Athlete Burnout and the Risk of Dropout Among Young Elite Handball Players

Abstract

The negative feelings that are part of burnout syndrome may prompt athletes to drop out of their sport. The objective of the current study was therefore to examine the influence of athlete burnout profiles on playing status 6 years later. The participants of this study were 458 boys and girls between 14 and 18 years old (M = 15.44; SD =.95) enrolled in elite handball training centers. Cluster analysis on athlete burnout and multinomial logistic regressions on the playing status were conducted. The results suggest that those individuals with a "higher burnout" profile at Time 1 were more likely to have stopped playing handball 6 years later. It therefore seems important to develop strategies to prevent burnout in young athletes enrolled in elite training structures and to promote long-term engagement and well-being in elite sporting activity.

Full text

Running Head: ATHLETE BURNOUT AND
DROPOUT 1
1
2
3
4
5
6
7
8
9
10
Athlete Burnout and the Risk of Dropout among Young Elite Handball Players
11
Date of submission: 10/10/2014
12
Date of submission of the 1st revision: 24/01/2015
13
Date of submission of the 2nd revision: 17/06/2015
14
15

Running Head: ATHLETE BURNOUT AND
DROPOUT 2
Abstract
16
The negative feelings that are part of burnout syndrome may prompt athletes to drop
17
out of their sport. The objective of the current study was therefore to examine the influence of
18
athlete burnout profiles on playing status six years later. The participants of this study were
19
459 boys and girls between 14 and 18 years old (M = 15.44; SD = .95) enrolled in elite
20
handball training centers. Cluster analysis on athlete burnout and multinomial logistic
21
regressions on the playing status were conducted. The results suggest that those individuals
22
with a “higher burnout” profile at Time 1 were more likely to have stopped playing handball
23
six years later. It therefore seems important to develop strategies to prevent burnout in young
24
athletes enrolled in elite training structures and to promote long-term engagement and well-
25
being in elite sporting activity.
26
Key Words: cluster analysis, multinomial logistic regression, sport participation,
27
withdrawal
28
29

Running Head: ATHLETE BURNOUT AND
DROPOUT 3
Athlete Burnout and the Risk of Dropout among Young Elite Handball Players
30
Becoming an elite athlete requires full commitment (i.e., athletes have to be totally
31
invested in their training and competition and they must be present at every training of the
32
entire competitive year) and years of intense training. In many countries, designated training
33
centers are an important part of the talent development system (Röger, Rütten, Zeimainz, &
34
Hill, 2010). The young athletes in these centers are in a context where achievement is of
35
prime importance, and this is reflected by the multiple demands from coaches, parents,
36
teammates and the system itself (e.g., sport- and school-related workloads, pressure from
37
significant others). When the athletes are unable to cope with these demands, negative
38
adjustments are to be expected, such as loss of motivation and burnout, which may lead to
39
decreased performance and sport dropout. Yet although it is assumed that decreased
40
performance and dropout may be consequences of burnout syndrome (Smith, 1986), no
41
studies have specifically tested these associations. The objective of the current study was
42
therefore to examine athletes’ level of performance (through their competitive level) and the
43
risk of dropping out of sport participation, based on their level of burnout six years earlier
44
while they were enrolled in these elite training centers.
45
Athlete burnout has been conceptualized as “a multidimensional construct consisting
46
of three dimensions: (a) emotional/physical exhaustion which is characterized by feelings of
47
emotional and physical fatigue stemming from the psychosocial and physical demands
48
associated with training and competing; (b) reduced sense of accomplishment which is
49
characterized by feelings of inefficacy and a tendency to evaluate oneself negatively in terms
50
of sport performance and accomplishments; and (c) sport devaluation which is defined as a
51
negative, detached attitude toward sport, reflected by lack of concern about sport and
52
performance quality” (Raedeke & Smith, 2009, p. 1). To date, this definition has been the
53
most widely used in sport settings.
54

