Motivational clusters and performance in a real-life setting
Nicolas Gillet Æ Robert J. Vallerand Æ
Published online: 7 January 2009
Ó Springer Science+Business Media, LLC 2009
Abstract The present research investigated whether
assessing adolescent elite athletes’ motivational proﬁles at
the beginning of the season would allow us to predict their
subsequent performance over the course of a competitive
season. In two studies, athletes completed the French ver-
sion of the Sport Motivation Scale (Brie
re et al., Int J Sport
Psychol 26:465–489, 1995) at the beginning of the season.
Objective levels of performance were recorded for the
following season, as well as for the season prior to ques-
tionnaire completion. In Study 1, the sample comprised a
total of 170 French junior national tennis players
(M = 13.42 years). Results revealed the presence of a
four-cluster solution, reﬂecting different levels of autono-
mous and controlled motivations. Results revealed that
controlling for past performance, athletes with the least
self-determined motivational proﬁle obtained lower levels
of subsequent tennis performance than those in the three
other (more self-determined motivational) clusters. In
Study 2, there were a total of 250 French junior national
fencers aged 15 years. Results revealed a three-cluster
solution very similar to that of Study 1. In addition, in line
with Study 1, results revealed that the least self-determined
motivational proﬁle led to the lowest level of performance.
Overall, these ﬁndings suggest that cluster analysis is
useful in the understanding of the complex relationship
between motivation and performance in elite sport.
Keywords Motivation Performance Sport
Cluster analysis Self-determination theory
Motivation has been repeatedly reported as one of the most
important contributors to high-level performance. For
instance, in sport, athletes’ success is often explained as a
function of motivation (Gould et al. 2002; Williams and
Krane 2001). In that light, it is not surprising that a con-
siderable amount of research in the last 20 years has
focused on sport motivation (see Chatzisarantis et al. 2003;
Roberts 2001; Vallerand et al. 1987). Among the different
theories proposed to explain motivated behavior and out-
comes in sport, self-determination theory (SDT; Deci and
Ryan 2000; Ryan and Deci 2007) has been found to be
especially inﬂuential. Numerous studies have supported
postulates from this theory in the sport setting (see Hagger
and Chatzisarantis 2007; Vallerand 2007a).
Among the numerous postulates of the theory, SDT
posits the existence of three major types of motivational
constructs, namely intrinsic motivation, extrinsic motiva-
tion, and amotivation. Intrinsic motivation refers to
engaging in activities for themselves, out of pleasure, fun,
and enjoyment. On the other hand, extrinsic motivation
refers to engaging in activities for outcomes that are separate
from the activity. Four forms of extrinsic motivation have
been proposed. First, external regulation involves engaging
in an activity to obtain rewards or avoid punishment. Sec-
ond, introjected regulation refers to behaviors performed to
avoid guilt and internal pressure and entails the
N. Gillet (&) E. Rosnet
Laboratoire de Psychologie Applique
Champagne-Ardenne, 57, rue Pierre Taittinger, 51096 Reims
R. J. Vallerand
partement de psychologie, Laboratoire de Recherche sur le
Comportement Social, Universite
Montreal, QC, Canada
Motiv Emot (2009) 33:49–62
internalization of past external controls. Third, in identiﬁed
regulation, individuals engage out of choice in the activity
that is not interesting per se. Finally, the last form of
extrinsic motivation is integrated regulation. It deals with
behaviors that while not emitted out of fun, are nevertheless
fully internalized in the individual’s self and value system.
In addition to intrinsic and extrinsic motivation, a third
motivational construct described by Deci and Ryan (1985)is
amotivation. When amotivated, individuals do not perceive
contingencies between their actions and subsequent out-
comes. Amotivation can be seen as the relative lack of
motivation to engage in a certain behavior (Vallerand 1997).
Self-determination theory further posits that these dif-
ferent forms of motivation can be aligned on a continuum
of increasing self-determination from amotivation to
external, introjected, identiﬁed, and integrated regulations,
and to intrinsic motivation. Furthermore, because self-
determination is a prerequisite for adaptive functioning
(Deci 1980), consequences should be increasingly positive
as one moves from amotivation to intrinsic motivation.
Much research supports this hypothesis in a variety of life
contexts (see Deci and Ryan 2000; Vallerand 1997), as
well as in sports (for reviews, see Vallerand 2007a, b;
Vallerand and Losier 1999). Thus, the more autonomous or
self-determined forms of motivation (intrinsic motivation
as well as identiﬁed and integrated regulations) have been
found to lead to a number of positive cognitive (e.g.,
concentration; e.g., Brie
re et al. 1995), affective (e.g.,
positive affect, ﬂow; e.g., Kowal and Fortier 1999; Pelletier
et al. 1995), and behavioral (e.g., persistence; e.g., Pelletier
et al. 2001; Sarrazin et al. 2002) outcomes in sport settings.
Conversely, forms of controlled or non self-determined
motivation (introjected and external regulations) and
especially amotivation have been found to typically yield
negative outcomes (e.g., anxiety, distraction, dropping out,
negative affect; e.g., Brie
re et al. 1995; McDonough and
Crocker 2007; Pelletier et al. 2001).
Previous research examining the relationships between
motivation and outcomes has used one of two strategies: (1)
assessing the relationship of each type of motivation inde-
pendently (e.g., intrinsic motivation) or (2) using the self-
determination index which entails giving weights to each
construct as a function of placement on the self-determi-
nation continuum and summing all products into one score
(e.g., Fortier et al. 1995; Grolnick and Ryan 1987; Guay and
Vallerand 1997; Ryan and Connell 1989; Sarrazin et al.
2002). While the ﬁrst strategy is limited because a number
of motives are typically at play in life settings (e.g., Pintrich
2003), the second one may not be optimal either as it
imposes a unique proﬁle based on theory where a high self-
determined motivational proﬁle is theoretically posited to
be the ideal one. Indeed, SDT assumes that a truly self-
determined motivational proﬁle exists and that such a
proﬁle should lead to the most positive outcomes. Yet, as
Vallerand (1997) suggested, a number of motivational
proﬁles may exist in actual life domains such as sport (e.g.,
high levels of both autonomous and controlled motivation)
and it is possible that in certain contexts, proﬁles that
deviate from the one proposed by SDT, lead to equivalent or
even more positive outcomes. It would thus appear impor-
tant to better understand the different clusters that exist in
certain life contexts such as in competitive sports, using
SDT as a theoretical framework, and determine how each
cluster relates to outcomes such as performance.
While increasing research has looked at the motivational
clusters that emerge in sport (e.g., Harwood et al.
Hodge et al. 2008; McNeill and Wang 2005), exercise
(Cumming and Hall 2004; Matsumoto and Takenaka 2004;
Wang and Biddle 2001), and physical education (Ntou-
manis 2002; Wang et al. 2002) settings, researchers have
typically used a mix of theoretical approaches in deter-
mining the clusters within the purview of the same study.
Such a methodological strategy makes it difﬁcult to
determine how SDT constructs uniquely contribute to the
clusters. Furthermore, other research has used both SDT
constructs and outcomes to determine the nature of the
clusters (e.g., Ntoumanis 2002). Because both motivation
and outcomes are used in creating the clusters, it becomes
impossible to determine from such clusters how motivation
independently predicts outcomes.
One study that has looked at emerging clusters using
strictly SDT constructs is that of Vlachopoulos et al.
(2000). These authors examined motivational proﬁles with
two samples of sport participants (n = 590 and n = 555).
Cluster analysis results revealed the presence of two pro-
ﬁles. The ﬁrst was characterized by a high self-determined
motivational proﬁle (high autonomous but low controlled
motivation). The second comprised athletes who had both
high autonomous and controlled motivation. Results with
outcomes revealed that participants in the second cluster
reported signiﬁcantly higher levels of enjoyment, effort,
positive affect, strength of intention to continue, and sat-
isfaction but lower levels of negative affect than
participants in the ﬁrst cluster. These results do not fully
corroborate SDT’s predictions. Indeed, athletes character-
ized by a mixed motivational proﬁle experienced more
positive outcomes than those characterized by a self-
determined motivational proﬁle. SDT would have pre-
dicted just the opposite.
