Grit: Perseverance and Passion for Long-Term Goals

Abstract

The importance of intellectual talent to achievement in all professional domains is well established, but less is known about other individual differences that predict success. The authors tested the importance of 1 noncognitive trait: grit. Defined as perseverance and passion for long-term goals, grit accounted for an average of 4% of the variance in success outcomes, including educational attainment among 2 samples of adults (N=1,545 and N=690), grade point average among Ivy League undergraduates (N=138), retention in 2 classes of United States Military Academy, West Point, cadets (N=1,218 and N=1,308), and ranking in the National Spelling Bee (N=175). Grit did not relate positively to IQ but was highly correlated with Big Five Conscientiousness. Grit nonetheless demonstrated incremental predictive validity of success measures over and beyond IQ and conscientiousness. Collectively, these findings suggest that the achievement of difficult goals entails not only talent but also the sustained and focused application of talent over time.

Full text

PERSONALITY PROCESSES AND INDIVIDUAL DIFFERENCES
Grit: Perseverance and Passion for Long-Term Goals
Angela L. Duckworth
University of Pennsylvania
Christopher Peterson
University of Michigan
Michael D. Matthews and Dennis R. Kelly
United States Military Academy, West Point
The importance of intellectual talent to achievement in all professional domains is well established, but less
is known about other individual differences that predict success. The authors tested the importance of 1
noncognitive trait: grit. Defined as perseverance and passion for long-term goals, grit accounted for an average
of 4% of the variance in success outcomes, including educational attainment among 2 samples of adults (N⫽
1,545 and N⫽690), grade point average among Ivy League undergraduates (N⫽138), retention in 2 classes
of United States Military Academy, West Point, cadets (N⫽1,218 and N⫽1,308), and ranking in the
National Spelling Bee (N⫽175). Grit did not relate positively to IQ but was highly correlated with Big Five
Conscientiousness. Grit nonetheless demonstrated incremental predictive validity of success measures over
and beyond IQ and conscientiousness. Collectively, these findings suggest that the achievement of difficult
goals entails not only talent but also the sustained and focused application of talent over time.
Keywords: achievement, success, personality, persistence, performance
Compared with what we ought to be, we are only half awake. Our
fires are damped, our drafts are checked. We are making use of only
a small part of our possible mental resources. . .men the world over
possess amounts of resource, which only exceptional individuals push
to their extremes of use. (William James, 1907, pp. 322–323)
In 1907, William James proposed “a program of study that might
with proper care be made to cover the whole field of psychology” (p.
332). James encouraged psychologists to address two broad problems:
First, what are the types of human abilities and, second, by what
diverse means do individuals unleash these abilities?
In the century that has passed since James’s suggestion, psycho-
logical science has made impressive progress in answering the first of
these two questions. In particular, we know a great deal about intel-
ligence, or general mental ability, a construct for which formal study
was initiated by a British contemporary of James, Sir Francis Galton.
Notwithstanding vigorous debates over the dimensionality and origins
of intelligence, we know more about IQ— how to measure it reliably
and precisely and what outcomes it predicts—than any other stable
individual difference. In contrast, we know comparatively little about
why, as James put it, most individuals make use of only a small part
of their resources, whereas a few exceptional individuals push them-
selves to their limits.
In this article, we reiterate James’s second question in the following
terms: Why do some individuals accomplish more than others of
equal intelligence? In addition to cognitive ability, a list of attributes
of high-achieving individuals would likely include creativity, vigor,
emotional intelligence, charisma, self-confidence, emotional stability,
physical attractiveness, and other positive qualities. A priori, some
traits seem more crucial than others for particular vocations. Extra-
version may be fundamental to a career in sales, for instance, but
irrelevant to a career in creative writing. However, some traits might
be essential to success no matter the domain.
1
We suggest that one
personal quality is shared by the most prominent leaders in every
field: grit.
We define grit as perseverance and passion for long-term goals.
Grit entails working strenuously toward challenges, maintaining
1
In this article, we are concerned with objective accomplishments. That
is, we are interested in vocational and avocational achievements that are
recognized by other people, in contrast to those that are primarily of
subjective value to the individual. We do not examine success in other
important domains of life, such as parenting, citizenship, friendship, and so
on. Thus, we use terms like success and achievement to refer to the
accomplishment of widely valued goals.
Angela L. Duckworth, Department of Psychology, University of Penn-
sylvania; Christopher Peterson, Department of Psychology, University of
Michigan; Michael D. Matthews, Department of Behavioral Sciences and
Leadership, United States Military Academy, West Point; Dennis R. Kelly,
Institutional Research and Analysis Branch, United States Military Acad-
emy, West Point.
This research was supported by a National Science Foundation Graduate
Fellowship and a grant from the John Templeton Foundation. For helpful
comments on a draft of this article, we thank Sigal Barsade, Dianne
Chambless, Martha Farah, Gary Latham, Paul Rozin, Richard Shell, Dean
Simonton, and especially Martin Seligman. We are grateful to Robert
Gallop and Paul McDermott for guidance on statistical analyses. Finally,
we thankfully acknowledge the efforts of Paige Kimble, Edgar Knizhnik,
Patty Newbold, Patrick Quinn, and Cybelle Weeks in data acquisition and
management.
Correspondence concerning this article should be addressed to Angela
L. Duckworth, Department of Psychology, University of Pennsylvania,
3451 Walnut Street, Philadelphia, PA 19104. E-mail: angela_duckworth@
yahoo.com
Journal of Personality and Social Psychology, 2007, Vol. 92, No. 6, 1087–1101
Copyright 2007 by the American Psychological Association 0022-3514/07/$12.00 DOI: 10.1037/0022-3514.92.6.1087
1087

effort and interest over years despite failure, adversity, and pla-
teaus in progress. The gritty individual approaches achievement as
a marathon; his or her advantage is stamina. Whereas disappoint-
ment or boredom signals to others that it is time to change
trajectory and cut losses, the gritty individual stays the course.
Our hypothesis that grit is essential to high achievement evolved
during interviews with professionals in investment banking, paint-
ing, journalism, academia, medicine, and law. Asked what quality
distinguishes star performers in their respective fields, these indi-
viduals cited grit or a close synonym as often as talent. In fact,
many were awed by the achievements of peers who did not at first
seem as gifted as others but whose sustained commitment to their
ambitions was exceptional. Likewise, many noted with surprise
that prodigiously gifted peers did not end up in the upper echelons
of their field.
More than 100 years prior to our work on grit, Galton (1892)
collected biographical information on eminent judges, statesmen,
scientists, poets, musicians, painters, wrestlers, and others. Ability
alone, he concluded, did not bring about success in any field.
Rather, he believed high achievers to be triply blessed by “ability
combined with zeal and with capacity for hard labour” (p. 33).
Similar conclusions were reached by Cox (1926) in an analysis of
the biographies of 301 eminent creators and leaders drawn from a
larger sample compiled by J. M. Cattell (1903). Estimated IQ and
Cattell’s rank order of eminence were only moderately related (r⫽
.16) when reliability of data was controlled for. Rating geniuses on
67 character traits derived from Webb (1915), Cox concluded that
holding constant estimated IQ, the following traits evident in
childhood predicted lifetime achievement: “persistence of motive
and effort, confidence in their abilities, and great strength or force
of character” (p. 218).
As context for the current research, we briefly review more
recent research on individual differences that bear on success. We
leave aside for the moment questions about how goals are set and
maintained, how values and expectancies affect goal attainment,
and so on. We also omit from our review situational factors and
social and cultural variables that influence achievement. For
a broader review than is possible here, we refer the reader to
Simonton (1994) and Latham and Pinder (2005).
Talent and Achievement
Intelligence is the best-documented predictor of achievement
(Gottfredson, 1997; Hartigan & Wigdor, 1989). Reliable and valid
measures of IQ have made it possible to document a wide range of
achievement outcomes affected by IQ, including college and grad-
uate school grade point average (GPA; e.g., Bridgeman,
McCamley-Jenkins, & Ervin, 2000; Kuncel, Hezlett, & Ones,
2001), induction into Phi Beta Kappa (Langlie, 1938), income
(Fergusson, Horwood, & Ridder, 2005), career potential and job
performance (Kuncel, Hezlett, & Ones, 2004), and choice of
occupation (Chown, 1959). The predictive validities of intelli-
gence rise with the complexity of the occupation considered. When
corrected for attenuation due to reliability of measures and restric-
tion on range, correlations between IQ and these various outcomes
can be as high as r⫽.6, meaning that IQ may account for up to
one third of the variance in some measures of success (Neisser et
al., 1996).
However, in the Terman longitudinal study of mentally gifted
children, the most accomplished men were only 5 points higher in
IQ than the least accomplished men (Terman & Oden, 1947). To
be sure, restriction on range of IQ partly accounted for the slight-
ness of this gap, but there was sufficient variance in IQ (SD ⫽
10.6, compared with SD ⫽16 in the general population) in the
sample to have expected a much greater difference. More predic-
tive than IQ of whether a mentally gifted Terman subject grew up
to be an accomplished professor, lawyer, or doctor were particular
noncognitive qualities: “Perseverance, Self-Confidence, and Inte-
gration toward goals” (Terman & Oden, 1947, p. 351). Terman and
Oden, who were close collaborators of Cox, encouraged further
inquiry into why intelligence does not always translate into
achievement: “Why this is so, what circumstances affect the fru-
ition of human talent, are questions of such transcendent impor-
tance that they should be investigated by every method that prom-
ises the slightest reduction of our present ignorance” (p. 352).
Reviewing the biographical details of Darwin, Einstein, and
other geniuses, Howe (1999) disputed the assumption that high
achievement derives directly from exceptional mental ability:
“Perseverance is at least as crucial as intelligence....The most
crucial inherent differences may be ones of temperament rather
than of intellect as such” (p. 15). Likewise, summarizing an
extensive body of research on the development of expertise, Erics-
son and Charness (1994) concluded that in chess, sports, music,
and the visual arts, over 10 years of daily “deliberate practice” set
apart expert performers from less proficient peers and that 20 years
of dedicated practice was an even more reliable predictor of
world-class achievement. Like Howe, Ericsson and Charness sug-
gested that inborn ability is less important than commonly thought:
“More plausible loci of individual differences are factors that
predispose individuals toward engaging in deliberate practice and
enable them to sustain high levels of practice for many years” (p.
744).
Personality and Achievement
The Big Five model has provided a descriptive framework for
much of the contemporary empirical work on traits that predict
success (Goldberg, 1990; John & Srivastava, 1999; McCrae &
Costa, 1987; Tupes & Christal, 1992). In a 1991 meta-analysis,
Barrick and Mount concluded that Big Five Conscientiousness
related more robustly to job performance than did Big Five Ex-
traversion, Openness to Experience, Neuroticism, or Agreeable-
ness (Barrick & Mount, 1991). Uncorrected correlations between
conscientiousness and job performance ranged from r⫽.09 to r⫽
.13, depending on the occupational group. In a meta-analysis of
confirmatory studies of personality measures as predictors of job
performance, Tett, Jackson, and Rothstein (1991) observed a
sample-weighted mean correlation between conscientiousness and
job performance of r⫽.12.
One might conclude from these meta-analyses that at best, any
given personality trait accounts for less than 2% of variance in
achievement. If so, compared with IQ, personality would seem
inconsequential. Alternatively, it is possible that more narrowly
defined facets of Big Five factors may more robustly predict
particular achievement outcomes (Paunonen & Ashton, 2001). It is
also possible that there exist important personality traits not rep-
resented as Big Five facets. A serious limitation of the Big Five
1088 DUCKWORTH, PETERSON, MATTHEWS, AND KELLY