Running Head: ATHLETE BURNOUT AND
DROPOUT 4
However, despite this definition, no fixed threshold has been established for what
55
constitutes high burnout based on scores on the Athlete Burnout Questionnaire (ABQ;
56
Raedeke & Smith, 2009). Researchers thus have to proceed cautiously when classifying
57
individuals as being burned out or healthy, especially when their intent is to test the influence
58
of burnout level on psychological, social and behavioral factors (Eklund & Cresswell, 2007;
59
Hodge, Lonsdale, & Ng, 2008). It is generally acknowledged that a person-oriented approach
60
is more suitable to examine the functional nature of a multidimensional construct (Bergmann,
61
Magnusson, & El Khoui, 2003; Magnusson, 1998). This approach enables researchers to
62
examine patterns and interactions among the individual features that collectively define the
63
multidimensional construct (Gotwals, 2011). Raedeke (1997) conducted a cluster analysis of
64
a group of swimmers and found that 11% of them could be described as having high burnout,
65
as their mean scores were near the scale midpoint (3: “sometimes”) on the three ABQ
66
dimensions. Another group consisting of 44% of the swimmers had low burnout scores (i.e.,
67
scores under 2: “rarely” on the three subscales). More recently, Isoard-Gautheur, Guillet-
68
Descas and Duda (2013) also used cluster analysis and found that 18% of the athletes in their
69
study had mean scores near the scale midpoint on the three dimensions and may have
70
experienced high burnout, and that 28% of them had low scores (i.e., scores under 2 on the
71
three subscales).
72
Nevertheless, the overall levels of burnout in past studies have been quite low. This
73
has been viewed as the “healthy athlete effect,” according to which investigated athletes are
74
relatively healthy since those who are not healthy have already left the sport (Gustafsson,
75
Kenttä, Hassmén, & Lundqvist, 2007). In addition, the tendency in athlete burnout studies is
76
to refer to “higher/lower” burnout scores instead of “high/low” scores as the scores identified
77
in past studies have generally been “moderate” and not “high” (i.e., scores near 3 on the three
78
subscales of the questionnaire, Isoard-Gautheur, et al., 2013).
79

Running Head: ATHLETE BURNOUT AND
DROPOUT 5
Athlete burnout is a major preoccupation in sport, as is sport dropout (Guzmán &
80
Kingston, 2012), which is considered as a negative motivational consequence. Indeed,
81
research on sport dropout has consistently shown that it is predicted by low levels of self-
82
determination (Balish, Rainham, Blanchar & McLaren, 2014; Sarrazin, Vallerand, Guillet,
83
Pelletier, & Cury, 2002). Moreover, when elite athletes drop out of their sport, they suffer on
84
a personal level, but there is also a negative impact on the talent development system itself
85
because talented athletes have been lost (Gustafsson et al., 2007). This point is particularly
86
important because highly motivated individuals – that is, those with the potential to develop
87
into elite athletes – have been shown to be most at risk of burnout (Gustafsson, Hassmén,
88
Kenttä, & Johansson, 2008). Thus, investigating the link between burnout and dropout is
89
crucial from the perspectives of both the individual and talent development.
90
In his early conceptualization, Smith (1986) defined athlete burnout as the
91
psychological, emotional, and sometimes physical withdrawal from an activity in response to
92
excessive stress or dissatisfaction, highlighting that it may lead to sport dropout. Smith thus
93
essentially suggested that burnout is the result of chronic stress: according to his model, the
94
process of burnout represents the situational, cognitive, physiological and behavioral
95
components of stress (Smith, 1986). Athletes must deal with various demands and restrictions
96
related to their status: among them, training, pressure from significant others, low social
97
support, and low autonomy. In response, negative physiological responses may appear (e.g.,
98
tension, anxiety, fatigue), which in turn can lead to maladaptive behaviors as the athletes
99
attempt to cope with the situational, cognitive and physiological components of stress. The
100
athletes might then experience a drop in performance even though they remain engaged in
101
their sport, with dropping out being the ultimate outcome of the burnout process. Coakley
102
(1992; p. 276) further suggested that burnout among young elite athletes is a social
103
phenomenon, with these young people leaving competitive sport for two main reasons: (a) a
104