The above study by Vlachopoulos et al. (2000)is
important because it would appear to be the only study to
use strictly SDT motivational variables to identify clusters
in sport and how these relate to outcomes. However, three
points are in order with respect to this study. First, par-
ticipants of this study were non elite sport participants.
Thus, we do not know what types of clusters may exist with
50 Motiv Emot (2009) 33:49–62
other populations such as elite athletes. More than two
clusters might be expected with elite performers as a
number of motivational patterns may explain participation
in a competitive setting. Second, the outcomes used in the
Vlachopoulos et al.’s (2000) study were self-reports of
affect, cognition, and behavioral intentions. Objective
levels of performance were not assessed in the Vlachopo-
ulos et al.’s study. Finally, all variables were assessed at
the same point in time. It is thus difﬁcult to determine if
motivational clusters can predict changes in outcomes,
such as performance, that may take place over time.
We believe that it is important to study performance as it
represents one of the key outcomes in sport, especially at
the elite level and yet, performance has been sorely
neglected in past motivation research (Roberts 1992;
Vallerand 2001, 2007a; Vallerand and Rousseau 2001).
Past research in other life contexts such as education has
shown that high levels of autonomous motivation toward
education lead to high academic performance (e.g., Boiche
et al. 2008; Burton et al. 2006; Gottfried et al. 1994). For
instance, in three studies, Ratelle et al. (2007) have
investigated the nature of students’ motivational proﬁles
toward education and examined how these clusters differed
on a variety of outcomes, including performance. Results
of two studies with high school students (Studies 1 and 2)
revealed three distinct clusters: (1) a low autonomous–high
controlled motivation cluster; (2) a moderate autonomous–
moderate controlled motivation cluster; and (3) a high
autonomous–high controlled motivation cluster. Thus, a
truly self-determined motivational cluster was not obtained
with high school students. Results also showed that high
school students with the least self-determined motivational
cluster (i.e., low autonomous–high controlled motivation)
had lower grades than those in the two other clusters that
did not differ between them (Ratelle et al. 2007, Study 2).
These ﬁndings with performance were replicated in Study
3 with college students (although a truly self-determined
motivational cluster was found in this study).
In light of the above, there were three purposes to the
present research. First, using cluster analysis, we sought to
identify the motivational proﬁles that exist in elite junior
tennis (‘‘Study 1’’) and fencing (‘‘Study 2’’). We limited
ourselves to fencing and tennis for homogeneity purposes
and also because reliable objective performance data were
available for these sport activities. This person-oriented
approach is interesting because it also provides opportu-
nities for researchers to determine the number of athletes
characterized by distinct motivational proﬁles while cor-
relation or regression analyses do not (Ratelle et al. 2007).
Furthermore, it should provide additional information on
motivational proﬁles as they actually exist in an achieve-
ment context such as sport and not simply as theoretically
proposed by SDT and exempliﬁed by the self-
determination index that imposes the high autonomous–
low controlled motivation conﬁguration.
A second purpose was to relate the motivational clusters
to indices of objective performance. Such a strategy should
allow us to determine if certain motivational proﬁles are
more conducive to performance than others. It should be
underscored that research in sport has yielded equivocal
ﬁndings, with both autonomous (e.g., Biddle and Brooke
1992) and controlled forms of motivation (e.g., Chantal
et al. 1996) being positively related to performance.
However, such research has typically used a cross-sectional
design and has not used cluster analysis. Clearly, a pro-
spective design would be necessary to more clearly
determine the role of motivational clusters in the prediction
of changes in subsequent objective sport performance. This
constituted the third purpose of the present research.
We ﬁrst examined athletes’ motivational proﬁles in a
sample of junior elite tennis players. In line with past
research using cluster analysis in education and sport (e.g.,
Ratelle et al. 2007; Vlachopoulos et al. 2000), it was
expected that at least three clusters would be uncovered:
(1) a high autonomous–low controlled motivation cluster;
(2) a high autonomous–high controlled motivation cluster;
and (3) a low autonomous–high controlled motivation
cluster. We then examined the links between these moti-
vational proﬁles and tennis performance. In line with SDT
and past studies using cluster analysis in education (e.g.,
Ratelle et al. 2007), it was hypothesized that the least self-
determined cluster would display the lowest level of per-
formance. This is because research with athletes from a
variety of sports (e.g., Brie
re et al. 1995; Pelletier et al.
1995) has shown positive relationships between autono-
mous forms of motivation and concentration which may
represent one the most important predictors of performance
(Vallerand 2007a). Furthermore, athletes who engage in
sport and put forth effort mostly when told to do so by the
coach (external regulation) or who only go through the
motions with little conviction (amotivation) may not work
as hard overall, and thus should improve and perform less,
than those who engage in sports because they love it and
feel that it is their personal choice to do so.
Participants and procedure
The sample was comprised of 170 French junior national
tennis players (71 females and 99 males). Participants were
Motiv Emot (2009) 33:49–62 51
either 13 or 14 years of age, with a mean age of
13.42 years (SD = .49 year). The number of years that
they had been practicing tennis ranged from 3 to 11, with a
mean of 6.94 years (SD = 1.48 years). They also reported
practicing tennis for an average of 9.33 hours a week
(SD = 3.70 hours).
This study received ethical approval from the French
Tennis Federation. Athletes and their family were told that
they were completely free to participate or not. Parental
consent was obtained and conﬁdentiality was ensured. In
order to ensure that the participants would be elite per-
formers, only players who were among the top 150 of
France for their respective age group were contacted. Out
of the 300 athletes, 39 could not be reached. Thus, before
the beginning of the tennis season, a total of 261 athletes
received a questionnaire by mail to assess their motivation
for the activity at the beginning of the season and were
asked to complete it. A prepaid reply envelope was also
provided. One hundred and seventy questionnaires were
returned, for a 65% return rate. Objective performance was
later secured from the French Tennis Federation.
The French version of the Sport Motivation Scale (SMS;
re et al. 1995) was used to measure athletes’ motiva-
tion toward tennis. The SMS contains a total of 28 items,
with four items per subscale. These items assess the con-
structs of intrinsic motivation to know, intrinsic motivation
to experience stimulation, intrinsic motivation toward
accomplishments, identiﬁed regulation, introjected regula-
tion, external regulation and amotivation. Participants
responded to items based on a seven-point Likert scale
ranging from ‘‘Does not correspond at all’’ (1) to ‘‘Corre-
sponds exactly’’ (7). Because, we did not have any speciﬁc
hypotheses about the different types of intrinsic motivation,
the three intrinsic motivation subscales were combined in
an index of intrinsic motivation. While past research has
conﬁrmed the validity and reliability of the SMS (e.g., Li
and Harmer 1996; Pelletier et al. 1995, 2007; Pelletier and
Sarrazin 2007), some authors have criticized its factorial
structure (e.g., Mallett et al. 2007b; Martens and Webber
2002; Reimer et al. 2002). Consequently, we have con-
ducted a conﬁrmatory factor analysis (CFA) with the
present data. The results appear below.
Five objective performance indices were used in the pres-
ent study. First, the ratio between the number of victories
and the number of matches played during the tennis season
following questionnaire completion was used as a perfor-
mance measure (i.e., Performance 1). For example, the
performance for a player who has won 4 of 10 matches
would be equal to .40. This tennis performance was
obtained via the French Tennis Federation. Participants
played an average of 60.6 matches (SD = 21.2 matches).
Second, we used the same method to calculate a total
performance score for the two seasons following ques-
tionnaire completion (i.e., Performance 2; M = 117.5
matches; SD = 41.1). We also obtained two other mea-
sures of performance with respect to the scores as
determined by the French Tennis Federation for the tennis
season following data collection (i.e., Performance 3), and
for the next two seasons (i.e., Performance 4). With the
Federation scores, the higher the level of the opponent, the
more points are won following a win, whereas losses
against players ranked below are penalized. Finally, the
ratio between the number of victories and the number of
matches played (M = 63.1 matches; SD = 18.4) during
the tennis season prior to data collection was used as a
performance score for the previous season.