taxonomy derives from its roots in the factor analyses of adjec-
tives. Traits for which there are fewer synonyms (or antonyms)
tend to be omitted. We agree with Paunonen and Jackson (2000)
that
the ultimate test of whether a dimension of behavior is important to
the understanding of human behavior depends not on the size of the
factor in the language of personality. . .if such dimensions are able to
account for criterion variance not accounted for by the Big Five
personality factors, then those dimensions need to be considered
separately in any comprehensive description of the determinants of
human behavior. (p. 833)
Thus, although we recognize the utility of the Big Five taxonomy
as a descriptive framework in which newly characterized person-
ality traits should be situated, we do not believe that it provides an
exhaustive list of traits worth studying.
Conscientious individuals are characteristically thorough, care-
ful, reliable, organized, industrious, and self-controlled. Whereas
all of these qualities bear a plausible contribution to achievement,
their relative importance likely varies depending upon the type of
achievement considered. For example, Galton (1892) suggested
that self-control—the ability to resist temptation and control im-
pulses—is a surprisingly poor predictor of the very highest
achievements:
People seem to have the idea that the way to eminence is one of great
self-denial, from which there are hourly temptations to di-
verge....This is true enough of the great majority of men, but it is
simply not true of the generality of those who have gained great
reputations. Such men, biographies show to be haunted and driven by
an incessant instinctive craving for intellectual work. (p. 36)
Consistent with Galton’s distinction, Hough (1992) distinguished
between achievement and dependability aspects of conscientious-
ness. According to Hough, the achievement-oriented individual is
one who works hard, tries to do a good job, and completes the task
at hand, whereas the dependable person is self-controlled and
conventional (p. 144). In a meta-analysis, Hough found scales
classified as measuring achievement orientation predicted job pro-
ficiency (r⫽.15) and educational success (r⫽.29) better than did
dependability (r⫽.08 and r⫽.12, respectively).
Grit overlaps with achievement aspects of conscientiousness but
differs in its emphasis on long-term stamina rather than short-term
intensity. The gritty individual not only finishes tasks at hand but
pursues a given aim over years. Grit is also distinct from depend-
ability aspects of conscientiousness, including self-control, in its
specification of consistent goals and interests. An individual high
in self-control but moderate in grit may, for example, effectively
control his or her temper, stick to his or her diet, and resist the urge
to surf the Internet at work—yet switch careers annually. As
Galton (1892) suggested, abiding commitment to a particular vo-
cation (or avocation) does not derive from overriding “hourly
temptations.”
Grit also differs from need for achievement, described by Mc-
Clelland (1961) as a drive to complete manageable goals that allow
for immediate feedback on performance. Whereas individuals high
in need for achievement pursue goals that are neither too easy nor
too hard, individuals high in grit deliberately set for themselves
extremely long-term objectives and do not swerve from them—
even in the absence of positive feedback. A second important
distinction is that need for achievement is by definition a noncon-
scious drive for implicitly rewarding activities and, therefore,
impossible to measure using self-report methods (McClelland,
Koestner, & Weinberger, 1992). Grit, in contrast, can entail ded-
ication to either implicitly or explicitly rewarding goals. Further,
we see no theoretical reason why individuals would lack aware-
ness of their level of grit.
Development of the Grit Scale
The aforementioned reasoning suggests that grit may be as
essential as IQ to high achievement. In particular, grit, more than
self-control or conscientiousness, may set apart the exceptional
individuals who James thought made maximal use of their abili-
ties. To test these hypotheses, we sought a brief, stand-alone
measure of grit that met four criteria: evidence of psychometric
soundness, face validity for adolescents and adults pursuing goals
in a variety of domains (e.g., not just work or school), low
likelihood of ceiling effects in high-achieving populations, and
most important, a precise fit with the construct of grit.
We reviewed several published self-report measures but failed
to find any that met all four of our criteria. The only stand-alone
measure of perseverance we found, the Perseverance Scale for
Children (Lufi & Cohen, 1987), is not face valid for adults. The
Passion Scale (Vallerand et al., 2003) assesses commitment to a
subjectively important activity but not perseverance of effort. The
tenacity scale used by Baum and Locke (2004) and derived from
Gartner, Gatewood, and Shaver (1991) was developed for entre-
preneurs and is not face valid for adolescents. Similarly, the Career
Advancement Ambition Scale (DesRochers & Dahir, 2000) refers
to attitudes toward one’s “profession” and “firm.” Cassidy and
Lynn (1989) developed a need for achievement questionnaire that
taps work ethic and desire for excellence, which are consonant
with the construct of grit, but also several irrelevant qualities such
as the needs for money, domination of others, superiority over
competitors, and social status. Finally, the goal commitment scale
by Hollenbeck, Williams, and Klein (1989) assesses state-level,
not trait-level, goal commitment.
The Present Research
In the absence of adequate existing measures, we developed and
validated a self-report questionnaire called the Grit Scale. We
expected grit to be associated with Big Five Conscientiousness and
with self-control but, in its emphasis on focused effort and interest
over time, to have incremental predictive validity for high accom-
plishment over and beyond these other constructs.
We also tested the hypothesis that grit would be unrelated to IQ.
Whereas personality and IQ represent independently flourishing
literatures, few contemporary investigations have incorporated
both kinds of measures. Thus, we have learned surprisingly little
about how personality traits and intelligence are related and about
their relative contributions to performance. There are notable ex-
ceptions to this trend (cf. Ackerman & Heggestad, 1997;
Chamorro-Premuzic & Furnham, 2005), but in general, psychol-
ogy has ignored the recommendations of Wechsler (1940) and
R. B. Cattell and Butcher (1968), who cautioned that the indepen-
dent study of either noncognitive or cognitive individual differ-
ences, to the exclusion of the other, would be impoverished.
1089
PERSEVERANCE AND PASSION

