefailure differences
between the groups in enjoyment and persistence.
Furthermore, the contrasting reactions of the groups to setbacks were not caused by differences in their prefailure expectations for performance. One-way ANO\As were conducted on
children's expectations of future success as well as on their
judgments of their current performance. No significant differences were noted for children's expectations, F(2, 48) = 1.01,
ns; children in the intelligence, effort, and control conditions
displayed equivalent expectations (M = 5.50, M = 5.06, and M
= 5.12, respectively). In addition, no significant differences
were found between the performance judgments for children in
the intelligence (M = 5.25), effort (M = 4.94), and control
(M = 4.53) conditions, F(2, 48) = 2.04, ns.
The differential effects of praise for intelligence and praise
for hard work on children's reactions to challenge thus are not
caused by disappointment brought on by children's different
expectations for their future performance.
The findings of Study 2, taken together with those of Study
1, continue to provide support for the notion that children's
postfailure responses are not attributable to differences in their
judgments of their performance. In Study 1, children were asked
to rate their overall performance after completing two sets of

progressive matrices and receiving both positive feedback and
negative feedback. As noted earlier, no significant differences
were found between the groups on this measure. In Study 2,
children were again asked to rate their performance, but this
time only on the first set of problems, on which they received
praise. No significant differences were found between the effort
and ability conditions. These results indicate that effort praise
and intelligence praise do not lead children to judge their performance differently. That is, children praised for intelligence do
not appear to rate their poor performance more harshly or their
good performance more highly than do children praised for
effort. This conclusion is consistent with the fact that all children
received the same degree of praise for their performance and
differed only in the attributions that they received for their
success.
The findings of Study 2, then, support the view that the experimental groups' contrasting responses to failure were likely
caused by whether children learn to read their effort or their
ability from their performance as a consequence of praise for
hard work or intelligence.
Study 3
We designed Study 3 to satisfy two main goals. The first aim
was to replicate the finding of Study 1 that praise for intelligence
and praise for hard work had different consequences for children's achievement goals and postfailure responses. The second
aim was to extend the investigation of intelligence praise and
effort praise by examining children's goals and concerns in
several new ways.
What do children care most about after failure: finding out
how to master problems better or finding out how well their
peers performed? If children are more oriented toward a learning
goal, then they should display a greater interest in acquiring
additional strategies so that they can better understand the problems (Butler, 1993; cf. Ruble & Frey, 1991). If, on the other
hand, they are more oriented toward a performance goal, they
should be more interested in knowing the scores of others in
order to assess or validate their ability (Butler, 1993; cf. Ruble &
Frey, 1991). Thus, children praised for effort, who prefer learning goals, were expected to seek information about learning.
Children praised for intelligence, who prefer performance goals,
were expected to seek information related to the performance
of others. This result would also suggest that these children
were truly performance oriented and not simply trying to retain
the experimenter's high opinions of their intelligence by choosing goals that ensured continued good performance.
Next, how do children praised for intelligence versus those
praised for effort represent their task performance to others? To
answer this question, children were asked to report their scores
confidentially to an anonymous child at another school. To the
extent that they had learned that performance reflected their
intelligence, we hypothesized that children praised for their ability would be more likely to misrepresent their performance in
a favorable direction. In contrast, to the extent that children
praised for effort did not see their performance as reflecting
negatively on fundamental aspects of the self, they should have
less reason to distort their performance. Both of these results
would suggest that intelligence praise does not simply teach
41
children that the adult experimenter is measuring them from

their performance. Rather, they would support the idea that children learn to measure their own ability from their performance.
Otherwise, why would they bother to find out the scores of
unfamiliar others and misrepresent their scores to an unidentified
child?
Method
participated in this study. Seventy-four percent were from one public
elementary school in a small midwestern town, and 26% were from two
in age from 9 to 11 years; their mean age was 10.3 years {SD = 0.5).
Seventy-four percent of the children were Caucasian, 8% were African
American, and 18% were Hispanic. Informed consent for the participation of all children was given by parents, teachers, and school principals.
Measures and procedure. In addition to the measures used in Study
1, two additional measures were added at the end of the original procedure to test the hypothesis that children praised for ability respond
differently to difficulty than do children praised for effort. A total of 29
children received intelligence praise, 30 received effort praise, and 29
made up the control group.
First, in order to determine whether children praised for ability were
more likely to misrepresent their actual performance after setbacks than
were children praised for effort, their self-reports were assessed. Participants were asked to write a description of the third set of problems for
children in another state. They were asked to state in their descriptions
how many of the third set of problems they had answered correctly. The
accuracy of their score reports was determined by subtracting the number
of problems they claimed to have solved from the actual number they
answered correctly.
In addition, postfailure information seeking was measured to test the
hypothesis that children praised for ability would be more interested in
measuring their performance than in mastering the problems and that
children praised for effort would show the opposite preference. Children
were presented with two identical folders; one contained "interesting
new strategies" for solving the problems (strategy information), and
the other contained the "average scores" of unfamiliar children (performance information). Children were asked to choose only one of the two
folders to read.

Goal choice. As shown in Figure lc, children's goal choice
was clearly affected by the content of the praise that they were
given. A chi-square analysis revealed a significant difference in
children's choice of achievement goals after praise, x 2 ( 2 , N =
88) = 6.55, p < .05. Whereas 55% of children who received
intelligence feedback chose performance goals, only 23% of
children who received effort feedback preferred these goals;
34% of children in the control condition elected to pursue performance, rather than learning, goals. Thus, again, intelligence
praise led children to wish to continue looking smart, whereas
effort praise led children to want to learn new things.
Postfailure Dependent Measures

Replication of Study 1. As shown in Table 1, one-way
ANOVAs revealed that the groups differed in their endorsements
of low effort, F(2, 85) = 27.54,p < .001, and low ability, F(2,
85) = 22.68, p < .001, as causes of their failure.
42
MUELLER AND DWECK
Children praised for hard work assigned greater weight to

low effort (M = 14.83, SD = 7.70) than did children praised
for intelligence (M = 4.07, SD = 3.43) and children in the
control condition (M = 7.97, SD = 4.87). Follow-up t tests
showed that children praised for hard work preferred effort
attributions significantly more than did children praised for ability, f(57) = -6.90, p < .001, and children in the control condition, f(56) = 4.08, p < .001. Further, children in the ability
and control conditions also differed significantly from each other
on this measure, t(56) = -3.52, p = .001.
In addition, children praised for intelligence after success
attributed relatively more of their failure to a lack of ability (M
= 19.79, SD = 7.18) than did children praised for hard work
(M = 7.70, SD = 6.20) and children in the control condition (M
= 12.28, SD = 7.43). Two-tailed t tests indicated that children
in the intelligence condition preferred low-ability attributions
significantly more than did children in the effort, t(51) = 6.93,
p < .001, and control, r(56) = 3.92, p < .001, conditions.
Children praised for hard work also differed significantly from
those in the control condition, ;(57) = -2.57, p < .05. Again,
intelligence praise for success seems to lead students to measure
the level of their ability rather than how hard they worked from
their performance.
Four one-way ANO\As were performed on the dependent
measures of task persistence, task enjoyment, task performance,
and performance judgments. As shown in Table 2, significant
differences were found between the groups for children's ratings
of two of the dependent measures. The findings were consistent
with the results of Study 1.
A one-way ANO\A revealed a significant difference in children's postfailure task persistence, F(2, 85) = 25.62, p < .001.
Follow-up t tests indicated that the persistence of children
praised for intelligence (M = 3.24, SD = 0.83) was significantly
lower than that of children praised for effort (M = 5.20, SD =
1.00), f(57) = -8.19, p < .001, and children in the control
condition (M = 4.28, SD = 1.29), t(56) = -3.65, p = .001.
Further, children in the effort condition showed a greater desire
to persist than did those in the control condition, 7(57) = 3.10,
p < .005.
A second one-way ANOVA revealed a significant difference
in the postfailure task enjoyment of the groups, F(2, 85) =
12.95, p < .001. Children praised for intelligence (M = 3.86,
SD = 1.01) enjoyed the tasks less than did children praised for
effort (M = 4.99, SD = 0.55); again, children in the control
condition (M = 4.49, SD = 0.94) fell in between the other two
groups. As indicated by t tests, children praised for intelligence
enjoyed the problems significantly less than did children praised
for effort, f(57) = -5.36, p < .001, and children in the control
condition, ?(56) = 2.48, p < .05. A significant difference was
also found between children in the effort and control conditions,
r(57) = 2.48, p < .05.
A third one-way ANOVA revealed significant differences between the groups' performance from the first to the third problem set, F(2, 85) = 6.58, p < .005, even though the children's
performance did not differ on the first set of problems, F(2,
85) = 0.28, ns. As shown in Figure 2b, intelligence-feedback
children's scores on the problems given after failure dropped
an average of 0.37 (SD = 1.42) from their scores on the first set
of problems. Children in the effort condition, however, improved
their prefailure scores by 1.23 (SD = 1.50) problems. Children