A two-stage cluster analysis procedure was used (Gore
2000; Hair et al. 1998) because it allows researchers to
constitute clusters with high internal and external homo-
geneities (Hair and Black 2000). Whereas hierarchical
cluster analysis represents a mean of obtaining the optimal
number of clusters, non-hierarchical k-means cluster
analysis is used as a way of further ﬁne-tuning the pre-
liminary cluster solution trough an iterative process by
minimizing the within-cluster variance and by maximizing
the between-cluster variance. Of importance is the fact that
cluster analysis is more adaptive than the typical median-
or mean-split procedures where much of the variance is lost
in the process of creating groups. A better understanding of
such motivational clusters is important because it under-
scores the role of motivational regulations considered
jointly rather than in isolation (Ntoumanis 2002). We
decided to use a cluster-analytic approach rather than
testing for interaction effects in a multiple regression
framework because we believe that it would have been
difﬁcult to analyze all potential conﬁgurations involving
the different forms of motivation in a regression analysis.
The univariate distributions of the various variables were
examined for normality (i.e., via skewness and kurtosis
52 Motiv Emot (2009) 33:49–62
values and the Kolmogorov–Smirnov statistic). The per-
formance variables were normally distributed except for
Performance 1 (Kolmogorov–Smirnov statistic, P \ .05;
skewness =-1.29, kurtosis = 2.73). Similarly, the moti-
vational variables were normally distributed in all instances,
except for the amotivation variable (Kolmogorov–Smirnov
statistic, P \ .01; skewness = 3.24, kurtosis = 12.06). Past
research has also revealed that the amotivation subscale
often displays a skewed distribution where low means are
reported. This is to be expected as individuals who engage in
the activity should report low levels of amotivation. How-
ever, non-normality in the data does not pose problem as the
amotivation subscale has been repeatedly found to be the
best predictor of various outcomes, including behavioral
persistence both in education (Vallerand and Bissonnette
1992; Vallerand et al. 1997) and sports (Pelletier et al.
2001). Furthermore, recent cluster analyses using the amo-
tivation subscale (e.g., Ratelle et al. 2007, Studies 1–3) have
revealed that results were not inﬂuenced by the non-nor-
mality of the data. In the present study, the correlations
among the motivation variables ranged between -.21 and
.53 (see Table 1). Thus, multicollinearity was not a problem
because only correlation values of .90 and above display
signiﬁcant collinearity (Hair et al. 1998).
We next examined the factor structure for the French
version of the SMS (Brie
re et al. 1995) via CFA. While the
number of parameters relative to the number of participants
imposes some limits on the model being tested, results of
the CFA yielded acceptable ﬁt indices, v
(318) = 485.60,
P \ .001, CFI = .92, NNFI = .91, RMSEA = .052 [.041;
.062]. All factor loadings were signiﬁcant. Furthermore, the
Cronbach alphas were all adequate (between .74 and .91)
and inspection of the correlations among the SMS sub-
scales provided support for the self-determination
continuum. Speciﬁcally, all correlations among the SMS
subscales revealed a simplex-like pattern, with stronger
positive correlations between adjacent factors on the self-
determination continuum and weaker correlations between
more distal factors (see Table 1). The present results are in
agreement with those obtained by Brie
re et al. (1995) and
Pelletier et al. (1995) and provide additional support for the
construct validity of the French version of the SMS.
To identify subgroups of athletes based on their motivation,
a cluster analysis was conducted. Cluster analysis allows
researchers to examine different solutions, and then select
the solution that best ﬁts the data (Cumming et al. 2002;
Hodge and Petlichkoff 2000). First, a hierarchical cluster
analysis using Ward’s linkage method with the squared
Euclidian distance measure was performed. Ward’s hierar-
chical method was chosen because it trivializes the within-
cluster differences found in other methods (Aldenderfer and
Blashﬁeld 1984). Hierarchical cluster analysis is an
exploratory data reduction technique designed to create
groups in such a way that participants in the same cluster
display a similar motivational proﬁle (Jobson 1992). The
clustering variables were intrinsic motivation, identiﬁed
regulation, introjected regulation, external regulation, and
amotivation. We used the raw scores because all variables
shared the same metric (i.e., a 7-point Likert scale). The
agglomeration coefﬁcient and dendrograms suggested that a
four-cluster solution was the most appropriate.
In the second stage, a k-means cluster analysis using the
cluster centers resulting from the hierarchical seed points
was conducted to validate the four-cluster solution. The
results of the hierarchical method were conﬁrmed because
the ﬁnal centroids in the k-means analysis were similar to
the initial seed points. Means of the motivation subscales
for the four-cluster solution are reported in Table 2 and
Fig. 1 displays the motivational subscales as a function of
clusters. Results from chi-square analyses revealed that the
proportion of gender in each cluster did not differ [(Cluster
Table 1 Correlations among the study variables (‘‘Study 1’’ )
1. Intrinsic motivation
2. Identiﬁed regulation .50**
3. Introjected regulation .36** .53**
4. External regulation .14 .42** .44**
5. Amotivation -.21* .09 .13 .26**
6. Previous performance .10 -.05 -.08 -.07 -.16*
7. Performance 1 .07 -.05 -.04 -.15* -.16* .33**
8. Performance 2 .08 -.05 -.06 -.16* -.19* .38** .83**
9. Performance 3 (Log) .10 -.01 -.02 -.09 -.26* .31** .77** .58**
10. Performance 4 (Log) .13 -.05 -.04 -.10 -.26* .41** .67** .75** .69**
* P \ .05, ** P \ .001
Motiv Emot (2009) 33:49–62 53
1 had 13 females and 17 males); (Cluster 2 had 31 females
and 39 males); (Cluster 3 had 20 females and 27 males);
and (Cluster 4 had 7 females and 16 males)].
A one-way multivariate analysis of variance (MANO-
VA) was conducted with the ﬁve types of motivation as
dependent variables and the four clusters as the indepen-
dent variable in order to identify the motivational content
of each cluster. Results revealed signiﬁcant differences
among the four groups, F(15, 448) = 36.92, P \ .001,
= .52. A one-way ANOVA was conducted on each
dependent variable as a follow-up to the MANOVA. The
ANOVAs revealed a number of signiﬁcant differences
among the four clusters [for intrinsic motivation, F(3,
166) = 14.66, P \ .001, g
= .21, identiﬁed regulation,
F(3, 166) = 60.62, P \ .001, g
= .52, introjected regu-
lation, F(3, 166) = 87.40, P \ .001, g
= .61, external
regulation, F(3, 166) = 55.28, P \ .001, g
= .50, and
amotivation, F(3, 166) = 28.58, P \ .001, g
= .34, see
Table 2 for the complete picture of all signiﬁcant differ-
ences among the four clusters]. Overall, these differences
support the distinction among the four clusters.
Scores on the various motivation subscales allow us to
label the four clusters. Participants in the ﬁrst cluster rep-
resented 18% of the sample (n = 30) and included athletes
who displayed high levels of intrinsic motivation, identiﬁed
regulation, introjected regulation, and external regulation,
but low levels of amotivation. Thus, this cluster was
labeled the high autonomous–high controlled cluster (high
AU–high C group). The second cluster represented 41% of
the sample (n = 70) and included athletes whose motiva-
tional proﬁle was characterized by relatively moderate
levels of intrinsic motivation and identiﬁed regulation, but
relatively low levels of controlled motivation and amoti-
vation. This second cluster was thus labeled the moderate
autonomous–low controlled motivation group (mod AU–
low C group). The third cluster represented 28% of the
sample (n = 47) and included athletes who displayed a
high level of autonomous motivation, and low to moderate
levels of controlled motivation and amotivation. This third
cluster was thus labeled the high autonomous–moderate
controlled motivation group (high AU–mod C group).