Running Head: ATHLETE BURNOUT AND
DROPOUT 6
constrained set of life experiences leading to the development of an unidimensional self-
105
concept, and (b) power relationships in and around sport that seriously restrict young athletes'
106
control over their lives.
107
In line with these conceptualizations of athlete burnout, an integrative model for
108
athlete burnout was recently proposed (Gustafsson, Kenttä, & Hassmen, 2011). In this model,
109
major antecedents, early signs, entrapment, vulnerability factors, key dimensions and
110
maladaptive consequences of athlete burnout were identified. The authors particularly
111
highlighted that burnout is caused by antecedents (e.g., excessive training, school/work
112
demands, stressful social relationships, negative performance, lack of recovery, early
113
success), entrapment (e.g., unidimensional athlete identity, high investment, social
114
constraints, performance-based self-esteem, low alternative attractiveness), and vulnerability
115
factors (e.g., perfectionism, trait anxiety, low social support, low autonomy, lack of coping
116
skills, goal orientation, motivational climate). As in the conceptualizations of Smith (1986)
117
and Coakley (1992), this model also assumed that high levels of athlete burnout lead to
118
maladaptive consequences such as long-term performance impairment and sport dropout.
119
However, these three models have only a theoretical view on the influence of athlete burnout
120
on sport dropout and have not yet been specifically tested.
121
The assumption that burnout is associated with a higher risk of sport dropout has
122
nevertheless received partial support by a few studies in the sport domain. These studies
123
found that this phenomenon was associated with amotivation, indicating that the athletes had
124
neither intrinsic nor extrinsic motives for sport participation (Gould Udry, Tuffey, & Loehr,
125
1996; Lemyre, Treasure, & Roberts, 2006). Moreover, in past studies on the influence of
126
motivational regulation on dropout, amotivation was consistently linked to sport dropout
127
(Sarrazin et al., 2002; Vallerand, 1997). As a result, it was assumed that athlete burnout,
128
through its negative impact on motivation, has an impact on sport dropout, but no study has
129

Running Head: ATHLETE BURNOUT AND
DROPOUT 7
ever tested this assumption. In a qualitative study, interviews with burned-out athletes
130
revealed that initially highly motivated athletes can develop severe burnout and ultimately
131
leave their sport (Gustafsson, et al., 2008). This study was retrospective, however, and
132
focused on a limited sample (i.e., 10 subjects). Furthermore, most of the studies on athlete
133
burnout have used a variable-centered approach to examine the relationships between this
134
multidimensional syndrome and expected antecedents and consequences. A person-oriented
135
approach may nevertheless be more appropriate to examine a multidimensional construct like
136
burnout because the focus is placed on the pattern of individual features of the syndrome
137
instead of being oriented on the variable itself (Gotwals, 2011).
138
In line with the studies in the sport domain, researchers on occupational burnout have
139
suggested that the effects can be chronic as relatively stable burnout levels were measured in
140
several longitudinal studies (Shirom, 2005; Taris, Le Blanc, Schaufeli, , & Schreurs, 2005).
141
Burnout also increases the odds of quitting the job (Alameddine, Saleh, El-Jardali, Dimassi,
142
& Mourad, 2012) or absence for sickness (Eriksson, Engstrom, Starrin, & Janson, 2011).
143
Thus, burnout is associated with leaving the job, which is relevant to professional sport as
144
sport is the “job” of the athletes. Moreover, research in the workplace has also linked burnout
145
with objective job performance. In a meta-analysis of 16 studies, Taris (2006) highlighted the
146
mechanisms by which the three dimensions of job burnout can influence performance. First,
147
exhaustion can lead to lower performance because the individual has no more energetic
148
resources to cope with work demands. Depersonalization (i.e., referring to a negative and
149
excessively detached response to the job) can also impair job performance through a
150
motivational mechanism by which the individual is no longer willing to expend effort. Last,
151
diminished personal accomplishment can also lead to reduced performance due to low levels
152
of perceived self-efficacy, which may lead to passivity and low motivation. Yet although
153
researchers have made theoretical assumptions about the potential negative influence of
154