in the control group improved scores somewhat (M 0.34, SD
= 2.13). As indicated by t tests, children praised for intelligence
differed significantly from those in the effort condition, t(51)
= 4.23, p = .001, but not those in the control condition, t(56)
= 1.52, ns. In addition, there was a trend for children praised
for hard work to improve their postfailure performance more
than children in the control condition, t(57) = 1.86, p < .10.
A fourth one-way ANOVA was conducted to examine children's perceptions of their task performance. As in preceding
studies, no differences were noted on this measure, F(2, 85) =
2.70, ns.
Extension of Study 1. As shown in Figure 3, an additional
chi-square analysis revealed significant differences in children's
information-seeking behaviors after setbacks, x 2 ( 2 , N = 88) =
24.24, p < .001. Eighty-six percent of children praised for
ability chose to read information related to the performance of
others over information related to problem-solving strategies.
Only 23% of children praised for effort made the same choice,
and the rest, the majority, opted for strategy information that
might help them learn more about alternate task solutions. Sixtytwo percent of children in the control condition preferred performance information over strategy information.
Thus, children in the ability condition were more likely to
seek information that might bolster them but would not further
their learning or aid their performance should they ever be asked
to perform tasks similar to those presented here again. This
finding provides further support for the notion that, when allowed to pursue their interests, children praised for intelligence
adopt a performance orientation; their preference for informa-
i.uuE3 Study 3
j
0.75-
0.50-
o
Q.
o
0.25-
Study 4
0.00Intelligence
Control
Effort

Figure 3. Proportion of children who selected performance (rather
than strategy) information plotted as a function of the type of praise
given.
43
tion about scores rather than strategies indicates that they focus
on performance in their own right and not just in order to
document their abilities to an evaluator.
Further, the accuracy of children's self-reports of their performance on the third set of problems was measured. The difference between the number of problems that children reported
that they solved to other fifth graders and the number of problems that they actually answered correctly was examined. As
shown in Figure 4, a chi-square analysis revealed significant
group differences on this measure, x 2 ( 2 , N = 88) = 6.61, p <
.05. A full 38% of the children praised for ability misrepresented
their scores to unfamiliar children, whereas only 13% of the
children praised for effort and 14% of the children in the control
condition did so. On average, the children praised for intelligence added 0.45 point (SD = 1.22) to their actual score. All
children's misrepresentations increased their actual scores, suggesting that these inaccuracies were not attributable to random
errors in children's memory of their scores.
This finding suggests that over one third of children praised
for intelligence cared so much about their performance and how
it reflected on them that they lied about their performance to
another child rather than admitting to a lower score. The unwillingness of these children to report their true scores in this way
seems particularly striking when one considers that on this measure, children were reporting their scores to completely unfamiliar children, whom they were certain never to meet.
The tendency of children praised for intelligence to misrepresent their scores offers support for the contention that they indeed learned to measure themselves from their performance in
their own right. Because great care was taken to ensure that
"5
Intelligence
Control
Effort

Figure 4. Proportion of children who misrepresented their scores on
the experimental task plotted as a function of the type of praise given.
children believed that adults would never see their score reports,
any overstatements of their performance were unlikely to have
been for the benefit of the experimenter. Instead, it seems plausible to suggest that children praised for intelligence were led to
exaggerate their own scores by some internal judgments that
they had madethat is, the judgment that their performance
was an index of their intelligence. Children praised for hard
work, in contrast, did not necessarily share this type of selfjudgment and so would have felt less pressure to artificially
increase the reports of their scores.
Overall then, the results of this study demonstrated that praising children for intelligence after good performance can backfire
by making children highly performance-oriented and extremely
vulnerable to the effects of subsequent setbacks. Children's performance orientation was indicated by their tendency to sacrifice
learning goals and strategy information in favor of performance
goals and normative information as well as by their tendency
to misrepresent their actual scores after failure. Their vulnerability to setbacks was demonstrated by their relatively low enjoyment, persistence, and performance as well as their low-ability
attributions after failure. Praising children for effort, meanwhile,
appears to allow them to focus on the opportunities for mastery
provided by learning goals and strategy information and to avoid
the vulnerabilities (i.e., low enjoyment, persistence, and performance) that can be associated with task difficulties. Indeed,
praise for intelligence appears to teach children to measure their
ability from their performance on a task in a way that praise
for hard work does not.
Study 4
Study 4 was intended to replicate and extend the findings of
Study 2. We hypothesized that, after a successful performance,
children praised for intelligence would not differ from those
praised for effort or those in the control condition in their task
enjoyment, persistence, expectations, and performance judgments. Also, we hypothesized that children praised for intelligence would endorse a performance goal in greater numbers
than would children praised for effort, as had been found in the
three preceding reported studies.
In Study 4 we also further investigated the contrasting information-seeking behaviors exhibited by children praised for intelligence and those praised for effort in Study 3. We hypothesized that children praised for intelligence would again prefer
information about the performance of others that would allow
them to measure themselves. In this study, however, we measured children's information preferences after their initial successful performance, when they may well have expected to work
on additional problems. In this case, then, a preference for performance-related information would effectively rob children of
useful problem-solving strategies that might improve their performance on later trials. We expected that children praised for
effort would again prefer information about new strategies that
would help them master the tasks.
We also designed Study 4 to extend the findings of the first
three studies by investigating the influence that praise might
have on children's beliefs about the nature of intelligence. How
might praise for intelligence orient children toward an entity
view of intelligence? By judging intelligence exclusively from
44
MUELLER AND DWECK
a small sample of work (e.g., ""feu got a high score; you must
be smart"), an adult might imply that children's intelligence is
an internal, perhaps stable, trait that readily displays itself in
performance. That is, the feedback may convey that intelligence
is a stable dimension that can be reliably measured.
Praise for hard work, on the other hand, does not appear to
carry this message of intelligence as a fixed trait to children.
Instead, because this type of praise emphasizes effort as playing
a fundamental role in achievement, it is likely to orient children
toward the development of their abilities. This focus is consistent
with an incremental view that characterizes intelligence as
something that can be developed.
Method
Participants. A total of 51 fifth graders (29 girls and 22 boys) from
a public elementary school in a large northeastern city participated in
this study. Children ranged in age from 9 to 11 years; their mean age
was 9.9 years (SD = 0.5). Two percent of the children were Caucasian,
69% were African American, and 29% were Hispanic. Seventeen children were assigned to each feedback group.
Measures and procedure. The first part of this study was designed
to replicate Study 2. Then, children's implicit theories of intelligence
were measured in order to investigate how praise might influence children's beliefs about the fixedness of intelligence. Specifically, children
were asked to rate a statement taken from the Implicit Theory Scale
(">fou have a certain amount of intelligence and really can't do much
to change it") on a scale from one (not at all true) to six (very true)
(see Dweck, Chiu, & Hong, 1995, for a review).
Next, participants' information-seeking preferences were measured as
in Study 3. Children were asked to choose to read either a folder containing strategy-related information or one containing performance-related information. However, in this study the assessment was made after
the first (success) trial and not after a failure task.
was likely to have an adverse effect on their future problem