Finally, the fourth cluster represented 13% of the sample
(n = 23) and included participants who displayed moder-
ate levels of autonomous motivation, but moderate to high
levels of controlled motivation and amotivation. In fact,
this group obtained the highest level of amotivation. Thus,
Table 2 Means for the study variables as a function of clusters (‘‘Study 1’’ )
Cluster Cluster 1
‘‘High AU–high C’’
(n = 30)
‘‘Mod AU–low C’’
(n = 70)
‘‘High AU–mod C’’
(n = 47)
‘‘Mod AU–high C’’
(n = 23)
Intrinsic motivation 5.86
14.66 .001 .21
Identiﬁed regulation 5.49
60.62 .001 .52
Introjected regulation 6.04
87.40 .001 .61
External regulation 4.29
55.28 .001 .50
28.58 .001 .34
Self-determination index 9.50
15.25 .001 .22
Previous performance .56
1.67 .18 .03
Performance 1 .57
5.89 .001 .10
Performance 2 .57
5.61 .001 .09
Performance 3 (Log) 6.98
4.59 .01 .08
Performance 4 (Log) 7.61
7.63 .001 .12
Note For each dependent variable, means with different subscripts indicate a signiﬁcant difference at P \ .05 using Fisher’s LSD test
AU autonomous, C controlled
IM IDR INR EXR AMO
Cluster 1 "High AU-High C"
Cluster 2 "Mod AU-Low C"
Cluster 3 "High AU-Mod C"
Cluster 4 "Mod AU-High C"
Fig. 1 Motivation subscales as a function of clusters (‘‘Study 1’’).
AU autonomous, C controlled, IM intrinsic motivation, IDR identiﬁed
regulation, INR introjected regulation, EXR external regulation, AMO
54 Motiv Emot (2009) 33:49–62
this cluster was labeled the moderate autonomous–high
controlled motivation group (mod AU–high C group).
For exploratory purposes, we next compared the four
clusters on their scores on the self-determination index.
Means for the clusters appear in Table 2. Results from an
ANOVA, F(3, 166) = 15.25, P \ .001, g
= .22, followed
by post hoc tests using Fisher’s LSD revealed that partici-
pants in the mod AU–high C group (M = 4.73, SD = 3.72)
exhibited signiﬁcantly lower scores on the self-determina-
tion index compared to those in the high AU–high C group
(M = 9.50, SD = 2.54), the high AU–mod C group
(M = 8.82, SD = 2.03), and the mod AU–low C group
(M = 7.84, SD = 2.89). Furthermore, the mod AU–low C
group was lower than the high AU–high C and high AU–
mod C clusters which did not differ between them.
A series of ANOVAs and post hoc tests using Fisher’s
LSD were conducted to determine whether the motiva-
tional proﬁle groups differed signiﬁcantly with respect to
subsequent performance. A prior analysis on performance
during the previous season revealed no difference among
clusters, F(3, 166) = 1.67, P = .18, g
= .03. The results
of the ANOVA on Performance 1 (ratio of win/loss for the
ﬁrst season following questionnaire completion) revealed a
signiﬁcant effect for clusters, F(3, 166) = 5.89, P \ .001,
= .10. Post hoc tests indicated that the least self-deter-
mined cluster, the mod AU–high C group, obtained
signiﬁcantly lower levels of performance (M = .46) than
all other clusters that did not differ among them. Overall,
athletes in the mod AU–high C cluster lost between 11 and
12% more matches than those in the three other clusters
during the following (ﬁrst) tennis season.
Additional analyses were also conducted with the three
other performance indices. We conducted ANOVAs in
order to assess the difference among the four clusters with
respect to (1) the ratio of victories to matches played in the
following two seasons (Performance 2), (2) the scores as
determined by the French Tennis Federation given to each
athlete for the subsequent season (Performance 3; such
scores take into consideration the ranking of opponents that
the player has defeated, as well as lost to) and (3) and
scores from the French Tennis Federation for the next two
seasons (Performance 4). All results were the same as those
mentioned above for Performance 1 (all ps \ .01 and effect
sizes varied from .08 to .12).
All the performance scores
for each cluster appear in Table 2.
Furthermore, a series of ANCOVAs were conducted on
all four performance variables controlling for the previous
season’s performance. The results were the same as those
of the ANOVAs (all ps \ .05 and effect sizes varied from
.07 to .11). In addition, another series of ANCOVAs were
conducted controlling for the previous season’s perfor-
mance, years of tennis experience, and hours of training per
week. This is a highly conservative test as the number of
hours of training can be inﬂuenced by motivation. Still,
once again, all results were the same as those reported
above (all ps \ .05, except for Performance 3, P = .08;
effect sizes varied from .05 to .08). Overall, these results
provide support for the validity of the present ﬁndings and
suggest that the motivational clusters allow us to explain a
sizeable portion of variance in the change of objective
performance that took place over time.
The purpose of this study was to use cluster analysis in
order to examine the nature of motivational proﬁles that
exist in elite sport and then assess the relationship between
these clusters and changes in subsequent objective perfor-
mance. The results revealed the existence of four readily
interpretable clusters: a high autonomous–high controlled
group (high AU–high C group), a moderate autonomous–
low controlled group (mod AU–low C group), a high
autonomous–moderate controlled group (high AU–mod C
group), and a moderate autonomous–high controlled group
(mod AU–high C group). Further, the present ﬁndings
showed that athletes with the least self-determined moti-
vational proﬁle at the beginning of the tennis season (i.e.,
mod AU–high C proﬁle), displayed the lowest level of
subsequent performance. Athletes in this cluster lost
between 11 and 12% more matches than those of the other
three clusters. Over the course of a 60 match season, this
amounts to winning 6 or 7 matches less than participants
with a more self-determined motivational conﬁguration.
This result on the role of motivation in performance is
particularly striking when one considers that these athletes
were among the best of their respective age group in a
country recognized for its high level of tennis development
(for instance, the International Tennis Federation world
men junior rankings for the year 2006 showed that France
was the only country to have two athletes in the top 10).
It should be noted that because of the extremely high level of
variance in the scores of the French Federation of Tennis, both the
Performance 3 and Performance 4 scores were subjected to a log
All Fs, Ps, and g
can be obtained through the authors.
We also calculated the rank of each player relative to those within
the present study (higher ranks = lower performance). Thus, based
on the scores from the French Tennis Federation, each player was
ranked from 1 to 170 for the ﬁrst season, as well as for both the ﬁrst
and second seasons combined. Then the mean rank of each cluster
was compared. Results for the ﬁrst season revealed the following
ranks: high AU–high C cluster = 77.4; mod AU–low C
group = 70.3; high AU–mod C group = 70.2; mod AU–high
C = 98.9. In line with the other results from this study, the least
self-determined cluster (mod AU–high C cluster) was found to be
Motiv Emot (2009) 33:49–62 55
However, the present ﬁndings were obtained with 13 and
14 years old elite tennis players from France. Therefore,
future research is needed to replicate the present ﬁndings
with athletes from different sport activities because as
suggested by Ratelle et al. (2007), athletes’ motivational
proﬁles might be context sensitive.
While our ﬁrst prospective investigation has provided us
with important information regarding athletes’ motiva-
tional proﬁles in elite sport, it seems imperative to conduct
an additional study with elite athletes practicing another
competitive activity. Thus, the purpose of this second study
was twofold. First, we sought to verify that the motiva-
tional conﬁgurations found in ‘‘Study 1’’ could be
generalized to another sample, this time elite fencers.
Second, we examined the relationships between these
motivational proﬁles and sport performance. According to
SDT and results from ‘‘Study 1’’, it was hypothesized that
athletes with the least self-determined motivational proﬁle
would obtain the lowest levels of subsequent fencing per-
formance, controlling for athletes’ prior performance.
Participants and procedure
The sample was composed of 250 French junior national
fencers (107 females and 143 males) aged 15 years. There
were 88 epe
e fencers, 92 foil fencers, and 70 sabre fencers
among the top 30 of France for their respective age and
weapon group. As in ‘‘Study 1’’, this second investigation
received ethical approval from athletes and the French
Fencing Federation. Each participant volunteered to com-
plete a questionnaire before the beginning of the fencing
season. Data were collected before the beginning of the
following seasons: 2001–2002, 2003–2004, 2004–2005,
2005–2006 and 2006–2007. At the end of each season, a
measure of sport performance was obtained via the French
The same scale as in ‘‘Study 1’’ was used.