Study 1
Study 1 was a cross-sectional study for which the major purpose
was to develop and validate a self-report measure of grit in a large
sample of adults aged 25 years or older. The predictive validity of
grit was assessed by its association with higher levels of lifetime
schooling among individuals of identical age.
The broad age range of the adults in Study 1 allowed us to
venture a second question: Does grit grow with age? Although
personality traits are by definition relatively stable over time, Big
Five Conscientiousness and stability of vocational interests both
increase over the life span (McCrae et al., 1999; Srivastava, John,
Gosling, & Potter, 2003; Swanson, 1999). Thus, we expected older
adults to be slightly higher in grit than younger individuals.
Method
Participants and procedure. Beginning in April 2004, we set
up a link on the www.authentichappiness.org website inviting
visitors to help validate the Grit Scale. This noncommercial, public
website provides free information about psychology research and
access to a variety of self-report measures to over 500,000 regis-
tered users. All participants indicated how old they were (25 to 34
years, 35 to 44 years, 45 to 54 years, 55 to 64 years, and 65 years
and older) and their level of education (some high school, high
school graduate, some college, Associate’s degree, Bachelor’s
degree, or postcollege graduate degree). By October 2005, we
collected data on 1,545 participants aged 25 and older (M⫽45
years; 73% women, 27% men).
Development of the Grit Scale. We began by generating a pool
of 27 items tapping the construct of grit. Our overarching goal for
scale development was to capture the attitudes and behaviors
characteristic of the high-achieving individuals described to us in
early, exploratory interviews with lawyers, businesspeople, aca-
demics, and other professionals. We intentionally wrote items that
would be face valid for both adolescents and adults and that did not
specify a particular life domain (e.g., work, school). We included
items that tapped the ability to sustain effort in the face of adver-
sity (e.g., “I have overcome setbacks to conquer an important
challenge,” “I finish whatever I begin”). We also considered that
some people sustain effort not because of subjective interest but
rather because they are afraid of change, compliant with the
expectations of others, or unaware of alternative options. Thus,
several Grit Scale items ask about the consistency of interests over
time. For example, two reverse-scored items were “My interests
change from year to year” and “I have difficulty maintaining my
focus on projects that take more than a few months to complete.”
Items are rated on a 5-point scale from 1 ⫽not at all like me to 5 ⫽
very much like me.
We considered item-total correlations, internal reliability coef-
ficients, redundancy, and simplicity of vocabulary to eliminate 10
items. On the remaining 17 items, we conducted an exploratory
factor analysis on half of the observations chosen at random (n⫽
772). We sought a solution that satisfied tests for number of factors
(e.g., R. B. Cattell’s scree test), retained 5 or more items with
loadings of at least .40, yielded internally consistent factors that
made psychological sense, and best approximated simple structure.
A two-factor oblique solution with promax rotation satisfied these
criteria. See Table 1 for the 12 retained items and corrected
item-total correlations with each item’s respective factor. We
considered the possibility that these two factors were an artifact of
positively and negatively scored items but were convinced that the
factor structure reflected two conceptually distinct dimensions.
The first factor contained 6 items indicating consistency of inter-
ests, and the second factor contained 6 items indicating persever-
ance of effort. Because we expected that stamina in the dimensions
of interest and effort would be correlated, we accepted this oblique
solution in which the two factors were correlated at r⫽.45.
To test the integrity of the final two-factor solution, we con-
firmed that the specificity of each factor (i.e., the portion of
reliable variance not shared by the other factor) was larger than the
error variance for that factor. Further, confirmatory factor analysis
with the remaining 773 observations in our sample supported this
two-factor solution (comparative fit index ⫽.83 and root-mean-
square error of approximation ⫽.11). The resulting 12-item Grit
Table 1
Common Factor Analysis of Grit Scale With Promax Rotation
Factor and Grit Scale item Promax loading Item-total r
Consistency of Interests
I often set a goal but later choose to pursue a different one.
a
.61 .51
New ideas and new projects sometimes distract me from previous ones.
a
.77 .54
I become interested in new pursuits every few months.
a
.73 .59
My interests change from year to year.
a
.69 .51
I have been obsessed with a certain idea or project for a short time but later lost interest.
a
.66 .44
I have difficulty maintaining my focus on projects that take more than a few months to complete.
a
.47 .62
Perseverance of Effort
I have achieved a goal that took years of work. .65 .62
I have overcome setbacks to conquer an important challenge. .68 .53
I finish whatever I begin. .54 .68
Setbacks don’t discourage me. .58 .59
I am a hard worker. .44 .70
I am diligent. .64 .82
Note. The last column displays the corrected item-total correlations for each item with its respective factor (i.e., either Consistency of Interests or
Perseverance of Effort).
a
Item was reverse scored.
1090 DUCKWORTH, PETERSON, MATTHEWS, AND KELLY

Scale demonstrated high internal consistency (␣⫽.85) for the
overall scale and for each factor (Consistency of Interests, ␣⫽.84;
Perseverance of Effort, ␣⫽.78). In subsequent analyses, neither
factor was consistently more predictive of outcomes than the other,
and in most cases, the two together were more predictive than
either alone. Therefore, we proceeded using total scores from the
full 12-item scale as our measure of grit.
Results and Discussion
As we predicted, more educated adults were higher in grit than
were less educated adults of equal age. We treated age and edu-
cational attainment as categorical variables. Two-way analysis of
variance models were used to test for differences in grit by edu-
cation and age. The interaction term was not significant, indicating
that differences in grit for levels of education were not differential
across age and that the differences in grit for levels of age were not
differential across education. We therefore fit a reduced model
excluding the interaction term. Main effects for each term indi-
cated a highly significant difference in grit for the levels of each
term adjusted for the other effect, F(5, 1535) ⫽15.48, p⬍.001, ␩
p
2
⫽
0.05, for education; F(4, 1535) ⫽11.98, p⬍.001, ␩
p
2
⫽0.03, for age.
As illustrated in Figure 1, post hoc comparisons revealed that when
age is controlled for, postcollege graduates were higher in grit than
most other groups. Similarly, participants with an Associate’s degree
were significantly higher in grit than those with less education and,
interestingly, also higher in grit than those with a Bachelor’s degree,
although this difference failed to reach significance.
Figure 2 shows that when education level is controlled for, grit
increased monotonically with age; however, 25- to 34-year-olds
did not differ significantly from 35- to 44-year-olds, and 45- to
54-year-olds did not differ significantly from 55- to 64-year-olds.
We confirmed that this effect was not an artifact of older participants
simply having more life experience and, therefore, a greater likelihood
of endorsing Grit Scale items asking about past experiences (e.g., “I
have overcome setbacks to conquer an important challenge.”) Exclud-
ing the 3 items phrased in the past tense did not change the relation-
ship between Grit Scale scores and age. Summary statistics for
Study 1 and all subsequent studies can be found in Table 2.
The cross-sectional design of Study 1 limits our ability to draw
strong causal inferences about the observed positive association
between grit and age. Our intuition is that grit grows with age and
that one learns from experience that quitting plans, shifting goals,
Figure 1. Grit as a function of educational attainment, controlling for age in Study 1 participants. Error bars
represent 95% confidence intervals of the mean.
1091
PERSEVERANCE AND PASSION

and starting over repeatedly are not good strategies for success. In
fact, a strong desire for novelty and a low threshold for frustration
may be adaptive earlier in life: Moving on from dead-end pursuits
is essential to the discovery of more promising paths. However, as
Ericsson and Charness (1994) demonstrated, excellence takes
time, and discovery must at some point give way to development.
Alternatively, McCrae et al. (1999) speculated that maturational
changes in personality, at least through middle adulthood, might be
genetically programmed. From an evolutionary psychology per-
spective, certain traits may not be as beneficial when seeking
mates as when providing for and raising a family. A third possi-
bility is that the observed association between grit and age is a
consequence of cohort effects. It may be that each successive gener-
ation of Americans, for social and cultural reasons, has grown up less
gritty than the one before (cf. Twenge, Zhang, & Im, 2004).
Similarly, we interpret the observed association between grit
and education as evidence that sticking with long-range goals over
time makes possible completion of high levels of education. But,
it is also possible that when evaluating one’s ability to stay focused
on goals, overcome setbacks, and so on, personal academic ac-
complishments were particularly salient and, therefore, spuriously
inflated grit scores. Finally, because all information in Study 1 was
self-reported and because grit was not compared with other traits,
we cannot rule out the possibility that observed positive associa-
tions were the consequence of social desirability bias.
Study 2
In Study 1, grit was associated with educational attainment and
age. The purpose of Study 2 was to test whether these relationships
Figure 2. Grit as a function of age (in years), controlling for educational attainment in Study 1 participants.
Error bars represent 95% confidence intervals of the mean.
Table 2
Summary Statistics for Grit Scale Across Studies
Sample characteristics ␣NMSD
Study 1: Adults aged 25 and older .85 1,545 3.65 0.73
Study 2: Adults aged 25 and older .85 690 3.41 0.67
Study 3: Ivy League undergraduates .82 138 3.46 0.61
Study 4: West Point cadets in Class of
2008 .77 1,218 3.78 0.53
Study 5: West Point cadets in Class of
2010 .79 1,308 3.75 0.54
Study 6: National Spelling Bee
finalists .80 175 3.50 0.67
1092 DUCKWORTH, PETERSON, MATTHEWS, AND KELLY