solving. Ironically, then, the children who are typically most
concerned with their performance were most likely to handicap
themselves by sacrificing an opportunity to gain potentially beneficial strategy information.
In addition, as shown in Figure 5, an ANOV\ revealed significant differences in the degree to which the groups endorsed
an entity theory, F(2, 47) = 4.98, p < .05. Children praised
for intelligence were more likely to rate intelligence as being
fixed (M = 4.24, SD = 1.79) than were children praised for
effort (M = 2.19, SD = 1.52); children in the control condition
fell in between children in the other two conditions (M = 3.47,
SD = 2.24). Follow-up t tests showed that children praised for
intelligence were significantly more likely to endorse an entity
theory than were children praised for effort, f(31) = 3.54, p =
.001. However, children praised for effort, f(31) = -1.92, ns,
and those praised for intelligence were not found to differ significantly from those in the control condition, /(32) = 1.10, ns.
These findings provide preliminary evidence for the contention that praise for intelligence after high performance can
lead children to believe that what is being measured is fixed
intelligence more than praise for hard work does.
Study 5
The first four studies provided substantial support for the
strong effect that different types of praise can have on children's
achievement motivation. However, they do not eliminate several
alternate explanations that may account for some of the findings.
We designed Study 5 to investigate two of these alternate

Replication of Study 2. Unlike the findings of the three preceding studies, children who received intelligence, effort, and
control praise were not found to differ in the achievement goals
that they chose to pursue, X 2 (2, N = 51) = 5.51, ns.
However, the four one-way ANO\5\s replicated the findings
of Study 2. No significant differences were found between the
groups in interest in task persistence, F(2, 47) = 1.03, ns,
task enjoyment, F(2, 47) = 0.68, ns, judgments of current
performance, F(2, 47) = 0.07, ns, and expectations of future
success, F(2, 47) = 1.41, n s.
Extension of Study 2. As shown in Figure 3, the finding
from Study 3 that children praised for different aspects of their
performance displayed different information-seeking behaviors
was also noted in this study, x 2 ( 2 , iV = 51) = 9.92, p < .01.
Seventy-six percent of children praised for intelligence chose
to read information related to the performance of others over
information related to problem-solving strategies. Only 24% of
children praised for hard work made the same choice. Fiftynine percent of children in the control condition also preferred
performance information over strategy information. In this
study, children's information-seeking preferences were assessed
after success. The preoccupation children praised for ability had
with measurement was thus even more striking than that in
Study 3, because their neglect of strategy-related information
0.0
Intelligence
Control
Effort

Figure 5. Rating of agreement with an entity theory (intelligence is
fixed) plotted as a function of the type of praise given. Error bars reflect
the standard error of the mean.

hypotheses. First, children's contrasting postfailure responses
might have been attributable to the fact that a single experimenter administered both positive feedback and negative feedback. Children praised for intelligence and children praised for
effort might have had distinct interpretations of the experimenter's comments. Children told that they "must be smart" might
have believed that the experimenter would continue to measure
their intelligence from subsequent poor performance. In this
case, the behaviors displayed by children praised for ability
might have been caused by their expectations of the experimenter's evaluations of their intelligence and not their own internal
judgments. On the other hand, children told that they "must
have worked hard" might have assumed that the experimenter
measured only their effort from performance and thus would not
have expected her to make ability assessments from subsequent
failure.
Studies have demonstrated that when children believe that an
evaluator perceives their ability to be high, they tend to feel
that this evaluator expects them to continue to perform well
(Baumeister, 1984; Baumeister et al., 1990; Meyer et al., 1979).
Although this may be true in many cases, in this research we
proposed that children reacted to their own assessments of the
meaning of their performance rather than to the expectations of
an evaluator. Study 5 tested this hypothesis by having two different experimenters administer the positive feedback and negative
feedback to children. The first experimenter introduced children
to the first experimental task and praised them for their good
performance in one of the three ways described for the preceding
studies. Then the children became acquainted with a second
experimenter, who was blind to their experimental condition,
who scored their second task, and who gave them the negative
feedback after failure.
Because the children knew that the second experimenter had
no knowledge of their initial good performance and positive
feedback, they could not believe that she would have high expectations for their future performance. The children's different
reactions to failure thus could not be attributed to their different
interpretations of an experimenter's judgments of and disappointment in their poor performance. Instead, children's responses would more likely be brought on by their own tendency
to use the message conveyed by different types of praise to
judge themselves from their performance.
A second alternate hypothesis was also tested in Study 5. It
is possible that, although the type of praise that children received
did indeed influence their achievement goals, it was the public
endorsement of these goals after a success that led children to
exhibit contrasting responses after subsequent failure. That is,
children who had openly expressed a desire for good performance might be more dejected after subsequent poor performance than children who had expressed a desire for learning.
In order to eliminate this influence from the experimental setting, children were not asked to report any goal preferences in
Study 5.
Method
participated in this study. Twenty-six percent were from one public
elementary school in a small midwestern town, and 74% were from two
45

in age from 10 to 11 years; their mean age was 10.3 years (SD = 0.5).
Twenty-four percent of the children were Caucasian American, 44% were
African American, 30% were Hispanic, and 2% were Asian American.
Children were randomly assigned to one of three categories of praise
ability, effort, and controlwith 16 children receiving ability praise,
15 receiving effort praise, and 15 receiving control praise.
Measures and procedure. In essence, Study 5 replicated the procedure of Study 1 with two different experimenters. Immediately after they
were introduced to the first experimenter and escorted to the experimental classroom, participants were given 4 min to work on the first set of
10 Standard Progressive Matrices (Raven, 1976). They then received
feedback and praise from the experimenter.
Next, children were introduced to a second experimenter and asked
to work on the second set of problems. It was made clear to the children
that the second experimenter had no knowledge of their performance on
the first set of problems. In addition, the second experimenter was blind
to the type of praise that the children had been given by the first experimenter. The first experimenter remained out of sight for the rest of the
session so that the children did not feel that she would be able to listen
to and judge their later performance.
Children were informed that they had performed poorly on the second
set of problems by the second experimenter. After this feedback, the
procedure followed the original outline: Children were asked to rate
their desire to persist on the problems, their enjoyment of the problems,
the quality of their performance, and their failure attributions, as described for Study 1. They subsequently were given 4 min to work on a
third set of progressive matrices that was equal to the first set in level
of difficulty.