Two objective measures of performance were used in the
present study. The ﬁrst performance score reﬂects the
national ranking in fencing of each athlete at the end of the
season following data collection (i.e., Performance 1).
Second, the national ranking determined by the French
Fencing Federation was also used as a performance mea-
sure for the previous season. The rankings were reversed so
that high values were indicators of a high level of perfor-
mance. Thus, the lower the national ranking (e.g., ﬁrst), the
worse the athletes’ performance was during the fencing
The internal consistency for all the subscales of the French
version of the SMS (Brie
re et al. 1995) was satisfactory
(between .73 and .87). As in ‘‘Study 1’’, an inspection of the
correlations among the SMS subscales provided support for
the self-determination continuum with stronger positive
correlations between adjacent factors on the self-determi-
nation continuum and weaker correlations between more
distal factors (see Table 3). As in ‘‘Study 1’’, the study
variables were normally distributed except for the amoti-
vation variable (Kolmogorov–Smirnov statistic, P \ .01;
skewness = 1.60, kurtosis = 1.68). Treatment of outliers
involved deleting four cases with a distance from the mean
greater than three times the value of the standard deviation.
First, scores of each motivation subscale of the SMS
(intrinsic motivation, identiﬁed regulation, introjected
regulation, external regulation, and amotivation) were
included in an exploratory cluster analysis using Ward’s
method of linkage and a squared Euclidian distance. The
agglomeration schedules and dendrograms suggested that a
three-cluster solution was the most appropriate because the
agglomeration coefﬁcient showed a large increase from
three to two clusters. Then, a k-means clustering method
Footnote 3 continued
ranked lower than the other three that did not differ among them, F(3,
166) = 2.47, P = .06, g
= .04. These results were replicated with
performance for the two seasons combined with mean ranks of 75.7,
71.7, 75.7, and 104.3 for the four clusters in that order, F(3,
166) = 2.80, P \ .05, g
= .05. What these results reveal is that
national tennis players with the least self-determined motivational
proﬁle assessed before the beginning of the ﬁrst season were ranked
signiﬁcantly lower (some 20 ranks lower after one season, and even
30 ranks lower after two seasons!) than other tennis players their own
age who had a more self-determined motivational proﬁle. Clearly
motivation matters with respect to performance!
56 Motiv Emot (2009) 33:49–62
was used with the cluster centers resulting from the hier-
archical seed points. The ﬁnal centroids are shown in
Table 4, and these were similar to the initial seed points.
Figure 2 shows the three distinct motivational proﬁles.
A one-way MANOVA was carried out with the ﬁve
forms of motivation as dependent variables and the clusters
as the independent variable in order to identify the moti-
vational content of each cluster. Results revealed
signiﬁcant differences among the three groups, F(10,
464) = 67.47, P \ .001, g
= .59. A one-way ANOVA
was conducted on each dependent variable as a follow-up
to the MANOVA. The ANOVAs were all signiﬁcant
except for the amotivation subscale. Then, Fisher’s LSD
tests revealed that the three groups were signiﬁcantly dis-
tinct from each other on intrinsic motivation, identiﬁed
regulation, and introjected regulation. For external regu-
lation, athletes in the high AU–high C displayed the
highest score, while there were no signiﬁcant differences
between those in the two other groups (see Table 4 for
more details). Chi-square analyses showed that there were
no signiﬁcant gender differences in the classiﬁcation of
athletes into the three clusters.
The ﬁrst group (mod AU–low C) comprised 44% of the
sample (n = 108) and entailed moderate levels of auton-
omous motivation as well as low levels of controlled
motivation and amotivation. The second group (mod AU–
mod C), which constituted 33% of the sample (n = 81),
included athletes whose motivational proﬁle was charac-
terized by moderate levels of both autonomous and
controlled motivation and low levels of amotivation. The
third group (high AU–high C) included 57 athletes (23% of
the sample) whose proﬁle was characterized by high levels
of both autonomous and controlled motivation and low
levels of amotivation.
Results from an ANOVA on the self-determination
index revealed a signiﬁcant effect for clusters, F(2,
243) = 14.98, P \ .001, g
= .11. Speciﬁcally, athletes in
the mod AU–low C group (M = 6.65, SD = 2.48) exhib-
ited signiﬁcantly lower scores on the self-determination
index compared to those in the high AU–high C group
Table 3 Correlations between the study variables (‘‘Study 2’’ )
Variables 1 2 3 4 5 6 7
1. Intrinsic motivation
4. External regulation .29** .42** .45**
5. Amotivation -.09 .11 .01 .17**
.06 .07 -.02 .03 -.09
7. Performance 1 .24** .17* -.05 .12* -.11 .10
* P \ .05, ** P \ .001
Table 4 Descriptive statistics for the three-cluster solution (‘‘Study 2’’ )
Cluster Cluster 1
‘‘Mod AU–low C’’
(n = 108)
‘‘Mod AU–mod C’’
(n = 81)
‘‘High AU–high C’’
(n = 57)
Intrinsic motivation 4.37
41.35 .001 .25
Identiﬁed regulation 3.13
65.44 .001 .35
Introjected regulation 3.25
181.00 .001 .60
External regulation 2.07
177.75 .001 .60
2.84 .06 .02
Self-determination index 6.65
14.98 .001 .11
Previous performance (ranks) 17.15
1.21 .30 .01
Performance 1 (ranks) 28.25
3.20 .05 .03
Note. For each dependent variable, means with different subscripts indicate a signiﬁcant difference at P \ .01 using Fisher’s LSD test
AU autonomous, C controlled
IM IDR INR EXR AMO
Cluster 1 "Mod AU-Low C"
Cluster 2 "Mod AU-Mod C"
Cluster 3 "High AU-High C"
Fig. 2 Motivation subscales as a function of clusters (‘‘Study 2’’).
AU autonomous, C controlled, IM intrinsic motivation, IDR identiﬁed
regulation, INR introjected regulation, EXR external regulation, AMO
Motiv Emot (2009) 33:49–62 57
(M = 7.98, SD = 2.38) and the mod AU–mod C group
(M = 8.50, SD = 2.27). No signiﬁcant differences
emerged between the high AU–high C cluster and the mod
AU–mod C cluster.
We next examined the links between these three moti-
vational proﬁles and sport performance. Thus, a ﬁrst
ANOVA and Fisher’s LSD tests were conducted to deter-
mine whether the motivational proﬁle groups differed
signiﬁcantly with respect to the fencing performance dur-
ing the season prior to data collection (previous
performance). The analysis was non signiﬁcant, F(2,
236) = 1.21, P = .30, g
= .01. The results of a second
ANOVA on the subsequent performance (i.e., Performance
1) revealed a signiﬁcant effect for clusters, F(2,
236) = 3.20, P \ .05, g
= .03. Speciﬁcally, athletes in
the least self-determined cluster, the mod AU–low C
group, performed lower than those in the high AU–high C.
The mod AU–low C and the mod AU–mod C groups as
well as the mod AU–mod C and the high AU–high C
groups did not differ from each other on this performance
variable. Finally, an ANCOVA was conducted on Perfor-
mance 1 controlling for the previous season’s performance.
Results were the same as those mentioned above (P \ .05,
The ﬁrst purpose of the present study was to identify ath-
letes’ motivational proﬁles in a sample of junior elite
fencers. Contrary to the ﬁrst study in tennis, the present
results revealed the existence of three readily interpretable
clusters. It is interesting that although we replicated two of
the three motivational proﬁles found in ‘‘Study 1’’ (i.e.,
mod AU–low C and high AU–high C motivational pro-
ﬁles), we found one different motivational proﬁle that
combines moderate levels of autonomous and controlled
motivations (labeled mod AU–mod C), sharing some
similarities with the high AU–mod C group in ‘‘Study 1’’ .
We believe that there were two signiﬁcant differences
between these two studies that might explain these differ-
ent results. First, the nature of the sport was different.