Running Head: ATHLETE BURNOUT AND
DROPOUT 8
athlete burnout on performance (Gustafsson et al., 2011; Smith, 1986), to our knowledge, no
155
study has examined these links in the sport domain.
156
Given the limitations of previous studies, the objective of the current study was to fill
157
a gap in the sport literature with a partial test of the models of Smith (1986) and Gustafsson et
158
al. (2011). We therefore examined the influence of the burnout symptom profiles of young
159
elite handball players while they were in elite training centers on their playing status (i.e.,
160
reflecting performance and dropout from sport) six years later. In line with the theoretical
161
framework of Smith (1986) and Gustafsson et al. (2011), we hypothesized that a group of
162
players with a “higher burnout” profile at Time 1 would emerge through cluster analysis and
163
be more likely to have lower performance (i.e., playing at a lower level) or to have dropped
164
out of their sport six years later, and that a group of players with a “lower burnout” profile at
165
Time 1 would also emerge and be more likely to have higher performance (i.e., playing at a
166
higher level) and to remain engaged in sport participation six years later.
167
Method
168
Participants and Procedure
169
We present longitudinal data from a six-year study of 459 handball players. All were
170
in elite training programs at the beginning of the study (211 females; 248 males)
i
, with an age
171
range from 14 to 18 years old (M = 15.44; SD = .95). These athletes completed a
172
questionnaire measuring athlete burnout at Time 1 while they were in the elite training
173
programs. Institutional approval was gained before conducting the study. In accordance with
174
the recommendations of the ethics committee, written consent to participate in the study was
175
obtained from the parents of all minors. The coaches were informed by mail and contacted by
176
phone regarding the overall purpose of the study and the logistics of questionnaire
177
administration with their team members. The first author administered the questionnaire,
178

Running Head: ATHLETE BURNOUT AND
DROPOUT 9
providing instructions to the athletes and indicating that she would answer any questions they
179
had while responding to the scales.
180
Recently the French Handball Federation has digitized its game sheets and allowed
181
access to these data via the internet. We therefore were able to search online for the players
182
who had completed the questionnaire at Time 1 and determine whether they were still playing
183
handball and at which competitive level six years later.
184
Measure
185
Athlete Burnout. The athletes’ experience of burnout symptoms was measured at
186
Time 1 by the French validated version of the Athlete Burnout Questionnaire (ABQ, Raedeke
187
& Smith, 2009), the “Questionnaire du Burnout Sportif” (QBS; Isoard-Gautheur, Oger,
188
Guillet & Martin-Krumm, 2010). The three subscales of the questionnaire consisted of: four
189
items measuring reduced sense of accomplishment ( = .75; e.g., “It seems that no matter
190
what I do, I don’t perform as well as I should”), four items measuring physical and emotional
191
exhaustion ( = .83; e.g., “I am exhausted by the mental and physical demands of handball”),
192
and four items measuring sport devaluation ( = .72; e.g., “I feel less concerned about being
193
successful in handball than I used to”). With this measurement tool, the participants
194
responded on a 5-point Likert scale (1 = “almost never”, 5 = “most of the time”).
195
Dropout and Performance. In order to measure dropout and performance, the
196
playing status of the athletes six years after Time 1 was classified into four categories based
197
on the playing levels defined by the French Handball Federation. The first group was labeled
198
“Dropout” (n = 128) and included the players who had stopped playing. The second group
199
was labeled “Regional” and included the players who were playing at the regional level,
200
which represented the lowest level in this study (n = 135). The third group was labeled
201
“National” and included the players who were playing at the national level, which
202
represented the intermediate level in this study (n = 102). The last group was labeled
203