First, as shown in Table 1, significant differences again were
found in the attributions that children made for their poor performance on the second set of problems; groups differed in their
endorsements of low effort, F(2, 45) = 10.79, p < .001, and
low ability, F(2, 45) = 10.50, p < .001, as causes of their
failure. Children praised for effort assigned greater weight to
low effort (M = 20.06, SD = 11.32) than did children praised
for ability (M = 7.13, SD - 5.52) and children in the control
condition (M = 10.06, SD = 6.79), who fell in between the
other two groups. Follow-up t tests revealed significant differences between children in the effort and intelligence conditions,
f (30) = - 4 . 1 1 , p = .001, as well as between those in the effort
and control conditions, t(30) = 3.03,/? < .01, but no differences
were found between those in the intelligence and control conditions, f(30) = -1.34, ns.
In addition, children who were praised for intelligence after
success attributed relatively more of their failure to a lack of
ability (M = 20.94, SD = 7.17) than did children praised for
hard work (M = 7.75, SD = 9.50) and children in the control
condition (M = 12.06, SD = 8.06), who fell in between the
other two groups. Follow-up t tests yielded significant differences between the ability attributions of children praised for
ability and children in the effort, f(30) = 4.43, p < .001, and
control, f(30) = 3.29, p < .005, conditions. Children in the
effort and control conditions did not differ from each other
significantly in their endorsement of low-ability attributions,
f(30) = -1.38, ns.
As shown in Table 2, additional significant differences once
again were found between the groups for ratings of several other
dependent measures.
46
MUELLER AND DWECK
A one-way ANO\5\ revealed a significant difference in children's postfailure task persistence, F(2, 45) = 3.16, p = .05.
Children praised for intelligence were less likely to want to
persist on the problems after setbacks (M = 3.44, SD = 1.59)
than were children praised for effort (M = 4.62, SD = 1.63);
children in the control condition (M = 4.56, SD = 1.26) closely
resembled those in the effort condition. Follow-up t tests revealed significant differences between the intelligence condition
and the effort, *(30) = -2.09, p < .05, and control, t(30) =
2.22, p < .05, conditions but no difference between the effort
and control conditions, f(30) = 0.12, ns.
A second one-way ANOV\ disclosed a significant difference
in children's task enjoyment after setbacks, F(2, 45) = 8.64, p
< .005. Children praised for intelligence (M = 3.92, SD 0.95) enjoyed the tasks less than did children praised for effort
(M = 5.19, SD = .82); again, children in the control condition
(M = 4.90, SD = .95) fell in between the other two groups.
Follow-up t tests yielded significant differences between the
intelligence condition and the effort, ?(30) = 4.07, p < .001,
and control, t(30) = -2.92, p < .01, conditions but not between
the effort and control conditions, r(30) = 0.82, ns.
A third one-way AN0\A did not initially reveal significant
differences in children's performance change between the first
and third problem sets, perhaps because of the smaller sample
size of Study 5 than of Study 1 and Study 3, F(2, 45) =
2.13, ns. However, a planned comparison revealed significant
differences in the performance of children praised for intelligence and those praised for effort from the first to the third
problem set, t(30) = -2.10, p < .05, even though children's
performance did not differ on the first set of problems, z(30)
= 0.21, ns. Intelligence-feedback children's scores on the
problems given after failure dropped an average of 0.50 (SD =
2.16) from their scores on the first set of problems (Figure
2c). Children in the effort condition, however, improved their
prefailure scores by 1.00 (SD = 1.86) problems. Children in
the control group improved their scores slightly (M = 0.38, SD
= 2.16). Two additional t tests did not yield significant differences between the performance of children in the effort and
control conditions, ?(30) = 0.88, ns, or between that of children
in the intelligence and control conditions, f(30) = 1.15, ns.
As found in several of the preceding studies, a fourth oneway ANOV\ revealed no significant differences in how the three
groups perceived their performance after two problem sets, F(2,
45) = 0.59, ns.
Overall, the results of this study support the hypothesis that
the contrasting responses to failure demonstrated by children
praised for effort and those praised for ability were not caused
by children's reactions to the judgments of a single experimenter. Children praised for intelligence after success displayed
more negative responses to subsequent failure than did children
praised for effort, even when the experimenter who administered
the failure feedback was blind to the children's experimental
condition and previous performance.
Furthermore, because children were not asked to explicitly
state their achievement goals before they worked on the second
problem set and encountered failure, their statements of these
preferences could not in any way have influenced their later
expectations or reactions. It seems increasingly likely that children's different responses to failure after praise for effort and
after praise for ability are indeed attributable to differences in

their interpretations of the meaning of their achievement, as
originally hypothesized.
Study 6
We designed Study 6 primarily to examine yet another alternate interpretation of the findings of Studies 1 through 5. Earlier,
we proposed that the two types of praise led children to interpret
their performance in markedly different ways. We hypothesized
that praise for intelligence would teach children to use their
performance on an intellectual task as a measure of intellectual
ability. In contrast, we proposed that praise for effort would
lead children to emphasize the degree to which they worked on
a task when explaining their performance.
However, an alternate explanation for the differences found in
the preceding studies is that children learned that the particular
experimental task on which they worked measured ability and
not that performance on intellectual tasks, in general, measures
ability. For instance, it is possible that children praised for ability
after working on the first set of problems believed that the
experimental task was actually designed to serve as an intelligence test. This belief could account for the heightened desire
to do well, manifested in their endorsement of performance
goals after praise. Furthermore, for these children, subsequent
poor performance on the same task would have seemed to be a
real measure of low intelligence. This interpretation could have
resulted in the decreased subsequent motivation and performance that they displayed in the preceding studies. Children
praised for hard work, meanwhile, were not likely to view the
experimental task as being diagnostic of their intelligence, perhaps allowing them to avoid experiencing decrements in motivation and achievement.
We designed this study to assess the viability of this explanation by determining whether children praised for ability did
indeed acquire a general tendency to judge their intelligence
from their performance or whether they did so only for the task
on which their intelligence was praisedthe "diagnostic task."
After children worked on the initial task and were praised for
their intelligence, they were introduced to a novel task of unknown diagnosticity on which they encountered challenge. We
then assessed children's responses to their poor performance on
this new task. Because we initially hypothesized that children
praised for ability learn to make general judgments about their
skills from their performance, we expected these children to
display reactions similar on the novel task to those documented
in the preceding studies after failure. Similarly, we expected
children praised for effort to display the resiliency to setback
that they demonstrated in the preceding studies.
An additional aim of Study 6 was to examine whether the
effect of contrasting forms of praise on children's beliefs about
the nature of intelligence noted in Study 4 would be noted not
only right after the praise, but also later on, after setbacks. In
Study 4, children's theories about the malleability of intelligence
were measured only after a successful performance. It could
reasonably be argued that children who received a positive judgment of their intelligence from an adult experimenter might have
had a greater interest in defining intelligence as fixed (thereby
claiming a permanent positive judgment of this characteristic)
than might children praised for effort. If this were true, greater
endorsement of the entity view would not be expected in the
face of poor performance, because children would no longer
have any vested interest in believing that a low measurement of
ability reflects a fixed trait.
In this study, we used two methods to measure children's
theories of intelligence after they experienced a setback. First,
as in Study 4, children rated their agreement with a statement
about the nonmalleability of intelligence. In addition, they were
asked to define intelligence in an open-ended question. Because
we hypothesized that praise does indeed convey information on
the nature of intelligence (and does not simply motivate children
to adopt a belief that serves their interests), we expected to
replicate the finding of Study 4 that children praised for intelligence agreed with a statement that described intelligence as
fixed to a greater extent than did children praised for effort. We
also wished to determine whether children praised for intelligence after success would also define the nature of intelligence
more in terms of an ability or trait than would children praised
for effort.
Method
participated in this study. Eighty-one percent were from a public elementary school in a small midwestern town, and 19% were from a public
elementary school in a large northeastern city. Children ranged in age
from 10 to 12 years; their mean age was 10.8 years (SD = 0.6). Eightyfour percent of the children were Caucasian, 8% were African American,
and 8% were Hispanic.
Measures and procedure. Much of the basic procedure outlined in
Study 1 was used in this study. However, after the first set of Standard
Progressive Matrices (Raven, 1976), children in this study were asked
by the same experimenter to work on different kinds of problems for 4
min. These problems were taken from the revised Minnesota Paper Form
Board Test (Likert & Quasha, 1970), which requires individuals to
"perceive fragmented percepts as wholes" (Lezak, 1983) and which is
quite distinct in appearance from Raven's Standard Progressive Matrices.
The first set of these problems contained extremely difficult items,
whereas the subsequent set contained items of moderate difficulty. Children's scores were again based on the number of problems that they
solved.
Participants' desire to persist on the problems, their enjoyment of the
problems, their attributions for failure, and their judgments about the
quality of their performance after setbacks were assessed as in Study 1.
In addition, children's beliefs about the nature of intelligence were
measured as in Study 4. These beliefs were also assessed when children
were asked to define intelligence by completing the sentence "I think
intelligence is . . ." Two coders who were blind to the experimental
condition the children were in categorized children's responses to this
open-ended question according to the presence (coded 1) or absence
(coded 0) of two elements: their use of terms that emphasized the more
malleable or motivational components of intelligence (e.g., effort and
knowledge) and their use of terms that emphasized the trait-like nature
of intelligence (e.g., ability and smartness).
Results and
Discussion
Postfailure dependent measures. As found in the earlier