would hold when conscientiousness and other Big Five traits were
controlled for. That is, does grit provide incremental predictive
validity over and beyond Big Five traits? Also, is there evidence
that grittier individuals make fewer career switches than their less
gritty peers?
Method
Beginning in April 2006, we revised our online study on www
.authentichappiness.org. By September 2006, 706 participants
aged 25 and older completed the same measures as in Study 1. In
addition, participants indicated “the number of times I have
changed careers” and completed the Big Five Inventory (BFI; John
& Srivastava, 1999), a widely used 44-item questionnaire that has
demonstrated convergent validity with Costa and McCrae’s (1992)
NEO Five-Factor Inventory and Goldberg’s (1992) Trait Descrip-
tive Adjectives measures of Big Five traits. Participants endorse
items such as “I see myself as someone who is talkative” using a
5-point Likert scale, where 1 ⫽disagree strongly and 5 ⫽agree
strongly. Observed internal reliabilities of the BFI subscales mea-
suring conscientiousness, extraversion, neuroticism, agreeable-
ness, and openness to experience were ␣⫽.86, .89, .85, .82, and
.84, respectively. Only 16 participants (2%) reported as their
highest education level either “high school” or “some high
school.” Therefore these individuals were excluded from analysis.
The resultant sample comprised 690 participants (M⫽45 years,
SD ⫽11; 80% women, 20% men).
Results and Discussion
As we expected, grit related to Conscientiousness (r⫽.77, p⬍
.001) more than to Neuroticism (r⫽⫺.38, p⬍.001), Agreeable-
ness (r⫽.24, p⬍.001), Extraversion (r⫽.22, p⬍.001), and
Openness to Experience (r⫽.14, p⬍.001).
The incremental predictive validity of grit for education and age
over and beyond conscientiousness and other Big Five traits was
supported. In a two-way analysis of variance predicting grit from
education and age, both education, F(3, 682) ⫽11.54, p⬍.001,
␩
p
2
⫽.05, and age, F(4, 682) ⫽15.32, p⬍.001, ␩
p
2
⫽.08, were
significant predictors. When conscientiousness was added as a
covariate to the above model, both education, F(3, 657) ⫽10.63,
p⬍.001, ␩
p
2
⫽.05, and age, F(4, 657) ⫽8.45, p⬍.001, ␩
p
2
⫽.05,
remained significant predictors. Further, when neuroticism, agree-
ableness, extraversion, and openness to experience were added to
this analysis of covariance model as additional covariates, both
education, F(3, 653) ⫽11.48, p⬍.001, ␩
p
2
⫽.05, and age, F(4,
653) ⫽6.94, p⬍.001, ␩
p
2
⫽.04, remained significant predictors.
As illustrated in Figure 3, post hoc comparisons revealed that
individuals who had completed only “some college” were lower in
grit than any other group, and individuals who had earned an
Associate’s degree or a graduate degree were higher in grit than
individuals with a Bachelor’s degree. Figure 4 shows that grit was
lowest among 25- to 34-year-olds and highest among those 65
years and older.
Similarly, grit had incremental predictive validity for number of
lifetime career changes over and beyond age, conscientiousness,
and other Big Five traits. Because the distribution of lifetime
career changes was skewed right (M⫽2.25, SD ⫽2.04), we
performed a median split to compare individuals with high versus
low career changes. We also standardized all continuous predictor
variables prior to analysis to allow for a more intuitive understand-
ing of odds ratios (ORs). In a binary logistic regression predicting
high versus low career change from grit, age, and all Big Five
traits, grit was the only significant predictor (OR ⫽0.65, ␤⫽
⫺.44, p⫽.001). Individuals who were a standard deviation higher
in grit than average were 35% less likely to be frequent career
changers.
Study 3
Studies 1 and 2 established an association between grit and
educational attainment in two diverse sample of adults. Because
we were interested in predicting performance among high achiev-
ers, Study 3 tested whether grit was associated with cumulative
GPA among undergraduates at an elite university. Further, using
SAT scores as a measure of general mental ability, we tested
whether grit would be orthogonal to intelligence and, therefore,
explain variance in GPA over and beyond that explained by
intelligence.
Method
Participants. Participants were 139 undergraduate students
(69% women, 31% men) majoring in psychology at the University
of Pennsylvania. The average SAT score of this participant pool
was 1,415, a score achieved by fewer than 4% of students who take
the SAT.
Procedure and measures. Participants were recruited through
an e-mail invitation sent to approximately 350 psychology majors
in fall 2002. The invitation emphasized the voluntary and confi-
dential nature of the study and provided a website address where
participants could complete the Grit Scale and report additional
information, including current GPA, expected year of graduation,
gender, and SAT scores. Following Frey and Detterman’s (2004)
study, we used SAT scores as a measure of general mental ability.
Results and Discussion
Gritty students outperformed their less gritty peers: Grit scores
were associated with higher GPAs (r⫽.25, p⬍.01), a relation-
ship that was even stronger when SAT scores were held constant
(r⫽.34, p⬍.001). As we expected, SAT scores were also related
to GPA (r⫽.30, p⬍.001).
It is interesting to note that grit was associated with lower SAT
scores (r⫽⫺.20, p⬍.03), suggesting that among elite under-
graduates, smarter students may be slightly less gritty than their
peers. This finding was somewhat surprising given that Ackerman
and Heggestad (1997) found conscientiousness and IQ to be or-
thogonal. However, our result is consistent with that of Moutafi,
Furnham, and Paltiel (2005), who found in a large sample of job
applicants that conscientiousness and general intelligence were
inversely correlated at r⫽⫺.24. It is possible, as Moutafi et al.
have suggested, that among relatively intelligent individuals, those
who are less bright than their peers compensate by working harder
and with more determination.
Study 4
The question of what predicts success in the most challenging
environments is particularly important to military decision makers.
1093
PERSEVERANCE AND PASSION

The United States Military Academy, West Point, graduates more
than 900 new officers annually, about 25% of the new lieutenants
required by the Army each year. Admission to West Point is
extremely competitive. Candidates must receive a nomination
from a member of Congress or from the Department of the Army.
They are then evaluated on their academic, physical, and leader-
ship potential. Specifically, admission to West Point depends
heavily on a Whole Candidate Score, a weighted average of SAT
scores, class rank, demonstrated leadership ability, and physical
aptitude. Even with such a rigorous admissions process, about 1 in
20 cadets drops out during the first summer of training.
In Study 4, we expected grit to predict retention over the first
summer and, among those cadets who remained, military and
academic GPA 1 year later. Given the especially rugged experi-
ence of the summer regimen, we anticipated that grit would predict
retention better than would self-control. We expected grit to be
unrelated to IQ (as measured by SAT scores) or to physical
aptitude.
Method
Participants. Participants were 1,218 of 1,223 freshman cadets
who entered the United States Military Academy, West Point, in July
2004. This group was typical of recent West Point classes in terms of
gender (16% women, 84% men), ethnicity (77% Caucasian, 8%
Asian, 6% Hispanic, 6% Black, 1% American Indian, and 2% other
ethnicity), and age (M⫽19.05 years, SD ⫽1.1).
Procedure. Participants completed questionnaires during a
routine institutional group testing activity on the 2nd and 3rd days
after arrival to West Point in June 2004. The test administrator
informed cadets that participation in this study was voluntary and
that the information provided would be kept confidential. Sepa-
rately, official records were obtained for other data.
Measures
Grit. In the current sample, the Grit Scale had an internal
reliability coefficient of ␣⫽.79.
Self-control. The Brief Self-Control Scale (BSCS; Tangney,
Baumeister, & Boone, 2004) contains 13 items endorsed on a
5-point scale, where 1 ⫽not like me at all and 5 ⫽very much like
me (e.g., “I have a hard time breaking bad habits” and “I do certain
things that are bad for me, if they are fun”). In the current sample,
the BSCS had an internal reliability coefficient of ␣⫽.81.
Whole Candidate Score. The Whole Candidate Score is used
in conjunction with other information to admit applicants to West
Figure 3. Grit as a function of educational attainment, controlling for age and Big Five traits in Study 2
participants. Error bars represent 95% confidence intervals of the mean.
1094 DUCKWORTH, PETERSON, MATTHEWS, AND KELLY

Point. The Whole Candidate Score is a weighted composite of high
school rank; SAT score; Leadership Potential Score, which reflects
participation in extracurricular activities; and Physical Aptitude
Exam, a standardized physical exercise evaluation.
Summer retention. Summer retention was coded as a dichot-
omous variable where 1 ⫽retained and 0 ⫽separated as of the
first day of the fall semester. In the current sample, 94.2% of cadets
completed the summer training (n⫽1,152), and 5.8% dropped out
(n⫽71). To examine the individual effects of grit, self-control,
and other predictors on retention, we conducted separate binary
logistic regressions with retention as the dependent variable. For
each predictor, we report beta, which represents the change in the
log odds of retention due to a unit change in the predictor, and the
OR, which in the case of continuous predictor variables represents
the change in the odds of retention associated with a one-unit
change in the predictor.
Academic GPA. Academic GPA was calculated in spring 2005
as the cumulative average of grades in academic subjects.
Military Performance Score (MPS). MPS was calculated in
spring 2005 from performance ratings from military program
activities during the summer and academic year as well as grades
for military science courses. Activities completed at higher levels
of responsibility were weighted more heavily.
Results and Discussion
Grit was not related to Whole Candidate Score (r⫽.02, ns) nor
any of its components: SAT score (r⫽⫺.05, ns), high school class
rank (r⫽⫺.04, ns), Leadership Potential Score (r⫽.05, ns), and
Physical Aptitude Exam (r⫽.01, ns). As predicted, grit was
related to self-control (r⫽.63, p⬍.001).
Grit predicted completion of the rigorous summer training pro-
gram better than any other predictor. We conducted separate
binary logistic regression analyses predicting summer retention
from grit, self-control, and Whole Candidate Score. Predictor
variables were standardized before regression analysis to allow for
a more intuitive understanding of ORs. Cadets who were a stan-
dard deviation higher than average in grit were more than 60%
more likely to complete summer training (␤⫽.48, OR ⫽1.62,
p⬍.001), whereas cadets who scored a standard deviation above
average in self-control were only 50% more likely to complete the
summer course (␤⫽.41, OR ⫽1.50, p⬍.01). Whole Candidate
Figure 4. Grit as a function of age (in years), controlling for educational attainment and Big Five traits in Study
2 participants. Error bars represent 95% confidence intervals of the mean.
1095
PERSEVERANCE AND PASSION