Running Head: ATHLETE BURNOUT AND
DROPOUT 10
“Professional” and included the players who were playing at the professional level, which
204
represented the highest level in this study (n = 94). In the analysis section, we also pooled the
205
last three groups and named this group “Continued participation” in opposition to the group
206
of players who had dropped out of sport.
207
Data Analysis
208
Players with missing data were removed from the analysis using Listwise deletion
209
(Meyers, Gamst, & Guarino, 2013). Variables entered into the cluster analysis were burnout
210
z-scores (i.e., reduced sense of accomplishment, emotional and physical exhaustion, and sport
211
devaluation). Both hierarchical (Ward's method with squared Euclidian distances) and
212
nonhierarchical cluster (K-means clustering) methods were used in the analyses (Hair, Black,
213
Babin, & Anderson, 2010). With the hierarchical method, each observation starts out as its
214
own cluster. Subsequently, new clusters are formed by combining the most similar
215
observations until either all observations are grouped into a single cluster or the researcher
216
determines that a parsimonious solution has been achieved by examining the dendrogram and
217
the agglomeration schedule. The agglomeration schedule was examined to identify a
218
relatively large percentage of change between the agglomeration coefficients associated with
219
successive cluster solutions (Hair et al., 2010). Then a nonhierarchical method was used to
220
confirm the cluster solutions retained from the hierarchical cluster analysis (Hair et al., 2010).
221
ANOVA with the Games-Howell post-hoc comparisons were used to test for differences
222
across the clusters.
223
As in past research on job burnout (Alameddine et al., 2012; Eriksson et al., 2011),
224
two multinomial logistic regressions (adjusted odds ratio (OR) and 95% confidence intervals
225
(CI)) were used to predict continued participation versus the dropout risk, and the players’
226
status. Indeed, in the job burnout literature this data analysis strategy has been used to
227

Running Head: ATHLETE BURNOUT AND
DROPOUT 11
examine whether burnout increases the risk of dropout (Alameddine et al., 2012) or sickness
228
absence (Eriksson et al., 2011).
229
The burnout clusters were treated as independent variables. Cluster analyses were
230
carried out using Statistica® 7.1 and multinomial logistic regressions were carried out using
231
SPSS® 20.0. The level of significance was set at p < .05.
232
Results
233
Descriptive Statistics
234
The descriptive statistics of the three dimensions of athlete burnout at Time 1 revealed
235
a reduced sense of accomplishment with a mean value of 2.56 (SD = 0.52), physical and
236
emotional exhaustion with a mean value of 2.60 (SD = 0.70), and sport devaluation with a
237
mean value of 1.69 (SD = 0.66).
238
Custer Analysis on the Three Dimensions of Athlete Burnout
ii
239
A hierarchical cluster analysis was conducted on the three dimensions of burnout at
240
Time 1. As joining two very different clusters results in a large percentage of change in the
241
coefficients, we looked for large increases in the agglomeration coefficient (Hair, Anderson,
242
Tatham, & Black, 1998). The solution with three clusters was identified as the most adequate
243
(40% of change in the coefficients between the solution with two clusters and the solution
244
with three clusters and only 28% of change between the solution with three clusters and the
245
solution with four clusters). Then, a nonhierarchical technique (i.e., K-mean) was used to
246
adjust the results from the hierarchical procedure. Last, after repeating the same procedure
247
with our sample split into two equal groups, the adoption of three clusters was confirmed
248
(Table 1). Cluster 1 was labeled “lower burnout” as it exhibited significantly lower scores on
249
reduced sense of accomplishment and exhaustion than clusters 2 and 3, and lower scores on
250
sport devaluation than cluster 3. Cluster 2 was labeled “higher exhaustion” as it exhibited
251
significantly higher scores on the exhaustion dimension than clusters 1 and 3. Cluster 3 as
252