studies and as shown in Table 1, one-way ANOVAs revealed
that children differed in the attributions that they made for their
failures. As expected, children differed significantly in their
47
effort attributions, F(2,45) = 23.38,p < .001. Children praised

for effort {M = 20.81, SD = 9.42) attributed more of their low
score to effort than did children praised for intelligence (M =
7.25, SD = 5.34); children in the control condition weighted
effort essentially the same as did children in the intelligence
condition (M = 5.75, SD = 4.92). Follow-up t tests revealed
the expected significant difference between the effort and intelligence conditions, f(30) = 5.01,p < .001, as well as a significant difference between the effort and control conditions, f(30)
= 5.67, p < .001. No significant difference existed between the
intelligence and control conditions, t(30) = 0.83, ns.
Children also were found to differ in their ability attributions,
F(2, 45) = 5.57, p < .01. Children praised for intelligence (M
= 16.94, SD = 9.74) attributed more of their poor performance
to ability than did children praised for effort (M = 7.13, SD =
6.48); children in the control condition fell in between the other
two groups (M = 13.31, SD = 8.67). Follow-up t tests showed
that children in the intelligence condition differed significantly
from those in the effort condition, f(30) = 3.36, p < .005, but
not from those in the control condition, f(30) = 1.11, ns. Children praised for effort differed significantly from those in the
control condition, t(30) = -2.29, p < .05.
In addition, as shown in Table 2, children in the three groups
differed in several other responses to failure. As in preceding
studies, no differences were found in their performance judgments, F(2, 45) = 0.35, ns.
As in Studies 1, 3, and 5, a one-way ANOVA revealed a
significant difference in children's postfailure task enjoyment,
F(2, 45) = 6.38, p < .005. Children praised for intelligence
(M = 3.84, SD = 0.74) described themselves as having enjoyed
the problems less than did children praised for effort (M =
4.86, SD = 0.88); children in the control condition (M = 4.41,
SD = 0.80) rated themselves similarly to children in the effort
condition. Follow-up t tests indicated that children in the intelligence condition differed significantly from those in the effort,
r(30) = -3.54, p = .001, and control, f(30) = -2.07, p <
.05, conditions but that children in the effort condition did not
differ from those in the control condition, t(30) = 1.53, ns.
Another one-way ANOVA revealed significant differences in
children's performance on the problems after failure, F(2, 45)
= 6.18, p < .005. For this study, postfailure performance was
measured by examining the number of problems that children
solved on the third problem set. As shown in Figure 2d, although
they did not differ in the number of problems that they solved
on the first, F(2, 45) = 0.32, ns, and second, F(2, 45) = 0.54,
ns, problem sets, children praised for effort (M = 6.81, SD =
2.23) solved more problems after failure than did children
praised for intelligence (M = 4.38, SD = 2.16) and children in
the control condition (M = 4.94, SD = 1.84). Follow-up t tests
showed that the scores of children in the effort condition differed
significantly from those in the intelligence, r(30) = 3.43, p
< .005, and control, t(30) = 2.12, p < .05, conditions but that
the scores of children in the intelligence and control conditions
did not differ from each other, f(30) = - 1 . 4 1 , ns.
However, unlike in the other studies, no significant differences
were found in children's desire to persist after failure, F(2, 45)
= 2.49, ns.
Overall, these findings indicate that, in spite of the introduction of a novel experimental task for the failure experience,
48
MUELLER AND DWECK
children praised for intelligence continued to show greater dec rements in enjoyment and performance than did children praised
for effort. They also continued to attribute their failure to a lack
of ability to a greater degree than did children praised for effort.
Thus, children who were told that their high performance was
caused by high intelligence appeared to continue to read their
ability from subsequent performance, even on an unrelated task,
whereas children praised for effort did not appear to reach the
same conclusion. The differential effects of praise for hard work
and praise for intelligence were therefore not limited to only
one experimental task; the attributional messages conveyed by
praise generalized to children's experiences with a new set of
problems.
Definitions of intelligence. Figure 5 shows that a significant
difference was obtained for children's endorsement of an entity
versus an incremental theory of intelligence in this study, F(2,
45) = 4.41, p < .05. As in Study 4, children in the intelligence
condition clearly endorsed an entity theory, affirming the fixed
nature of intelligence (M = 4.13, SD = 1.20), whereas children
in the effort and control conditions distinctly preferred an incremental view vis-a-vis the malleability of intelligence (M = 2.56,
SD = 1.55, and M = 2.94, SD = 1.84, respectively). Followup t tests showed significant differences between the intelligence
condition and both the effort, f(30) = 3.19, p < .005, and
control, f(30) = 2.16, p < .05, conditions. No differences were
noted between the effort and control conditions, 1(30) =
-0.62, ns.
These findings demonstrate that children praised for intelligence after a success are led to endorse statements that describe
intelligence as a fixed trait that is not subject to development,
even after they have experienced setbacks on a problem-solving
task.
In a related vein, on the open-ended question, children in
the effort and ability conditions were found to differ in their
definitions of intelligence. The two raters showed 94% agreement on their coding of malleable or motivational terms (e.g.,
knowledge and effort) and 83% agreement on their coding of
trait terms (e.g., smartness and ability); differences were resolved through discussion.
A chi-square comparison between children's use of malleable
or motivational terms for intelligence (e.g., "studying hard,"
"trying your best," and "how much you know") revealed a
significant difference between the intelligence and effort conditions, x 2 ( 2 , N = 48) = 5.81, p = .05. Whereas 56% of children
praised for effort used these terms (e.g., "It is to work hard")
to describe the nature of intelligence, only 25% of children
praised for ability did so. The responses of children in the control condition resembled those of children in the intelligence
condition; only 23% of them explained intelligence in terms of
knowledge or effort.
However, although the differences were in the predicted direction, a chi-square analysis did not reveal significant differences
between the three groups in terms of their use of trait terms
(e.g., "It is smartness") to define intelligence, x 2 ( 2 , N = 48)
= 2.23, ns.
Still, the results of the open-ended descriptions offer some
corroboration for children's endorsement of the statement about
the nature of intelligence. Children praised for ability after good
performance were found to be somewhat more likely to later
describe intelligence as a trait and to see it as not being subject

to improvement than were children praised for effort, who preferred to define it in malleable or motivational terms and to view
it as something that is subject to development or improvement.
Children in the control condition were not oriented toward one
consistent view of intelligence; although they tended to agree
with an incremental theory of intelligence as malleable on the
statement, they also used fewer malleable or motivational terms
to describe intelligence in the open-ended question.
In summary, Study 6 demonstrated that the effects of praise
for ability after successful performance on one experimental
task generalized to children's responses when they encountered
setbacks on a different task. In addition, it provided further
evidence that praise after good performance influences children's beliefs about the nature of intelligence.
General Discussion
Taken together, the findings.from the six studies provide striking evidence for the differential effects that praise for intelligence and praise for hard work have on children's achievement
behaviors and beliefs.
These effects became apparent early in each experimental
session when children were asked to choose between performance and learning goals for their future problem-solving tasks.
Children praised for intelligence after success chose problems
that allowed them to continue to exhibit good performance (representing a performance goal), whereas children praised for
hard work chose problems that promised increased learning.
This finding was further supported by the interest that children
showed in different types of information after they worked on
the experimental tasks. Children praised for intelligence preferred to find out about the performance of others on the tasks
rather than to learn about new strategies for solving the problems, even when these strategies might have improved their
future performance. Children praised for effort, on the other
hand, demonstrated their continued interest in mastery by preferring to receive strategy-related information. Thus, praise for
intelligence seemed to teach children to value performance, even
when following their own information-seeking interests,
whereas praise for hard work seemed to lead children to value
learning opportunities.
Further, children who received ability feedback appeared to
learn to measure their intelligence from their performance in a
way that children who received effort feedback did not. After
they faced failure, these children used low-ability, rather than
low-effort, attributions to account for their poor performance
more than did children praised for hard work, who preferred to
ascribe their failures to low effort. Thus, the children who were
explicitly told that they were smart after success were the ones
who most indicted their ability on the basis of poor performance.
This indictment of ability also led children praised for intelligence to display more negative responses in terms of lower
levels of task persistence, task enjoyment, and performance than
their counterparts, who received commendations for effort. That
children praised for intelligence after success adopted the tendency to measure their ability from their performance also was
evident in the ways in which they reported their performance
to others. Children praised for intelligence showed a greater