Score, the composite score used by West Point to admit candi-
dates, did not predict summer retention (␤⫽.09, OR ⫽1.09, ns).
Further, when all three predictors were entered simultaneously into
a binary logistic regression model, grit (␤⫽.44, OR ⫽1.55, p⬍
.01) but neither self-control (␤⫽.12, OR ⫽1.13, ns) nor Whole
Candidate Score (␤⫽.11, OR ⫽1.11, ns) predicted retention
robustly.
However, grit was not the best predictor of cumulative first-year
academic GPA and MPS among cadets who remained at West
Point. Grit predicted MPS (r⫽.19, p⬍.001) about as well as did
self-control (r⫽.21, p⬍.001). However, self-control was a better
predictor of GPA (r⫽.13, p⬍.001) than was grit (r⫽.06, p⬍
.05); p⬍.001 for the difference in correlation coefficients. Even
more striking was the superior prediction by Whole Candidate
Score of both MPS (r⫽.42, p⬍.001) and GPA (r⫽.64, p⬍
.001). When Whole Candidate Score and self-control were held
constant, grit continued to predict MPS (partial r⫽.09, p⬍.01)
but not GPA (partial r⫽⫺.01, ns).
These findings support Galton’s (1892) contention that there is
a qualitative difference between minor and major accomplish-
ments. Earning good grades during the academic year at West
Point requires regulating effort from moment to moment, primarily
by resisting “hourly temptations” to procrastinate, daydream, or
indulge in unproductive diversions. Self-control may be constantly
taxed, but the workload is manageable and there is little temptation
to give up altogether. Staying at West Point through the first
summer training (sometimes referred to as Beast Barracks), in
contrast, calls upon a different sort of fortitude. Beast Barracks is
deliberately engineered to test the very limits of cadets’ physical,
emotional, and mental capacities. A reasonable response to the
unrelenting dawn-to-midnight trials of Beast Barracks would be to
exchange the goal of graduating from West Point for a more
manageable goal such as graduating from a liberal arts college.
Study 5
Study 4 showed that grittier cadets were more likely to complete
their first summer of training at West Point. Study 5 was a
replication and extension of Study 4 in which we tested whether
grit had incremental predictive validity for summer attrition over
and beyond Big Five Conscientiousness.
Method
On the day after their arrival at West Point in June 2006, 1,308
of 1,310 cadets in the Class of 2010 completed questionnaires.
Participants completed the Grit Scale (observed ␣⫽.79) and the
9-item Conscientiousness subscale of the Big Five Inventory (John
& Srivastava, 1999; observed ␣⫽.82). Official records including
Whole Candidate Scores and retention data were obtained in
September 2006. Summer retention for the Class of 2010 (95.3%)
was higher than for the Class of 2008 (94.2%).
Results and Discussion
Whole Candidate Score was related to conscientiousness (r⫽
.12, p⬍.001) but not to grit (r⫽.03, ns). As in Study 2, grit and
conscientiousness were highly related (r⫽.64, p⬍.001). Nev-
ertheless, summer retention was predicted better by grit (␤⫽.31,
OR ⫽1.36, p⬍.02) than by either conscientiousness (␤⫽.09,
OR ⫽1.09, ns) or Whole Candidate Score (␤⫽.02, OR ⫽1.02,
ns). When all three predictors were entered simultaneously into a
binary logistic regression model, grit predicted summer retention
(␤⫽.39, OR ⫽1.47, p⬍.03), but Conscientiousness (␤⫽⫺.17,
OR ⫽0.85, ns) and Whole Candidate Score (␤⫽.04, OR ⫽1.04,
ns) did not.
Study 6
Study 6 was a prospective, longitudinal investigation involving
finalists in the 2005 Scripps National Spelling Bee. This annual
competition involves thousands of children in the United States,
Europe, Canada, New Zealand, Guam, Jamaica, Puerto Rico, the
U.S. Virgin Islands, The Bahamas, and American Samoa. In 2005,
273 newspapers sponsored spelling bee programs in their commu-
nities; the champion of each sponsor’s spelling bee advanced to the
national competition in Washington, DC. The two outcomes of
interest were final round reached in the national competition and
number of prior competitions in which children participated.
We were interested in this competition for two reasons. First, we
were curious about the importance of grit to exceptional extracur-
ricular accomplishment—to avocational rather than vocational
pursuits. Second, Study 6 enabled us to test a hypothesis about the
mechanism of grit. We expected the effect of grit on final round to
be mediated by time on task, in this context operationally defined
as the number of hours spent studying for the current spelling bee
final competition and, in addition, the number of prior final com-
petitions entered.
Method
Participants. Of 273 finalists in the 2005 Scripps National
Spelling Bee, 175 (64%) elected to participate by returning signed
child and parent consent forms and self-report questionnaires in
April and May 2005, prior to the June final competition. Partici-
pants ranged in age from 7 to 15 years old (M⫽13.20, SD ⫽
1.23); 48% were girls, and 52% were boys. Of these 175 partici-
pants, 79 volunteered to take a verbal IQ measure over the tele-
phone. We were able to administer the verbal IQ test to 66
participants before the competition; the remaining 13 verbal IQ
tests were administered during the 2 weeks following the compe-
tition. Participants did not differ from nonparticipants on age,
gender, final round reached, or number of prior competitions.
Similarly, there were no systematic differences on these variables
between participants who completed the verbal IQ measure and
those who did not, nor between participants who completed the
verbal IQ measure before the final competition and those who
completed it afterward.
Measures
Grit. The Grit Scale had an internal reliability coefficient of
␣⫽.80 in this study.
Self-control. The BSCS had an internal reliability of ␣⫽.88
in the current sample.
Verbal IQ. The Similarities subtest of the Wechsler Intelli-
gence Scale for Children–III (Wechsler, 1991) was delivered over
the telephone to a subgroup of participants who indicated a will-
1096 DUCKWORTH, PETERSON, MATTHEWS, AND KELLY