Indeed, the present sample was composed of junior elite
fencers, while the ﬁrst study included junior elite tennis
players. Second, participants in the ﬁrst study were among
the top 150 of France for their respective age group, while
the sample in ‘‘Study 2’’ was composed of athletes who
were even more elite, being among the top 30 of France for
their respective age and weapon group. Thus, as mentioned
above, it is possible that the development of motivational
proﬁles may vary as a function of sport activities and levels
of expertise. However, future research is needed on this
Concerning the links between the three motivational
proﬁles and sport performance, results from ‘‘Study 2’’
replicated those of ‘‘Study 1’’. Indeed, results from ‘‘Study
2’’ also revealed that athletes with the least self-determined
proﬁle (i.e., the mod AU–low C proﬁle) obtained lower
performances than those obtained by the other groups. In
line with recent investigations (e.g., Ratelle et al. 2007,
Study 3), the present ﬁndings suggest that having a moti-
vational proﬁle characterized by low to moderate levels of
autonomous and controlled motivations toward an activity
is counterproductive with respect to performance in this
The ﬁrst aim of the present research was to examine ado-
lescents’ motivational proﬁles in a real-life setting, namely
elite sport. The second aim was to study the potential
inﬂuence of these motivational proﬁles on subsequent sport
performance controlling for previous performance. In
‘‘ Study 1’’ with junior elite tennis players, results revealed
the existence of four motivational proﬁles: (1) a high
autonomous–high controlled (high AU–high C) motiva-
tional proﬁle, (2) a moderate autonomous–low controlled
(mod AU–low C) motivational proﬁle, (3) a high autono-
mous–moderate controlled (high AU–mod C) motivational
proﬁle, and (4) a least self-determined proﬁle (mod AU–
high C) characterized by moderate levels of autonomous
motivation and high levels of controlled motivation.
Results of ‘‘Study 2’’ conducted with junior elite fencers
suggested a three-cluster solution to be suitable: (1) a
moderate autonomous–low controlled (mod AU–low C)
motivational proﬁle, (2) a high autonomous–high con-
trolled (high AU–high C) motivational proﬁle (these two
proﬁles were identical to those found in the sample of
tennis players), and (3) a moderate autonomous–moderate
controlled (mod AU–mod C) motivational proﬁle, which
shared some similarities with the high AU–mod C proﬁle
among the tennis players. The results of both studies
showed that athletes characterized by the least self-deter-
mined proﬁle obtained the worst sport performance during
the subsequent season.
The present results thus provide support for the propo-
sition on the relevant inﬂuence of motivation on sport
performance (e.g., Roberts 1992; Vallerand 2001). By
ﬁnding that the least self-determined proﬁle was associated
with the worst sport performance, the results of both
studies provide support for SDT and past investigations
that showed that non-self determined motivation was
associated with negative sport outcomes such as distraction
re et al. 1995; Pelletier et al. 1995), dropout (e.g.,
Pelletier et al. 2001) and burnout (e.g., Cresswell and
58 Motiv Emot (2009) 33:49–62
Eklund 2005a, b; Lemyre et al. 2006). It is also interesting
to note that the present ﬁndings are directly in line with
those obtained by Ratelle et al. (2007) in education. In
Studies 2 and 3, these authors showed that the cluster with
the lowest level of self-determined motivation always
yielded the lowest level of performance in education.
However, Ratelle et al. did not use a prospective design in
their research. What the present research adds to the Ra-
telle et al.’s results is that using a prospective design, a low
self-determined motivation cluster was shown to predict
changes (drops) in performance that took place over the
course of one and even two seasons. However, future
research is needed to better understand the nature of the
psychological processes triggered by controlled motivation
and amotivation that may undermine performance (see
Ntoumanis et al. 2004; Pelletier et al. 1999).
Bearing in mind that introjected and external regulations
are located toward the lower end of the self-determination
continuum (Deci and Ryan 1985), one might have antici-
pated that the best performers would report high levels of
autonomous motivation and low levels of controlled
motivation. Unexpectedly, this was not the case in our
research. In fact, in ‘‘Study 2’’, the best performance in
fencing was obtained by the high AU–high C cluster. Of
interest, Chantal et al. (1996) also found that higher per-
forming athletes (i.e., title and medal holders in national
and international events) exhibited high levels of both
autonomous and controlled motivations. Although these
results were obtained in a speciﬁc cultural context (Bul-
garia was under the controlling communist regime when
the Chantal et al. study was conducted in 1989), these
ﬁndings suggest that sport performance may be related to
high levels of both autonomous and controlled motivations.
In the present research, it would thus appear that the rel-
atively high levels of autonomous motivation that the high
AU–high C cluster displayed may have served a protective
function against controlled motivation. Such was not the
case for the mod AU–high C cluster in ‘‘Study 1’’ whose
autonomy level may have been too low to protect against
the high level of controlled motivation. Alternatively, other
authors (Amabile 1993; Lepper et al. 2005) suggest that
under certain conditions (e.g., when autonomous motiva-
tion is high), controlled motivation may act in synergy with
autonomous motivation in leading to positive outcomes.
Future research is clearly needed on this issue.
What is striking in the present ﬁndings, is that a truly self-
determined motivation cluster (i.e., a high autonomous–low
controlled motivation group) was not obtained. As we
mentioned above, it would thus appear that the prevailing
context may have an important impact on the types of
clusters or motivational conﬁgurations that are prevalent in a
given life domain. Contexts that are highly competitive,
achievement driven, and potentially controlling in nature
(such as competitive sports and high school education) may
not lend themselves to high levels of pure self-determined
motivation (e.g., high autonomous–low controlled motiva-
tion). However, future research is needed to test this
hypothesis in a variety of life contexts.
An alternative explanation for the fact that a true
autonomous cluster was not found may have to do with the
scale we used, namely the French version of the SMS
re et al. 1995). Indeed, although the present results
provided support for the construct validity of this scale,
some authors (e.g., Mallett et al. 2007a, b) have suggested
that the SMS may not assess some of the SDT regulatory
categories in a theory-consistent way. In particular, these
researchers suggest that the external regulation subscale
does not assess the more controlling dimensions of external
rewards or punishments, but rather focus on elements
dealing with seeking prestige and regard (e.g., ‘‘To show
others how good I am at my sport.’’, ‘‘Because it allows me
to be well regarded by people I know.’’). It is thus possible
that high scores on the external regulation subscale may
reﬂect the player’s desire to be famous, rather than external
control per se. While these items do reﬂect controlled
motivation, the absence of highly controlling external
regulation items (and the use of less controlling ones) may
explain why we and other researchers who have used the
SMS in cluster analyses (e.g., Vlachopoulos et al. 2000)
have found the presence of a high AU–high C cluster that
was associated with positive consequences (e.g., perfor-
mance, enjoyment, effort, positive affect). Therefore,
further investigations using other measures of athletes’
motivation (e.g., the Behavioral Regulation in Sport
Questionnaire; Lonsdale et al. 2008) are needed to provide
some more deﬁnitive answers to the present issue.
This study has some limitations. First of all, while the
present study used a very informative statistical technique,
namely cluster analysis, in conjunction with a prospective
design, it should be underscored that the design used was
nevertheless correlational in nature. Consequently, we
cannot infer causality from the ﬁndings. Future research
using an experimental design should be conducted to
reproduce the present ﬁndings under controlled conditions.
Second, the present ﬁndings were obtained with 13–
15 years old elite fencers and tennis players from France.
Future research is needed to replicate the present ﬁndings
with athletes from different ages, sports, levels (i.e., pro-
fessional or Olympic athletes), cultures, and other
achievement ﬁelds (e.g., music, work, dramatic arts).
Third, a large number of athletes from the target population
(35%) did not participate in our ﬁrst study. We have no
way of knowing if such athletes display a different moti-
vational cluster and how such a cluster would relate to
performance. Fourth, it should be noted that the number of
athletes who participated in the present studies was rather
Motiv Emot (2009) 33:49–62 59
low (n = 170 and n = 250). Future research is needed to
determine if the motivational clusters uncovered in the
present research can be replicated with a larger sample of
adolescent elite athletes as well as in other achievement
settings. Fifth, only the SMS was used to derive the
motivational clusters in both studies. In light of some of the
issues raised with respect to this scale (e.g., Mallett et al.