tendency to misrepresent their scores on the problems than did
children praised for effort, in spite of the fact that their reports
were anonymous and were not seen by the experimenter. This
result suggested that children praised for intelligence learned to
equate high performance with high ability in their own right
and not for the benefit of an evaluator. Thus, these children, on
their own, seemed to consider that low performance reflected
their intelligence in a way in which children praised for hard
work did not.
This belief in the power of performance to measure intelligence can become even more important to children's achievement responses when it is combined with an entity view of
intelligence as a fixed trait. Indeed, praise for ability was found
to orient children toward defining intelligence in terms of a
stable trait. That is, praise for high ability after success appeared to lead children to believe intelligence to be a fixed
trait whose level was measured from their performance on
academic tasks. It is thus not surprising that children who
received this type of intelligence feedback showed signs of
distress after they experienced a setback in their achievement.
Praise for hard work, on the other hand, appeared to lead
children to hold a more incremental theory of intelligence as
malleable and to define intelligence in terms of motivation and
knowledge. These children did not appear to consider intelligence to be determined from any single performance and were
found to avoid the postfailure achievement decrements of their
intelligence praise counterparts.
Children in the control condition received praise that contained a positive assessment of their work (i.e., "That's a really
high score") but no attributional component and that was similar to the performance-oriented praise used by Butler (1987)
(e.g., "Good j o b " ) . In general, control group children showed
more positive achievement motivation than did children praised
for intelligence but somewhat more negative achievement motivation than did children praised for effort. This finding is particularly interesting because it indicates that, even when children
are not taught to equate performance with intelligence, they
sometimes respond more negatively after praise for an outcome
or a product (i.e., "control" praise) than they do after praise
for process or effort (i.e., praise for hard work).
It is also important to note that virtually all of the findings
were replicated not only across genders but also across children
from several different ethnic groups in both rural and urban
communities. Thus, although other studies of achievement have
noted cultural differences in children's interpretations of and
responses to experimenter feedback (Garza & Lipton, 1978),
the present phenomenon seems important to the achievement of
all children and not just one specific cultural group.
Indeed, the differential effects of praise for effort and praise
for ability were found to be independent of actual ability level.
Children with high problem-solving scores were equally likely
to display the praise effects as were children with low scores.
Thus, even children who might reasonably be assumed to feel
confident in their skills (or who might be most likely to believe
the intelligence praise to be true) appear to be vulnerable to the
negative motivational effects of a focus on ability over effort.
49
praise for intelligence might have led children to have higher

judgments of their successful performance, higher expectations
for their future performance, or lower judgments of their poor
performance than might praise for effort. These different expectations or judgments then might have caused children praised
for intelligence to appear more dejected or less motivated after
failure than their effort praise counterparts. Studies 1, 2, and 4,
however, showed that children receiving the different attributions for their success did not differ in their judgments of their
past performance or their expectations for their future performance. Thus, their different postfailure persistence, enjoyment,
performance, and other achievement behaviors could not be explained in this way.
Second, the low motivation and achievement displayed by
children praised for intelligence after setbacks might have been
caused by their interpretations of the experimenter's assessments
of their ability after failure. That is, when these children were
told by the experimenter that they must be smart, they might
have believed that this experimenter would also measure their
intelligence from subsequent poor performance and be disappointed in this evidence of low ability. In this case, the negative
responses of these children to challenge might have been
brought on by their inability to live up to their perceptions of
the experimenter's judgments or expectations. This explanation
was made less plausible by the results of Study 5, in which two
experimenters administered praise for success and criticism for
failure. Children knew that the second experimenter, who gave
children the failure feedback, was not aware of their earlier
success and the attributional content of the praise that they were
given afterward and that the first experimenter had no knowledge
of their subsequent poor performance. Thus, it is unlikely that
children's negative postfailure behaviors and attributions were
based on their interpretations of the ability-praising experimenter' s disappointment in their abilities.
Finally, praise for intelligence might have led children to
believe that the experimental task was actually an intelligence
test that allowed the experimenter to diagnose their ability from
their performance. This belief in the diagnostic nature of the task
might have accounted for these children's negative reactions to
their poor performance on the second set of problems. However,
Study 6, in which a new task was used to provide the failure
experience, ruled out this explanation. Because children encountered challenge on a task that was unrelated to the first one, it
is unlikely that their subsequent negative responses could be
ascribed to beliefs about this novel task's diagnosticity.
Thus, it seems likely that praise for intelligence did indeed
lead children to learn that they could read trait information from
their performance on intellectual tasks in a way that praise for
effort did not. It is therefore not surprising that children exposed
to this intelligence feedback, with an emphasis on proving ability through high performance, were likely to respond negatively
when they faced achievement setbacks that prevented them from
attaining their performance goals. Children given effort feedback, on the other hand, who valued learning over performance,
were understandably less likely to fall apart when they experienced an isolated low performance.
Alternative Explanations
Implications and Future Research
Several possible explanations for the findings were eliminated

by the results of Studies 2 through 6. First, it was possible that
Although the encouragement of children with low achievement levels through the use of praise for their ability has received
50
MUELLER AND DWECK
widespread parental support (Mueller & Dweck, 1996; Phillips,

1984), the findings of our studies indicate that it could lead to
even more detrimental achievement beliefs and behaviors in
these children. Instead, the results presented here suggest that
when students succeed, attention and approbation should be
directed at their efforts or work strategies. That is, children
should be praised for the process of their work (e.g., focusing on
the task, using effective strategies, or persisting on challenging
problems) rather than for the end product and the ability that
produced it.
In addition, the findings may be used to shed light on a
persistent and puzzling paradox in achievement. Bright young
girls who are academic stars in grade school often seem most
vulnerable to later academic challenges (Cramer & Oshima,
1992; cf. Dweck, 1986; Dweck & Bush, 1976; Dweck, Davidson, Nelson, & Enna, 1978; Licht & Dweck, 1983, 1984). It is
possible that in their desire to bolster young girls' confidence in
their abilities, educators have gone out of their way to administer
intelligence praise. As shown in our studies, this well-intentioned approach could have an undesired impact on later motivation and performance.
The findings presented here may also have implications for
labeling children as talented or gifted, a form of intelligence
praise. That is, when children are so labeled, some may become
overly concerned with justifying that label and less concerned
with meeting challenges that enhance their skills. They may also
begin to react more poorly to setbacks because they worry that
mistakes, confusions, or failures mean that they do not deserve
to be labeled as gifted. It may therefore be especially important
in gifted-student programs to maintain an emphasis on meeting
challenges, applying effort, and searching for strategies.
Still, several limitations to our studies must be considered. For
instance, our studies all were conducted with novel experimental
tasks administered individually. Children may respond somewhat differently to praise for other types of tasks and in other
settings. Such potential experimental limitations must be thoroughly explored to determine the boundary conditions of our
findings. It is also important to examine the feedback context
in which the praise is given. Specifically, it might be that the
postfailure effects of praise for intelligence can be mitigated by
the attributional content of the failure feedback. For instance, it
is possible that the negative postfailure achievement effects of
intelligence praise may be avoided if effort attributions are used
as part of the failure feedback. Feedback that contains an effort
message may serve to reduce the harsh effects that failure can
have on the achievement of children praised for intelligence. It
could send an alternative message about the meaning of performance in the diagnosis of ability that may weaken the ability
orientation demonstrated in these studies. Further, it could give
children the option to attribute their poor performance to a
temporary state as opposed to a permanent trait. Thus, the effects
of the combination of effort attributions and ability praise on
children's responses to failure may be a fruitful area of future
research.
Future investigations of the impact of praise on children's
postfailure responses could lead to a closer examination of effort
praise itself and its effects on motivation. Is praise for effort,
strategy, or process always beneficial to children, or can it lead
them to greater disappointment if it is overemphasized or if hard
work fails to yield satisfactory results? Further, what impact