ingness to be called for this purpose. The Wechsler Intelligence
Scale for Children–III is a widely used measure of general intel-
ligence for children aged 6 to 16 years. The subtest comprises 19
word pairs that participants are asked to compare and bring under
a single concept (e.g., “Red and blue. How are they similar? How
are they the same?”) We chose the Similarities subtest in part
because it correlates highly with verbal IQ (r⫽.85) and full scale
IQ (r⫽.78). In addition, we considered that most participants
would be memorizing words in preparation for the spelling bee
competition and that this verbal subtest would be least confounded
with vocabulary learned explicitly for the competition. Wechsler
(1958) pointed out that
while a certain degree of verbal comprehension is necessary for even
minimal performance, sheer word knowledge need only be a minor
factor. More important is the individual’s ability to perceive the
common elements of the terms he is asked to compare and, at higher
levels, his ability to bring them under a single concept. (p. 73)
The Similarities subtest has a published average split-half reliabil-
ity coefficient of .81 and an average test–retest stability coefficient
of .81. The current sample scored more than a standard deviation
above average (mean scaled score ⫽13.83, SD ⫽2.38).
Study time. Participants reported how many hours per day they
studied for the spelling bee finals on weekdays and, separately,
how many hours per day they studied on weekends. Participants
studied for the spelling bee an average of 2.25 hr per day (SD ⫽
2.04) on weekends and 1.34 hr per day (SD ⫽1.50) on weekdays.
We interpret the higher mean and standard deviation for weekend
studying as indicating that on Saturdays and Sundays finalists had
fewer school-related and extracurricular obligations and, therefore,
more discretionary time for studying. Because weekend and week-
day studying hours were highly correlated (r⫽.62, p⬍.001), and
because of the greater variance in weekend studying hours, in
subsequent analyses we used weekend studying hours only.
Final round. The final competition of the Scripps National
Spelling Bee is an oral competition conducted in rounds until only
one speller remains. Rounds end after all spellers among those
remaining in competition have spelled for the judges one new
word. Beginning in Round 3, if a speller misspells a word, he or
she is eliminated. During the 2005 competition, the winner cor-
rectly spelled words during all 19 rounds, two children tied for
second place by correctly spelling words during the first 18 rounds,
and so on. For all participants in our study, we recorded the
number of rounds completed by a finalist prior to elimination.
Prior competitions. We recorded from records provided by the
Scripps National Spelling Bee the total number of times a child has
participated in the final competition. Of the 175 participants in our
study, 133 (76%) were first-time finalists, 34 (19%) had competed
once before, 4 had competed twice before (2%), and 4 had com-
peted in three prior competitions (2%).
Results and Discussion
The two dependent variables of interest in Study 6 —final round
and prior competitions—were ordinal. We therefore used ordinal
regression models (Scott, Goldberg, & Mayo, 1997) to test the
effect of each predictor. We report the statistical significance and
OR for each covariate, where the OR represents the likelihood of
being in the next category per unit increase in the covariate. To
facilitate interpretation of ORs, we standardized grit, self-control,
and verbal IQ scores before fitting ordinal regression models. Not
surprisingly, older children were more likely to have participated
in prior competitions ( p⬍.02), and there was a trend toward older
children advancing farther in competition ( p⬍.08). We therefore
include age as a covariate in all subsequent analyses.
As shown in Figure 5, grit predicted advancement to higher
rounds in competition. In an ordinal regression model with final
round as the dependent variable, grit (␤⫽.34, OR ⫽1.41, p⬍
.04) and age (␤⫽.28, OR ⫽1.32, p⬍.05) were significant
predictors, indicating that finalists with grit scores a standard
deviation above the mean for same-aged finalists were 41% more
likely to advance to further rounds.
Despite the sizable correlation between grit and self-control
(r⫽.66, p⬍.001), self-control (␤⫽.04, OR ⫽1.04, ns) failed
to predict performance when age was controlled for (␤⫽.27,
OR ⫽1.31, p⬍.06). When grit, self-control, and age were entered
as predictors of final round, grit (␤⫽.62, OR ⫽1.86, p⬍.01) and
age (␤⫽.29, OR ⫽1.33, p⬍.05) were the only significant
positive predictors.
Verbal IQ also predicted final round. In an ordinal regression
model with final round as the dependent variable, verbal IQ (␤⫽
.80, OR ⫽2.22, p⬍.003) but not age (␤⫽.20, OR ⫽1.22, ns)
was a significant predictor. Grit and verbal IQ were not strongly
related (r⫽.02, ns). Thus, we were surprised that in an ordinal
regression model predicting final round from grit, verbal IQ, and
age, grit was not a statistically significant predictor of final round.
Specifically, the regression coefficient for grit in this model was
␤⫽.19 and its OR was 1.21, suggesting that finalists who were a
standard deviation above the mean for finalists of the same age and
verbal IQ might be 21% more likely to advance to further rounds.
Because listwise deletion of participants who did not complete the
verbal IQ measure reduced the model’s degrees of freedom by
more than half, we speculate that grit would have been a signifi-
cant predictor had we obtained verbal IQ data on more children
and thus preserved statistical power. However, we cannot rule out
other explanations.
Gritty finalists outperformed their less gritty peers at least in
part because they studied longer. Specifically, weekend hours of
practice mediated the relationship between grit and final round.
Figure 5. Final round reached as a function of ranked quartiles of grit,
self-control, and IQ among National Spelling Bee finalists in Study 6. All
predictors are controlled for age.
1097
PERSEVERANCE AND PASSION

Several criteria must be met for a variable to be considered a
mediator: The independent variable must predict the mediator, the
mediator must predict the dependent variable when controlling for
the independent variable, and the independent variable must pre-
dict the dependent variable. In addition, mediation implies that
association of the independent variable and the dependent variable
is reduced in the presence of the mediator (see Figure 6).
We showed in the above ordinal regression model that grit
indeed predicted final round when holding age constant. Second,
in a simultaneous multiple regression with study time as the
dependent variable and age as a covariate, we found grit was a
significant predictor (␤⫽.28, p⬍.001). Finally, in a simulta-
neous ordinal regression model predicting final round, study time
(␤⫽.30, OR ⫽1.35, p⬍.001) and age (␤⫽.32, OR ⫽1.38, p⬍
.03) were both significant predictors, but grit (␤⫽.16, OR ⫽1.17,
ns) was not. Thus, although we do not know of an accepted test for
the significance of the decrement in the grit regression coefficient,
this set of regression analyses is consistent with weekend hours of
practice at least partially mediating the relationship between trait-
level grit and performance.
We followed a similar procedure to show that experience in
prior final competitions was also a partial mediator between grit
and final round. In contrast to self-control and verbal IQ, grit
robustly postdicted participation in prior national spelling bee final
competitions. In an ordinal regression model with prior competi-
tions as the dependent variable, grit (␤⫽.48, OR ⫽1.62, p⬍.02)
was a significant predictor when age was controlled for (␤⫽.30,
OR ⫽1.35, p⬍.07). The OR for grit was 1.62, indicating that
finalists who were a standard deviation above same-aged peers in
grit score were 62% more likely to have competed in an incre-
mental prior competition. In contrast, self-control only approached
significance as a postdiction variable ( p⫽.11), and verbal IQ
seemed entirely unrelated ( p⫽.82). In a simultaneous ordinal
regression predicting 2005 final round, number of prior competi-
tions (␤⫽1.21, OR ⫽3.36, p⬍.001) remained a significant
covariate when age was controlled for (␤⫽.20, OR ⫽1.22, ns),
but grit (␤⫽.20, OR ⫽1.22, ns) did not.
Study 6 suggests that gritty children work harder and longer
than their less gritty peers and, as a consequence, perform better.
The prospective, longitudinal design of this study gives us some
confidence that, indeed, an enduring personality characteristic we
call grit is driving the observed correlations with success outcomes
rather than the other way around. However, in all of the current
studies, it remains possible that a third variable drove both success
outcomes and responses to the Grit Scale. We discuss this limita-
tion in detail in the General Discussion.
General Discussion
Across six studies, individual differences in grit accounted for
significant incremental variance in success outcomes over and
beyond that explained by IQ, to which it was not positively related.
As summarized in Table 3, grit accounted for more variance in
outcomes than commonly observed for Big Five Conscientious-
ness. In Studies 1 and 2, we found that grittier individuals had
attained higher levels of education than less gritty individuals of
the same age. Older individuals tended to be higher in grit than
younger individuals, suggesting that the quality of grit, although a
stable individual difference, may nevertheless increase over the
life span. As we expected, grittier individuals made fewer career
changes than less gritty peers of the same age. In Study 3, under-
graduates at an elite university who scored higher in grit also
earned higher GPAs than their peers, despite having lower SAT
scores. In Studies 4 and 5, grit was a better predictor of first
summer retention at West Point than was either self-control or a
summary measure of cadet quality used by the West Point admis-
sions committee. However, among the cadets who persisted to the
fall semester, self-control was a better predictor of academic
performance. In our final study, grittier competitors in the Scripps
National Spelling Bee outranked less gritty competitors of the
same age, at least in part because of more accumulated practice.
In our view, achievement is the product of talent and effort, the
latter a function of the intensity, direction, and duration of one’s
exertions toward a goal. We speculate that individual differences
in the intensity dimension of effort are salient and, therefore,
described by many adjectives in the English language (e.g., ener-
getic, conscientious, dutiful, responsible, lazy). Whereas the
amount of energy one invests in a particular task at a given
moment in time is readily apparent both to oneself and to others,
the consistency of one’s long-term goals and the stamina with
which one pursues those goals over years may be less obvious.
Similarly, whereas the importance of working harder is easily
apprehended, the importance of working longer without switching
objectives may be less perceptible. Hence, it is possible that fewer
adjectives describe individual differences in the dimensions of
direction and duration of effort, both molar rather than molecular
concepts. This disparity in lexical representation may have resulted
in the omission of the grit construct from measures of Big Five
Conscientiousness.
As an example, consider two children learning to play the piano.
Assume that both children are equally talented in music and,
therefore, improve in skill at the same rate per unit effort. Assume
further that these children are matched in the intensity of effort
they expend toward musical training. Intensity in this case is
described by the extent to which attention is fully engaged during
practice time. Duration and direction of effort, on the other hand,
are described by the number of accumulated hours devoted to
musical study and, crucially, the decision to deepen expertise in
piano rather than to explore alternative instruments. Our findings
suggest that children matched on talent and capacity for hard work
may nevertheless differ in grit. Thus, a prodigy who practices
intensively yet moves from piano to the saxophone to voice will
likely be surpassed by an equally gifted but grittier child.
Twenty years prior to our research, the Personal Qualities
Project examined the effect on success in college of over 100
preadmissions variables, including expert ratings of community
Figure 6. Model of study time and prior spelling bee experience as
mediators between Grit and final round in Study 6.
1098 DUCKWORTH, PETERSON, MATTHEWS, AND KELLY