Running Head: ATHLETE BURNOUT AND
DROPOUT 12
labeled “higher burnout” as it exhibited significantly higher scores on reduced sense of
253
accomplishment and sport devaluation than clusters 1 and 2, and higher scores on exhaustion
254
than cluster 1. These three clusters differed significantly with regard to reduced sense of
255
accomplishment, and emotional and physical exhaustion. The “lower burnout” and “higher
256
exhaustion” clusters were marked by low levels of sport devaluation and did not differ from
257
each other on this dimension. Sport devaluation was a significant factor differentiating the
258
“lower burnout” and “higher exhaustion” clusters from the “higher burnout” clusters.
259
[Table 1 near here]
260
Multinomial Logistic Regression Analysis of the Clusters on the Playing Status
261
Table 2 shows the results of the first multinomial logistic regression analysis
262
examining continued participation versus the dropout risk. Those who were in the “lower
263
burnout” cluster were 2.21 times more likely to be in the “Continued participation” vs.
264
“Dropout” group (b = .79, Wald χ²(1) = 9.21, and p < .01), compared with those who were in
265
the “higher burnout” cluster. Those who were in the “higher exhaustion” cluster were 2.41
266
times more likely to be in the “Continued participation” vs. “Dropout” group (b = .88, Wald
267
χ²(1) = 11.29, and p < .001), compared with those who were in the “higher burnout” cluster.
268
Table 3 shows the results of the second multinomial logistic regression analysis
269
examining the players’ status. Those who were in the “lower burnout” cluster were 2.13 times
270
more likely to be in the “National” vs. “Dropout” group (b = .75, Wald χ²(1) = 4.87, and p <
271
.05), and 3.60 times more likely to be in the “Professional” vs. “Dropout” group (b = 1.28,
272
Wald χ²(1) = 11.06, and p < .001), compared with those who were in the “higher burnout”
273
cluster.
274
Those who were in the “higher exhaustion” cluster were 1.88 times more likely to be
275
in the “Regional” vs. “Dropout” group (b = .63, Wald χ²(1) = 4.21, and p < .05), 2.29 times
276
more likely to be in the “National” vs. “Dropout” group (b = .83, Wald χ²(1) = 5.92, and p <
277

Running Head: ATHLETE BURNOUT AND
DROPOUT 13
.05), and 3.98 times more likely to be in the “Professional” vs. “Dropout” group (b = 1.38,
278
Wald χ²(1) = 12.94, and p < .001), compared with those who were in the “higher burnout”
279
cluster.
280
[Table 2 & Table 3 near here]
281
Discussion
282
The aim of this study was to investigate the links between athlete burnout at Time 1
283
and the performance level and risk of having dropped out in the following six years. In line
284
with the burnout models of Smith (1986) and Gustafsson et al. (2011), we hypothesized that
285
higher burnout (i.e., high levels of burnout in all three subscales) would be associated with
286
lower performance and higher risk of dropout. This was shown to be the case, as the average
287
risk of having dropped out was 2.21 and 2.41 times higher for players with the “higher
288
burnout” profile than the two other groups (i.e., respectively, “lower burnout” and “higher
289
exhaustion”). This supports the assumption that a high level of burnout leads to a higher risk
290
of dropout (Gustafsson et al., 2011; Smith, 1986). The present results confirm the theoretical
291
assumptions formulated by Smith (1986) and Gustafsson et al. (2011) by showing that the
292
level of athlete burnout has behavioral consequences, especially sport dropout. Moreover, the
293
results confirm the findings of other studies on sport dropout (Balish et al., 2014; Sarrazin et
294
al., 2002) by showing that sport dropout is linked to a negative motivational state (i.e., athlete
295
burnout).
296
Our study also shows that the players with the “lower burnout” profile had an average
297
of 2.86 times more chance to have higher performance (i.e., playing at the national or
298
professional level), supporting the hypothesis that athlete burnout scores are linked to
299
performance. More precisely, this result is in line with the suggestion of Taris (2006) and
300
confirms that lower levels of burnout on the three dimensions lead to higher performance in
301
athletes.
302