might praise that describes effort in stable, trait terms (e.g.,
'"fou must be a hard worker") have on children's achievement?
Given the negative consequences of praise for ability, the use of
trait terms to describe effort might well impair the positive
effects that effort praise has been demonstrated to have on postfailure responses.
The present research speaks to recent work suggesting that
performance goals can promote intrinsic motivation and good
performance (Harackiewicz, Baron, Carter, Lehto, & Elliot,
1997). We would not dispute this possibility. Indeed, the intelligence praise group, which was predominantly performance goal
oriented, showed high intrinsic motivation after their initial success, in terms of both task enjoyment and a desire to take the
problems home to work on them further. However, on the basis
of our present findings we would suggest that performance goals
carry greater vulnerability than do learning goals: In the face
of failure, both the intrinsic motivation and the performance of
the intelligence praise group showed sharp decrements. In most
intrinsic motivation research, the hardiness of the intrinsic motivationsuch as its ability to withstand failureis not tested.
We suggest that this might be a highly fruitful avenue for future
research.
Our findings might also speak to the important issue of contingent self-worth, the belief that one's worth or basic competence
is dependent on performing well (Burhans & Dweck, 1995;
Dykman, 1998; Harter, 1990). The intelligence praise can be
seen as promoting a sense of contingent self-worth vis-a-vis
intelligence and a need for validation. Those receiving the intelligence praise opted for a task that would provide further validation of their intelligence and, more than the other groups, saw
failure as an invalidation of their intelligence. It would be highly
interesting to investigate further the role of trait-oriented feedback in establishing a sense of contingent self-worth (see also
Kamins & Dweck, 1997).
Overall, our studies illustrate the important, and often unsuspected, role that praise after success can play in children's later
achievement motivation. Well-meant praise for intelligence,
which is intended to boost children's enjoyment, persistence,
and performance during achievement, does not prepare them for
coping with setbacks. In fact, we have demonstrated that this
type of ability feedback can undermine children's motivation
when they are later confronted with challenge. Indeed, researchers, educators, and parents alike might be well advised to borrow
a guideline from the literature on criticism when they decide to
praise children. That is, as with criticism, it is better to separate
"the deed from the doer" by applying praise to children's strategies and work habits rather than to any particular trait. Because
children cannot be insulated from failure throughout their lives,
great care should be taken to send them motivationally beneficial
messages after success.
References
Baumeister, R. F. (1984). Choking under pressure: Self-consciousness
and paradoxical effects of incentives on skillful performance. Journal
of Personality and Social Psychology, 46, 610-620.
Baumeister, R. E, Hutton, D. G., & Cairns, K. J. (1990). Negative effects of praise on skilled performance. Basic and Applied Social Psy-
chology, 11, 131-148.

Bell, S. M., McCallum, R. S., Bryles, J., & Driesler, K. (1994). Attributions for academic success and failure: An individual difference investigation of academic achievement and gender. Journal of Psychoeducational Assessment, 12, 4-13.
Briggs, D. C. (1970). Your child's self-esteem. Garden City, NY:
Doubleday.
Brophy, J. (1981). Teacher praise: A functional analysis. Review of
Educational Research, 51, 5-32.
Burhans, K., & Dweck, C. S. (1995). Helplessness in early childhood:
The role of contingent worth. Child Development, 66, 1719-1738.
Butler, R. (1987). Task-involving and ego-involving properties of evaluation: Effects of different feedback conditions on motivational perceptions, interest and performance. Journal of Educational Psychology,
79, 474-482.
Butler, R. (1988). Enhancing and undermining intrinsic motivation: The
effects of task-involving evaluation on interest and performance. British Journal of Educational Psychology, 58, 1-14.
Butler, R. (1993). Effects of task- and ego-achievement goals on information seeking during task engagement. Journal of Personality and
Social Psychology, 65, 18-31.
Covington, M. V, & Omelich, C. L. (1984). It's best to be able and
virtuous too: Student and teacher evaluative responses to successful
effort. Journal of Educational Psychology, 71, 688-700.
Cramer, J., & Oshima, T. C. (1992). Do gifted females attribute their
math performance differently than other students? Journal for the
Education of the Gifted, 16, 18-35.
Delin, C. R., & Baumeister, R. F. (1994). Praise: More than just social
reinforcement. Journal for the Theory of Social Behaviour, 24, 219241.
Diener, C. I., & Dweck, C. S. (1978). An analysis of learned helplessness: Continuous changes in performance, strategy, and achievement cognitions following failure. Journal of Personality and Social
Psychology, 36, 451-461.
Diener, C. I., & Dweck, C. S. (1980). An analysis of learned helplessness: II. The processing of success. Journal of Personality and
Dweck, C. S. (1975). The role of expectations and attributions in the
alleviation of learned helplessness. Journal of Personality and Social
Dweck, C. S. (1986). Motivational processes affecting learning. American Psychologist, 41, 1040-1048.
Dweck, C. S., & Bush, E. (1976). Sex differences in learned helplessness: I. Differential debilitation with peer and adult evaluators.
Developmental Psychology, 12, 147-156.
Dweck, C. S., Chiu, C. Y., & Hong, Y. Y. (1995). Implicit theories and
their role in judgments and reactions: A world from two perspectives.
Psychological Inquiry, 6, 267-285.
Dweck, C. S., Davidson, W., Nelson, S., & Enna, B. (1978). Sex differences in learned helplessness: II. The contingencies of evaluative feedback in the classroom and IH. An experimental analysis. Developmental Psychology, 14, 268-276.
Dweck, C. S., & Leggett, E. (1988). A social cognitive approach to
motivation and personality. Psychological Review, 95, 256-273.
Dykman, B. M. (1998). Integrating cognitive and motivational factors
in depression: Initial tests of a goal-oriented approach. Journal of
Personality and Social Psychology, 74, 139-158.
Elliott, E. S., & Dweck, C. S. (1988). Goal: An approach to motivation
and achievement. Journal of Personality and Social Psychology, 54,
5-12.
Garza, R. T., & Lipton, J. P. (1978). Culture, personality, and reactions
to praise and criticism. Journal of Personality, 46, 743-761.
Harackiewicz, J. M., Baron, K. E., Carter, S. M., Lehto, A. T, & Elliot,
A. J. (1997). Predictors and consequences of achievement goals in
51
the college classroom: Maintaining interest and making the grade.