activities, athletic achievement, creative talent, personal statement
quality, talent in music, and leadership experience (Willingham,
1985). One quality, follow-through, captured the essence of grit:
“The follow-through rating involved evidence of purposeful, con-
tinuous commitment to certain types of activities versus sporadic
efforts in diverse areas” (p. 213). High school students who re-
ceived a 5-point rating for follow-through were involved for sev-
eral years in at least two different activities and, in each of these
domains, demonstrated significant advancement and achievement
(e.g., editor of the yearbook and captain of the varsity softball
team). Students who received a 1-point rating had no evidence of
a multiple-year involvement in any activity.
Among more than 3,500 participants attending nine different
colleges, follow-through was a better predictor than all other
variables, including SAT scores and high school rank, of whether
a student would achieve a leadership position in college. Follow-
through was also the single best predictor of significant accom-
plishment in science, art, sports, communications, organization, or
some other endeavor (Willingham, 1985, p. 213). Follow-through
was the third best predictor, after SAT scores and high school rank,
of who would graduate with academic honors. It is important to
note that ratings of follow-through were better than ratings of
overall high school extracurricular involvement in predicting suc-
cess outcomes. Consistent with our finding that grit was not
positively associated with IQ, follow-through was orthogonal to
SAT scores (␸⫽.01). Given that college grades are only modestly
correlated with adult success (Hoyt, 1966), we wonder whether
follow-through or, as we prefer to call it, grit, may in fact matter
more than IQ to eventual success in life.
Limitations
We see four major limitations to the current research. First, we
relied exclusively on a self-report questionnaire to measure grit.
The limitations of self-report instruments are well-known (e.g.,
Lucas & Baird, 2006; Paulhus, 1991). The Grit Scale is relatively
transparent and, therefore, particularly vulnerable to social desir-
ability bias. Although confidentiality was assured in all six studies,
some participants may have been more motivated than others by
the desire to look good. Studies 1 and 2 involved self-reported
educational attainment and GPA, respectively, and it could be
argued that in these studies, social desirability bias drove observed
positive correlations between outcomes and the Grit Scale. Against
this is the fact that grit was associated with educational attainment
when controlling for conscientiousness and other Big Five factors,
the scores of which would also reflect social desirability bias.
Further, how do we account for the sizable correlations between
grit and objective measures of success in Studies 4, 5, and 6? In
fact, if significant, social desirability bias suggests that the true
correlations between grit and achievement are higher than we
observed, strengthening our conclusions rather than weakening
them. Still, we believe that a multimethod, multisource approach to
measurement is preferable, and we plan to develop informant
report, content analysis, and biodata measures of grit in future
studies.
The second major limitation of the current work is that the Grit
Scale asks respondents to reflect on their characteristic approach to
goals, setbacks, and challenges (e.g., “Setbacks don’t discourage
me”). Such items, even when worded in the present tense, neces-
sitate retrospective reflection. A case could be made that the sum
total of our research is to show that past behavior predicts future
behavior. The strong version of this complaint would suggest there
is no stable individual difference called grit. Rather, there is
consistency of behavior across time, possibly reflecting consis-
tency of situation (Mischel, 1968). Of course, this claim questions
whether such a thing as personality exists at all. A discussion of
this debate is beyond the scope of this article, but we point out that
in Studies 3, 4, 5, and 6, we examined how individuals in a similar
situation respond differently.
An additional concern is that Studies 3, 4, 5, and 6 involved
select populations in which there was restriction of range on IQ,
resulting in attenuation of correlations between IQ and both grit
and achievement. Our findings suggest that among high achievers,
there is likely some degree of restriction of range on grit as well.
Thus, we may have underestimated the correlations among grit,
IQ, and achievement. Further, by focusing our attention on indi-
vidual differences among relatively high-IQ individuals, we have
necessarily limited the external validity of our investigation. We
are hesitant to extrapolate from the conclusions made here to less
talented populations, but our suspicion is that grit, like IQ, is of
ubiquitous importance in all endeavors in which success requires
months or even years of sustained effort and interest. To the extent
that the temptation to give up is greater for individuals of modest
ability, grit may matter more, not less. We found no significant
interactions between IQ and grit in Studies 3, 4, 5, and 6, but
recognize that more heterogeneous samples are needed to test
Table 3
Summary of Major Findings
Sample Study design Success measure
% variance
in success
explained by grit
rbetween
grit and
IQ
Study 1: Adults aged 25 and older Cross-sectional Educational attainment 4.8
a***
Study 2: Adults aged 25 and older Cross-sectional Educational attainment 4.8
a***
Study 3: Ivy League undergraduates Cross-sectional Grade point average 6.3
**
⫺.20
b*
Study 4: West Point cadets Class of 2008 Longitudinal (3 months) Retention 3.9
c***
⫺.05
b*
Study 5: West Point cadets Class of 2010 Longitudinal (3 months) Retention 1.4
c*
⫺.08
b**
Study 6: National Spelling Bee finalists Longitudinal (1 month) Final round 3.8
a,c*
.02
Note.
*
p⬍.05.
**
p⬍.01.
***
p⬍.001.
a
Controlling for age.
b
IQ measured by SAT score.
c
Percentage of variance estimated using Nagelkerke R
2
.
1099
PERSEVERANCE AND PASSION

whether IQ moderates the relationship between grit and achieve-
ment.
Finally, the current findings do not shed light on how grit relates
to other variables known to predict achievement, such as self-
efficacy (Bandura, 1977), optimistic explanatory style (Seligman
& Schulman, 1986), and locus of control (Rotter, 1966). Future
research is necessary to test whether these other variables are distal
factors that have an effect on achievement via grit. One possibility
is that the propensity to pursue long-term goals with perseverance
and passion may be determined in part by beliefs about one’s
capabilities, attributions of positive and negative events, and be-
liefs about the relative influence of external causes. However, it is
also possible that the effects of these other variables on perfor-
mance are mediated by some other mechanism and that grit is a
mere epiphenomenon. More generally, further research is needed
to elucidate the specific processes or behaviors set in motion by
grit and other variables associated with achievement.
Implications
In a qualitative study of the development of world-class pianists,
neurologists, swimmers, chess players, mathematicians, and sculp-
tors, Bloom (1985) noted that “only a few of [the 120 talented
individuals in the sample] were regarded as prodigies by teachers,
parents, or experts” (p. 533). Rather, accomplished individuals
worked day after day, for at least 10 or 15 years, to reach the top
of their fields. Bloom observed that in every studied field, the
general qualities possessed by high achievers included a strong
interest in the particular field, a desire to reach “a high level of
attainment” in that field, and a “willingness to put in great amounts
of time and effort” (p. 544). Similarly, in her study of prodigies
who later made significant contributions to their field, Winner
(1996) concluded, “Creators must be able to persist in the face of
difficulty and overcome the many obstacles in the way of creative
discovery....Drive and energy in childhood are more predictive
of success, if not creativity, than is IQ or some other more
domain-specific ability” (p. 293).
The qualitative insights of Winner (1996), Bloom (1985), and
Galton (1892), coupled with evidence gathered by the current
investigation and its forerunners, suggest that, in every field, grit
may be as essential as talent to high accomplishment. If substan-
tiated, this conclusion has several practical implications: First,
children who demonstrate exceptional commitment to a particular
goal should be supported with as many resources as those identi-
fied as “gifted and talented.” Second, as educators and parents, we
should encourage children to work not only with intensity but also
with stamina. In particular, we should prepare youth to anticipate
failures and misfortunes and point out that excellence in any
discipline requires years and years of time on task. Finally, liberal
arts universities that encourage undergraduates to sample broadly
should recognize the ineluctable trade-off between breadth and
depth. To paraphrase Benjamin Franklin, the goal of an education
is not just to learn a little about a lot but also a lot about a little.
References
Ackerman, P. L., & Heggestad, E. D. (1997). Intelligence, personality, and
interests: Evidence for overlapping traits. Psychological Bulletin, 121,
219 –245.
Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral
change. Psychological Review, 84, 191–215.
Barrick, M. R., & Mount, M. K. (1991). The Big Five personality dimen-
sions and job performance: A meta-analysis. Personnel Psychology, 44,
1–26.
Baum, J. R., & Locke, E. A. (2004). The relationship of entrepreneurial
traits, skill, and motivation to subsequent venture growth. Journal of
Applied Psychology, 89, 587–598.
Bloom, B. (1985). Developing talent in young people. New York: Ballan-
tine Books.
Bridgeman, B., McCamley-Jenkins, L., & Ervin, N. (2000). Predictions of
freshman grade-point average from the revised and recentered SAT I:
Reasoning Test (College Board Report 2000 –2001). New York: College
Entrance Examination Board.
Cassidy, T., & Lynn, R. (1989). A multifactorial approach to achievement
motivation: The development of a comprehensive measure. Journal of
Occupational Psychology, 62, 301–312.
Cattell, J. M. (1903). A statistical study of eminent men. Popular Science
Monthly, 359 –377.
Cattell, R. B., & Butcher, H. J. (1968). The prediction of achievement and
creativity. Oxford, England: Bobbs-Merrill.
Chamorro-Premuzic, T., & Furnham, A. (2005). Personality and intellec-
tual competence. Mahwah, NJ: Erlbaum.
Chown, S. M. (1959). Personality factors in the formation of occupational
choice. British Journal of Educational Psychology, 29, 23–33.
Costa, P. T., & McCrae, R. R. (1992). Revised NEO Personality Inventory
(NEO-PI-R) and NEO Five-Factor Inventory (NEO-FFI) professional
manual. Odessa, FL: Psychological Assessment Resources.
Cox, C. M. (1926). Genetic studies of genius: Vol. 2. The early mental
traits of three hundred geniuses. Stanford, CA: Stanford University
Press.
Desrochers, S., & Dahir, V. (2000). Ambition as a motivational basis of
organizational and professional commitment: Preliminary analysis of a
proposed career advancement ambition scale. Perceptual and Motor
Skills, 91(2), 563–570.
Ericsson, K. A., & Charness, N. (1994). Expert performance: Its structure
and acquisition. American Psychologist, 49, 725–747.
Fergusson, D. M., Horwood, L. J., & Ridder, E. M. (2005). Show me the
child at seven II: Childhood intelligence and later outcomes in adoles-
cence and young adulthood. Journal of Child Psychology & Psychiatry,
46(8), 850 – 858.
Frey, M. C., & Detterman, D. K. (2004). Scholastic assessment or g? The
relationship between the scholastic assessment test and general cognitive
ability. Psychological Science, 15(6), 373–378.
Galton, F. (1892). Hereditary Genius: An inquiry into its laws and conse-
quences. London: Macmillan.
Gartner, W. B., Gatewood, E., & Shaver, K. G. (1991). Reasons for starting
a business: Not-so-simple answers to simple questions. In G. E. Hills &
R. W. LaForge (Eds.), Research at the marketing/entrepreneurship
interface (pp. 90 –101). Chicago: University of Illinois at Chicago.
Goldberg, L. R. (1990). An alternative “description of personality”: The
Big-Five factor structure. Journal Of Personality And Social Psychol-
ogy, 59, 1216 –1229.
Goldberg, L. R. (1992). The development of markers for the Big-Five
factor structure. Psychological Assessment, 4, 26 – 42.
Gottfredson, L. S. (1997). Why g matters: The complexity of everyday life.
Intelligence, 24, 79 –132.
Hartigan, J., & Wigdor, A. (1989). Fairness in employment testing: Va-
lidity generalization, minority issues, and the general aptitude test
battery. Washington, DC: National Academy Press.
Hollenbeck, J. R., Williams, C. L., & Klein, H. J. (1989). An empirical
examination of the antecedents of commitment to difficult goals. Journal
of Applied Psychology, 74, 18 –23.
Hough, L. M. (1992). The “Big Five” personality variables— construct
1100 DUCKWORTH, PETERSON, MATTHEWS, AND KELLY