2007a, b), future research using other instruments is rec-
ommended to replicate the present ﬁndings. Finally,
although the athletes in the various clusters did not differ
with respect to prior performance in the two present studies
and other control variables in ‘‘Study 1’’ (i.e., years of sport
experience and hours of training per week), it is never-
theless possible that they differed with respect to other
variables that could account for the different levels of
performance as a function of clusters. Future research
controlling for variables such as having a personal coach,
experiencing injuries, etc. would appear in order.
In sum, the present ﬁndings represent what would
appear to be the ﬁrst to support the role of different
motivational proﬁles in predicting changes in objective
sport performance over time. Future research is needed,
however, in order to replicate and extend these ﬁndings
thereby allowing us to better understand the motivational
processes underlying elite performance.
Acknowledgments Preparation of this manuscript was supported in
parts by grants from the Fonds Que
cois pour la Recherche sur la
et la Culture (FQRSC) and the Social Sciences Humanities
Research Council of Canada (SSHRC) to the second author.
Aldenderfer, M. S., & Blashﬁeld, R. K. (1984). Cluster analysis.
Newbury Park: Sage.
Amabile, T. M. (1993). Motivational synergy: Toward new concep-
tualizations of intrinsic and extrinsic motivation in the
workplace. Human Resource Management Review, 3, 185–201.
Biddle, S., & Brooke, R. (1992). Intrinsic versus extrinsic motiva-
tional orientation in physical education and sport. The British
Journal of Educational Psychology, 62, 247–256.
, J. C. S., Sarrazin, P. G., Grouzet, F. M. E., Pelletier, L. G., &
Chanal, J. P. (2008). Students’ motivational proﬁles and
achievement outcomes in physical education: A self-determina-
tion perspective. Journal of Educational Psychology, 100, 688–
re, N. M., Vallerand, R. J., Blais, M. R., & Pelletier, L. G. (1995).
veloppement et validation d’une mesure de motivation
que et d’amotivation en contexte sportif :
chelle de Motivation dans les Sports (EMS). International
Journal of Sport Psychology, 26, 465–489. On the development
and validation of the French form of the Sport Motivation Scale.
Burton, K., Lydon, J., D’Alessandro, D., & Koestner, R. (2006). The
differential effects of intrinsic and identiﬁed motivation on well-
being and performance: Prospective, experimental, and implicit
approaches to self-determination theory. Journal of Personality
and Social Psychology, 91, 750–762. doi:10.1037/0022-3514.
Chantal, Y., Guay, F., Dobreva-Martinova, T., & Vallerand, R. J.
(1996). Motivation and elite performance: An exploratory
investigation with Bulgarian athletes. International Journal of
Sport Psychology, 27, 173–182.
Chatzisarantis, N. L. D., Hagger, M. S., Biddle, S. J. H., Smith, B., &
Wang, J. C. K. (2003). A meta-analysis of perceived locus of
causality in exercise, sport, and physical education contexts.
Journal of Sport & Exercise Psychology, 25, 284–306.
Cresswell, S. L., & Eklund, R. C. (2005a). Motivation and burnout
among top amateur rugby players. Medicine and Science in
Sports and Exercise, 37, 469–477. doi:10.1249/01.MSS.0000
Cresswell, S. L., & Eklund, R. C. (2005b). Motivation and burnout in
professional rugby players. Research Quarterly for Exercise and
Sport, 76, 370–376.
Cumming, J., & Hall, C. (2004). The inﬂuence of goal orientation on
self-efﬁcacy in exercise. Journal of Applied Social Psychology,
34, 747–763. doi:10.1111/j.1559-1816.2004.tb02568.x.
Cumming, J., Hall, C., Harwood, C., & Gammage, K. (2002).
Motivational orientations and imagery use: A goal proﬁling
analysis. Journal of Sports Sciences, 20, 127–136. doi:10.1080/
Deci, E. L. (1980). The psychology of self-determination. Lexington:
Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-
determination in human behavior. New York: Plenum.
Deci, E. L., & Ryan, R. M. (2000). The ‘‘what’’ and ‘‘why’’ of goal
pursuits: Human needs and the self-determination of behavior.
Psychological Inquiry, 11, 227–268. doi:10.1207/S15327
Fortier, M. S., Vallerand, R. J., & Guay, F. (1995). Academic
motivation and school performance: Toward a structural model.
Contemporary Educational Psychology, 20, 257–274. doi:
Gore, P. A. (2000). Cluster analysis. In H. Tinsley & S. Brown (Eds.),
Handbook of applied multivariate statistics and mathematical
modeling (pp. 298–321). San Diego: Academic Press.
Gottfried, A. E., Fleming, J. S., & Gottfried, A. W. (1994). Role of
parental motivational practices in children’s academic intrinsic
motivation and achievement. Journal of Educational Psychol-
ogy, 86, 104–113. doi:10.1037/0022-06126.96.36.199.
Gould, D., Dieffenbach, K., & Moffett, A. (2002). Psychological
talent and their development in Olympic champions. Journal of
Applied Sport Psychology, 14, 172–204. doi:10.1080/10413200
Grolnick, W. S., & Ryan, R. M. (1987). Autonomy in children’s
learning: An experimental and individual difference investiga-
tion. Journal of Personality and Social Psychology, 52, 890–898.
Guay, F., & Vallerand, R. J. (1997). Social context, students’
motivation, and academic achievement: Toward a process model.
Social Psychology of Education, 1, 211–233. doi:10.1007/
Hagger, M. S., & Chatzisarantis, N. L. D. (2007). Editorial: Advances
in self-determination theory research in sport and exercise.
Psychology of Sport and Exercise, 8, 597–599. doi:10.1016/
Hair, J. F., & Black, W. C. (2000). Cluster analysis. In L. G. Grimm &
P. R. Yarnold (Eds.), Reading and understanding more multi-
variate statistics (pp. 147–206). Washington, DC: American
Hair, J. F., Anderson, R. E., Tatham, R. L., & Black, W. C. (1998).
Multivariate data analysis (5th ed.). Upper Saddle River:
60 Motiv Emot (2009) 33:49–62
Harwood, C., Cumming, J., & Fletcher, D. (2004). Motivational
proﬁles and psychological skills usage within elite youth sport.
Journal of Applied Sport Psychology, 16, 318–332. doi:
Hodge, K., & Petlichkoff, L. M. (2000). Goal ‘‘proﬁles’’ in sport: A
cluster analysis. Journal of Sport & Exercise Psychology, 22,
Hodge, K., Allen, J. B., & Smellie, L. (2008). Motivation in masters
sport: Achievement and social goals. Psychology of Sport and
Exercise, 9, 157–176. doi:10.1016/j.psychsport.2007.03.002.
Jobson, J. D. (1992). Applied multivariate data analysis, Vol. 2.
Categorical and multivariate methods. New York: Springer-
Kowal, J., & Fortier, M. S. (1999). Motivational determinants of ﬂow:
Contributions from self-determination theory. The Journal of
Social Psychology, 139, 355–368.
Lemyre, N. P., Treasure, D. C., & Roberts, G. C. (2006). Inﬂuence of
variability in motivation and affect on elite athlete burnout
susceptibility. Journal of Sport & Exercise Psychology, 28,
Lepper, M. R., Corpus, J. H., & Iyengar, S. S. (2005). Intrinsic and
extrinsic motivational orientations in the classroom: Age differ-
ences and academic correlates. Journal of Educational
Psychology, 97, 184–196. doi:10.1037/0022-06188.8.131.52.
Li, F., & Harmer, P. (1996). Testing the simplex assumption
underlying the Sport Motivation Scale: A structural equation
modeling analysis. Research Quarterly for Exercise and Sport,
Lonsdale, C., Hodge, K., & Rose, E. A. (2008). The Behavioral
Regulation in Sport Questionnaire (BRSQ): Instrument devel-
opment and initial validity evidence. Journal of Sport &
Exercise Psychology, 30, 323–355.
Mallett, C. J., Kawabata, M., & Newcombe, P. (2007a). Progressing
measurement in sport motivation: A response to Pelletier,
Vallerand, and Sarrazin. Psychology of Sport and Exercise, 8,
Mallett, C. J., Kawabata, M., Newcombe, P., Otero-Rorero, A., &
Jackson, S. (2007b). Sport Motivation Scale-6: A revised six-
factor sport motivation scale. Psychology of Sport and Exercise,
8, 600–614. doi:10.1016/j.psychsport.2006.12.005.