Journal of Personality and Social Psychology, 73, 1284-1295.
Harter, S. (1990). Causes, correlates, and the functional role of global
self-worth: A life-span perspective. In R. J. Steinberg & J. Kolligan,
Jr. (Eds.), Competence considered (pp. 67-97). New Haven, CT!
Yale University Press.
Jagacinski, C. M., & Nicholls, J. G. (1984). Conceptions of ability and
related affects in task involvement and ego involvement. Journal of
Educational Psychology, 76, 909-919.
Kamins, M., & Dweck, C. S. (1997). The effect of praise and criticism
on children's coping with setbacks. Unpublished manuscript, Columbia University, New \brk.
Kanouse, D. E., Gumpert, P., & Canavan-Gumpert, D. (1981). The
semantics of praise. In J. H. Harvey, W. Ickes, & R. F. Kidd (Eds.),
New directions in attribution research (pp. 97-115). Hillsdale, NI:
Erlbaum.
Koestner, R., Zuckerman, M., & Koestner, J. (1987). Praise, involvement, and intrinsic motivation. Journal of Personality and Social
Leggett, E. L. (1986). Individual differences in effort/ability inference
rules and goals: Implications for causal judgments. Unpublished doctoral dissertation, Harvard University, Cambridge, MA.
Lezak, M. D. (1983). Neuropsychological assessment (3rd ed.). New
fork: Oxford University Press.
Licht, B. G., & Dweck, C. S. (1983). Sex differences in achievement
orientation: Consequences for academic choices and attainments. In
M. Marland (Ed.), Sex differentiation in schooling (pp. 72-97). London: Heinemann.
Licht, B.G., & Dweck, C. S. (1984). Determinants of academic
achievement: The interaction of children's achievement orientations
with skill area. Developmental Psychology, 20, 628-636.
Likert, R., & Quasha, W. H. (1970). The revised Minnesota Paper Form
Board Test. New \brk: Psychological Corporation.
Meyer, W. U. (1992). Paradoxical effects of praise and criticism on
perceived ability. In W. Stroebe & M. Hewstone (Eds.), European
review of social psychology (Vol. 3, pp. 259-283). Chichester, England: Wiley.
Meyer, W. U., Bachmann, M., Biermann, U., Hempelmann, M., Ploeger,
F. D., & Spiller, H. (1979). The informational value of evaluative
behavior: Influences of praise and blame on perceptions of ability.
Journal of Educational Psychology, 71, 259-268.
Meyer, W. U., Mittag, W., & Engler, U. (1986). Some effects of praise
and blame on perceived ability and affect. Social Cognition, 4, 293308.
Miller, R. L., Brickman, P., & Bolen, D. (1975). Attribution versus
persuasion as a means for modifying behavior. Journal of Personality
and Social Psychology, 31, 430-441.
Mueller, C. M., & Dweck, C. S. (1996, April). Implicit theories of intelligence: Relation of parental beliefs to children's expectations. Poster
session presented at Head Start's Third National Research Conference,
Washington, DC.
Mueller, C. M., & Dweck, C. S. (1997). [Implicit theories of intelligence: Malleability beliefs, definitions and judgments of intelligence].
Unpublished raw data.
Nicholls, J. G. (1975). Causal attributions and other achievement-related
cognitions: Effects of task outcome, attainment value, and sex. Journal of Personality and Social Psychology, 31, 379-389.
Nicholls, J. G. (1976). Effort is virtuous, but it's better to have ability:
Evaluative responses to perceptions of effort and ability. Journal of
Research in Personality, 10, 306-315.
Nicholls, J. G. (1984). Achievement motivation: Conceptions of ability,
subjective experience, task choice, and performance. Psychological
Review, 91, 328-346.
Phillips, R. H. (1984). Increasing positive self-referent statements to
52
MUELLER AND DWECK
improve self-esteem in low-income elementary school children. Journal of School Psychology, 22, 155-163.
Powers, S., Douglas, P., Cool, B. A., & Gose, K. F. (1985). Achievement
motivation and attributions for success and failure. Psychological Reports, 57, 751-754.
Raven, J. C. (1976). Standard progressive matrices. Oxford, England:
Oxford Psychologists Press.
Ruble, D. N., & Frey, K. S. (1991). Changing patterns of comparative
behavior as skills are acquired: A functional model of self-evaluation.
In J. Suls & T. A. Wills (Eds.), Social comparison: Contemporary
theory and research (pp. 79-113). Hillsdale, NJ: Erlbaum.
Schunk, D. H. (1983). Ability versus effort attributional feedback: Differential effects on self-efficacy and achievement. Journal of Educational Psychology, 75, 848-856.
Schunk, D. H. (1994). Self-regulation of self-efficacy and attributions
in academic settings. In D. H. Schunk & B. J. Zimmerman (Eds.),
Self-regulation of learning and performance: Issues and educational
applications (pp. 75-99). Hillsdale, NJ: Erlbaum.
Schunk, D. H. (1996, April). Attributions and the development of selfregulatory competence. Paper presented at the annual meeting of the
American Educational Research Association, New \fork.
Weiner, B. (1972). Theories of motivation: From mechanism to cognition. Chicago: Markham.
Weiner, B. (1985). An attributional theory of achievement motivation
and emotion. Psychological Review, 92, 548-573.
Weiner, B., Frieze, I., Kukla, A., Reed, L., Rest, S., & Rosenbaum, R. M.
(1971). Perceiving the causes of success and failure. Morristown,
NJ: General Learning Press.
Weiner, B., Russell, D., & Lerman, D. (1979). The cognition-emotion
process in achievement-related contexts. Journal of Personality and
Received June 4, 1997

Revision received December 19, 1997
Accepted December 22, 1997

Dweck - Praise For Intelligence

Uploaded by

Copyright:

Available Formats

Dweck - Praise For Intelligence

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Dweck - Praise For Intelligence

Uploaded by

Copyright:

Available Formats

Copyright 1998 by the American Psychological Association, Inc.

Journal of Personality and Social Psychology

Praise for Intelligence Can Undermine Children's

Praise for high ability is a common response to a job well

smart, the greater will be their enjoyment of and motivation for

Consequences of Praise for Ability Given

Claudia M. Mueller and Carol S. Dweck, Department of Psychology,

MUELLER AND DWECK

Previous Comparisons of Effort Feedback

Consequences of Praise for Effort Given

The Current Research

The six studies described in this article were designed to

PRAISE CAN UNDERMINE MOTIVATION

information about the scores of others over strategy information

MUELLER AND DWECK

by children's responses to the question "How much would you like to

progressive matrices equal to the first set in level of difficulty. This

Results and Discussion

PRAISE CAN UNDERMINE MOTIVATION

want to learn new things, while intelligence praise led children

reflected their ability and thus attributed low performance to

MUELLER AND DWECK

Fbllow-up t tests indicated that children praised for intelligence

Note. Within rows, means with different subscripts differ significantly.

tween the groups in their change in performance from the first

Note. Within rows, means with different subscripts differ significantly.

PRAISE CAN UNDERMINE MOTIVATION

finding indicates that the effects of praise for intelligence were

MUELLER AND DWECK

for children's motivation before their poor performance.

Results and Discussion

PRAISE CAN UNDERMINE MOTIVATION

children that the adult experimenter is measuring them from

Results and Discussion

Postfailure Dependent Measures

MUELLER AND DWECK

Children praised for hard work assigned greater weight to

their prefailure scores by 1.23 (SD = 1.50) problems. Children

Type of Praise Given

PRAISE CAN UNDERMINE MOTIVATION

Type of Praise Given

MUELLER AND DWECK

was likely to have an adverse effect on their future problem

Results and Discussion

Type of Praise Given

PRAISE CAN UNDERMINE MOTIVATION

public elementary schools in a large northeastern city. Children ranged

Results and Discussion

MUELLER AND DWECK

after praise for ability are indeed attributable to differences in

PRAISE CAN UNDERMINE MOTIVATION

Postfailure dependent measures. As found in the earlier

effort attributions, F(2,45) = 23.38,p < .001. Children praised

MUELLER AND DWECK

describe intelligence as a trait and to see it as not being subject

PRAISE CAN UNDERMINE MOTIVATION

praise for intelligence might have led children to have higher

Implications and Future Research