confusion: Description versus prediction. Human Performance, 5(1–2),
139 –155.
Howe, M. J. A. (1999). Genius explained. New York: Cambridge Univer-
sity Press.
Hoyt, D. P. (1966). College grades & adult accomplishment: A review of
research. Educational Record, 47, 70 –75.
James, W. (1907, March 1). The energies of men. Science, 25, 321–332.
John, O., & Srivastava, S. (1999). The Big Five trait taxonomy: History,
measurement, and theoretical perspectives. In L. A. Pervin & O. P. John
(Eds.), Handbook of personality: Theory and research (pp. 102–138).
New York: Guilford Press.
Kuncel, N. R., Hezlett, S. A., & Ones, D. S. (2001). A comprehensive
meta-analysis of the predictive validity of the graduate record examina-
tions: Implications for graduate student selection and performance. Psy-
chological Bulletin, 127, 162–181.
Kuncel, N. R., Hezlett, S. A., & Ones, D. S. (2004). Academic perfor-
mance, career potential, creativity, and job performance: Can one con-
struct predict them all? Journal of Personality and Social Psychology,
86, 148 –161.
Langlie, T. A. (1938). Intelligence-test scores and scholarship. Journal of
Higher Education, 9, 449 – 450.
Latham, G. P., & Pinder, C. C. (2005). Work motivation theory and
research at the dawn of the twenty-first century. Annual Review of
Psychology, 56, 485–516.
Lucas, R. E., & Baird, B. M. (2006). Global self-assessment. In M. Eid &
E. Diener (Eds.), Handbook of multimethod measurement in psychology
(pp. 29 –42). Washington, DC: American Psychological Association.
Lufi, D., & Cohen, A. (1987). A scale for measuring persistence in
children. Journal of Personality Assessment, 51(2), 178 –185.
McClelland, D. C. (1961). The achieving society. Oxford, England: Van
Nostrand.
McClelland, D. C., Koestner, R., & Weinberger, J. (1992). How do self-
attributed and implicit motives differ? New York: Cambridge University
Press.
McCrae, R. R., & Costa, P. T., Jr. (1987). Validation of the five-factor
model of personality across instruments and observers. Journal of Per-
sonality and Social Psychology, 52, 81–90.
McCrae, R. R., Costa, P. T., Jr., de Lima, M. P., Simo˜es, A., Ostendorf, F.,
& Angleitner, A., et al. (1999). Age differences in personality across the
adult life span: Parallels in five cultures. Developmental Psychology, 35,
466 – 477.
Mischel, W. (1968). Personality and assessment. New York: Wiley.
Moutafi, J., Furnham, A., & Paltiel, L. (2005). Can personality factors
predict intelligence? Personality and Individual Differences, 38(5),
1021–1033.
Neisser, U., Boodoo, G., Bouchard, T. J. J., Boykin, A. W., Brody, N., &
Ceci, S. J., et al. (1996). Intelligence: Knowns and unknowns. American
Psychologist, 51, 77–101.
Paulhus, D. L. (1991). Measurement and control of response bias. In J. P.
Robinson, & P. R. Shaver (Eds.), Measures of personality and social
psychological attitudes (Vol. 1; pp. 17–59). San Diego, CA: Academic
Press.
Paunonen, S. V., & Ashton, M. C. (2001). Big five predictors of academic
achievement. Journal of Research in Personality, 35(1), 78 –90.
Paunonen, S. V., & Jackson, D. N. (2000). What is beyond the big five?
Plenty! Journal of Personality, 68(5), 821– 835.
Rotter, J. B. (1966). Generalized expectancies for internal versus external
control of reinforcement. Psychological Monographs: General & Ap-
plied, 80(1), 1–28.
Scott, S. C., Goldberg, M. S., & Mayo, N. E. (1997). Statistical assessment
of ordinal outcomes in comparative studies. Journal of Clinical Epide-
miology, 50, 45–55.
Seligman, M. E., & Schulman, P. (1986). Explanatory style as a predictor
of productivity and quitting among life insurance sales agents. Journal
of Personality and Social Psychology, 50, 832– 838.
Simonton, D. K. (1994). Greatness: Who makes history and why. New
York: Guilford Press.
Srivastava, S., John, O. P., Gosling, S. D., & Potter, J. (2003). Develop-
ment of personality in early and middle adulthood: Set like plaster or
persistent change? Journal of Personality and Social Psychology, 84,
1041–1053.
Swanson, J. L. (1999). Stability and change in vocational interests. In
M. L. Savickas & A. R. Spokane (Eds.), Vocational interests (pp.
135–158). Palo Alto, CA: Davies-Black Publishing.
Tangney, J. P., Baumeister, R. F., & Boone, A. L. (2004). High self-control
predicts good adjustment, less pathology, better grades, and interper-
sonal success. Journal of Personality, 72(2), 271–322.
Terman, L. M., & Oden, M. H. (1947). The gifted child grows up:
Twenty-five years’ follow-up of a superior group. Oxford, England:
Stanford University Press.
Tett, R. P., Jackson, D. N., & Rothstein, M. (1991). Personality measures
as predictors of job performance: A meta-analytic review. Personnel
Psychology, 44(4), 703–742.
Tupes, E. C., & Christal, R. E. (1992). Recurrent personality factors based
on trait ratings. Journal of Personality, 60(2), 225–251.
Twenge, J. M., Zhang, L., & Im, C. (2004). It’s beyond my control: A
cross-temporal meta-analysis of increasing externality in locus of con-
trol, 1960 –2002. Personality and Social Psychology Review, 8(3), 308 –
319.
Vallerand, R. J., Blanchard, C., Mageau, G. A., Koestner, R., Ratelle, C.,
&Le´onard, M., et al. (2003). Les passions de l’A
ˆme: On obsessive and
harmonious passion. Journal of Personality and Social Psychology, 85,
756 –767.
Webb, E. (1915). Character and intelligence. British Journal of Psychol-
ogy, 1(3), 99.
Wechsler, D. (1940). Nonintellective factors in general intelligence. Psy-
chological Bulletin, 37, 444 – 445.
Wechsler, D. (1958). The measurement and appraisal of adult intelligence
(4th ed.). Baltimore: Williams & Wilkins.
Wechsler, D. (1991). Wechsler Intelligence Scale for Children—Third
Edition. San Antonio, TX: The Psychological Corporation.
Willingham, W. W. (1985). Success in college: The role of personal
qualities and academic ability. New York: College Entrance Examina-
tion Board.
Winner, E. (1996). Gifted children: Myths and realities. New York: Basic
Books.
Received October 31, 2005
Revision received December 27, 2006
Accepted January 10, 2007 䡲
1101
PERSEVERANCE AND PASSION