Martens, M. P., & Webber, S. N. (2002). Psychometric properties of
the Sport Motivation Scale: An evaluation with college varsity
athletes from the US. Journal of Sport & Exercise Psychology,
Matsumoto, H., & Takenaka, K. (2004). Motivational proﬁles and
stages of exercise behavior change. International Journal of
Sport and Health Science, 2, 89–96.
McDonough, M. H., & Crocker, P. R. E. (2007). Testing self-
determined motivation as a mediator of the relationship between
psychological needs and affective and behavioral outcomes.
Journal of Sport & Exercise Psychology, 29, 645–663.
McNeill, M. C., & Wang, C. K. J. (2005). Psychological proﬁles of
elite school sports players in Singapore. Psychology of Sport and
Exercise, 6, 117–128. doi:10.1016/j.psychsport.2003.10.004.
Ntoumanis, N. (2002). Motivational clusters in a sample of British
physical education classes. Psychology of Sport and Exercise, 3,
Ntoumanis, N., Pensgaard, A. M., Martin, C., & Pipe, K. (2004). An
idiographic analysis of amotivation in compulsory school
physical education. Journal of Sport & Exercise Psychology,
Pelletier, L. G., & Sarrazin, P. (2007). Measurement issues in self-
determination theory and sport. In M. S. Hagger & N. L. D.
Chatzisarantis (Eds.), Intrinsic motivation and self-determination
in exercise and sport (pp. 143–152). Champaign, IL: Human
Pelletier, L. G., Fortier, M. S., Vallerand, R. J., Tuson, K. M., Brie
N. M., & Blais, M. R. (1995). Toward a new measure of intrinsic
motivation, extrinsic motivation, and amotivation in sports: The
Sport Motivation Scale (SMS). Journal of Sport & Exercise
Psychology, 17, 35–53.
Pelletier, L. G., Dion, S., Tuson, K., & Green-Demers, I. (1999). Why
do people fail to adopt environmental behaviors? Towards a
taxonomy of environmental amotivation. Journal of Applied
Social Psychology, 29, 2481–2504. doi:10.1111/j.1559-1816.
Pelletier, L. G., Fortier, M. S., Vallerand, R. J., & Brie
re, N. M.
(2001). Associations among perceived autonomy support, forms
of self-regulation, and persistence: A prospective study. Moti-
vation and Emotion, 25, 279–306. doi:10.1023/A:1014
Pelletier, L. G., Vallerand, R. J., & Sarrazin, P. (2007). The revised
six factor Sport Motivation Scale (Mallett, Kawabata, New-
combe, Otero-Forero, & Jackson, 2007): Something old,
something new, and something borrowed. Psychology of Sport
and Exercise, 8, 615–621. doi:10.1016/j.psychsport.2007.03.006.
Pintrich, P. R. (2003). A motivational science perspective on the role
of student motivation in learning and teaching contexts. Journal
of Educational Psychology, 95, 667–686. doi:10.1037/0022-
Ratelle, C. F., Guay, F., Vallerand, R. J., Larose, S., & Sene
(2007). Autonomous, controlled, and amotivated types of
academic motivation: A person-oriented analysis. Journal of
Educational Psychology, 99, 734–746. doi:10.1037/0022-
Reimer, H., Fink, J. S., & Fitzgerald, M. P. (2002). External validity
of the Sport Motivation Scale. Avante, 8, 57–66.
Roberts, G. C. (1992). Motivation in sport and exercise: Conceptual
constraints and convergence. In G. C. Roberts (Ed.), Motivation
in sport and exercise (pp. 3–29). Champaign: Human Kinetics.
Roberts, G. C. (2001). Advances in motivation in sport and exercise.
Champaign: Human Kinetics.
Ryan, R. M., & Connell, J. P. (1989). Perceived locus of causality and
internalization: Examining reasons for acting in two domains.
Journal of Personality and Social Psychology, 57, 749–761. doi:
Ryan, R. M., & Deci, E. L. (2007). Active human nature: Self-
determination theory and the promotion and maintenance of
sport, exercise, and health. In M. S. Hagger & N. L. D.
Chatzisarantis (Eds.), Intrinsic motivation and self-determination
in exercise and sport (pp. 1–19). Champaign: Human Kinetics.
Sarrazin, P., Vallerand, R. J., Guillet, E., Pelletier, L. G., & Cury, F.
(2002). Motivation and dropout in female handballers: A 21-
month prospective study. European Journal of Social Psychol-
ogy, 57, 749–761.
Vallerand, R. J. (1997). Toward a hierarchical model of intrinsic and
extrinsic motivation. In M. P. Zanna (Ed.), Advances in
experimental social psychology (pp. 271–360). New York:
Vallerand, R. J. (2001). A hierarchical model of intrinsic and extrinsic
motivation in sport and exercise. In G. C. Roberts (Ed.),
Advances in motivation in sport and exercise (2nd ed., pp. 263–
319). Champaign: Human Kinetics.
Vallerand, R. J. (2007a). A hierarchical model of intrinsic and
extrinsic motivation for sport and physical activity. In M. S.
Hagger & N. L. D. Chatzisarantis (Eds.), Intrinsic motivation
and self-determination in exercise and sport (pp. 255–279).
Champaign: Human Kinetics.
Vallerand, R. J. (2007b). Intrinsic and extrinsic motivation in sport
and physical activity: A review and a look at the future. In G.
Tenenbaum & R. E. Eklund (Eds.), Handbook of sport psychol-
ogy (3rd ed., pp. 49–83). New York: Wiley.
Motiv Emot (2009) 33:49–62 61
Vallerand, R. J., & Bissonnette, R. (1992). Intrinsic, extrinsic, and
amotivational styles as predictors of behavior: A prospective
study. Journal of Personality, 60, 599–620. doi:10.1111/
Vallerand, R. J., & Losier, G. F. (1999). An integrative analysis of
intrinsic and extrinsic motivation in sport. Journal of Applied
Sport Psychology, 11, 142–169. doi:10.1080/104132099084
Vallerand, R. J., & Rousseau, F. L. (2001). Intrinsic and extrinsic
motivation in sport and exercise: A review using the hierarchical
model of intrinsic and extrinsic motivation. In R. Singer, H.
Hausenblas, & C. Janelle (Eds.), Handbook of sport psychology
(2nd ed., pp. 389–416). New York: Wiley.
Vallerand, R. J., Deci, E. L., & Ryan, R. M. (1987). Intrinsic
motivation in sport. In K. B. Pandolf (Ed.), Exercise and sport
sciences reviews (pp. 389–425). New York: MacMillan.
Vallerand, R. J., Fortier, M. S., & Guay, F. (1997). Self-determination
and persistence in a real-life setting: Toward a motivational
model of high school dropout. Journal of Personality and Social
Psychology, 72, 1161–1176. doi:10.1037/0022-35184.108.40.2061.
Vlachopoulos, S. P., Karageorghis, C. I., & Terry, P. C. (2000).
Motivation proﬁles in sport: A self-determination theory per-
spective. Research Quarterly for Exercise and Sport, 71,
Wang, C. K. J., & Biddle, S. J. H. (2001). Young people’s
motivational proﬁles in physical activity: A cluster analysis.
Journal of Sport & Exercise Psychology, 23, 1–22.
Wang, C. K. J., Chatzisarantis, N. L. D., Spray, C. M., & Biddle, S. J.
H. (2002). Achievement goal proﬁles in school physical
education: Differences in self-determination, sport ability
beliefs, and physical activity. The British Journal of Educational
Psychology, 72, 433–445. doi:10.1348/000709902320634401.
Williams, J. M., & Krane, V. (2001). Psychological characteristics of
peak performance. In J. M. Williams (Ed.), Applied sport
psychology: Personal growth to peak performance (4th ed., pp.
137–147). Mountain View: Mayﬁeld.
62 Motiv Emot (2009) 33:49–62