Running Head: ATHLETE BURNOUT AND
DROPOUT 14
Interestingly, the results also suggest that individuals with the “lower burnout” and
303
“higher exhaustion” profiles at Time 1 were less likely to have dropped out of their sport six
304
years later. The finding that fewer athletes with the “higher exhaustion” profile dropped out
305
seems contradictory and must be evaluated with caution. The items on the exhaustion
306
subscale include questions about physical fatigue, which can arise from overtraining but is
307
also a natural part of any training program. For example, research has failed to find a link
308
between the training load and the symptoms of burnout (Gustafsson et al., 2007). It has also
309
been convincingly argued that burnout is a multidimensional syndrome (Maslach, Schaufeli,
310
& Leiter, 2001; Lonsdale, Hodge, & Rose, 2009; Gustafsson et al., 2011). Thus, without the
311
other dimensions, exhaustion would be only a stress-related construct and insufficient to
312
characterize burnout syndrome (Lonsdale et al., 2009; Maslach et al., 2001). Our findings
313
also lend support to the idea that an athlete’s relationship with his or her sport (e.g., handball
314
in the present study) is important and thus sport devaluation appears to be an important factor
315
in the relationship between athlete burnout and sport dropout (i.e., “lower burnout” and
316
“higher exhaustion” profiles differed from “higher burnout” on this dimension). Indeed,
317
Cresswell and Eklund (2007) pointed out that sport devaluation might serve to disengage
318
self-worth when a valued activity becomes a source of frustrated accomplishment and chronic
319
exhaustion: it thus might be considered as a central dimension in the athlete burnout process.
320
Strengths and Limitations of the Present Study
321
The originality of this study is the demonstration of a link between the initial level of
322
burnout and the status of players six years later. Furthermore, our study includes the
323
measurement of actual behavioral change (i.e., dropout), which is an important aspect of the
324
syndrome (Smith, 1986), and this has been lacking in previous athlete burnout research
325
(Gustafsson et al., 2011). Last, the study design has already been used in job burnout studies
326
(Alameddine et al., 2012; Eriksson et al., 2011) but has never before been used in athlete
327

Running Head: ATHLETE BURNOUT AND
DROPOUT 15
burnout research. Despite the strengths of this study, however, a few limitations should be
328
noted. One is the data analysis procedure. The number of clusters required in cluster analysis
329
is not easy to determine, and the recommendation is to examine the dendogram or the
330
agglomeration coefficients. However, no standardized approach to conducting and reporting
331
cluster analysis has been established. We therefore chose to determine the number of clusters
332
by examining the agglomeration coefficients. Another limitation is the lack of monitoring
333
over the six years between the two data collection points, which precluded any determination
334
as to exactly why certain athletes dropped out of sport. Although this six-year interval
335
increased the risk of other factors affecting the results, it is important to note that researchers
336
in occupational burnout have suggested that burnout symptoms can be chronic (Shirom,
337
2005) and workplace studies have indeed shown similar symptoms over long time periods,
338
even extending up to eight years (Taris et al., 2005). Therefore, future studies should
339
investigate the influence of the burnout level on dropout by using a longitudinal design with
340
repeated measures (e.g., survival analysis, Lunn, 2010) and by examining the reasons why
341
athletes stop their participation in sport (e.g., because they start another sporting activity).
342
Conclusions and Applied Implications
343
The results of the current study show that young athletes are vulnerable to burnout,
344
which may lead to subsequent negative behaviors, such as impaired performance and dropout
345
from sport. These results open avenues for further research to establish the behavioral
346
consequences of athlete burnout over the years. Moreover, these results highlight burnout as a
347
key variable for predicting sport dropout. Strategies to monitor and prevent burnout should
348
thus be implemented early on, when young athletes are in elite training centers, to prevent
349
maladaptive outcomes like impaired performance and dropout. It might even be beneficial to
350
monitor athletes over the course of their careers for burnout level and other potential
351
predictors of sport dropout, such as motivational regulation. This monitoring should be
352

Running Head: ATHLETE BURNOUT AND
DROPOUT 16
conducted with short questionnaires – on key individual and social aspects of training context
353
– completed during periods when athletes have an increase in training volumes, and when
354
they have important competitions. However, it also seems important to monitor athletes
355
during periods when the training volumes decrease to examine if they also recover compared
356
busiest periods of the season. Such monitoring would help to identify negative trends in
357
athletes’ perceptions (i.e., increased level of burnout and amotivation) so that appropriate
358
interventions can be made to guide the athletes away from the negative behavioral
359
consequences. In addition, the strategies to prevent burnout in elite training centers should
360
include the development of tools (e.g., formations for coaches, prevention program aimed to
361
teach stress management for athletes) to encourage the athletes’ well-being and even
362
fulfillment through sport engagement. This in turn would limit the level of burnout, should it
363
occur, and therefore decrease the risk of performance decline and sport dropout.
364
365

Running Head: ATHLETE BURNOUT AND
DROPOUT 17
References
366
Alameddine, M., Saleh, S., El-Jardali, F., Dimassi, H., & Mourad, Y. (2012). The retention of
367
health human resources in primary healthcare centers in Lebanon: a national survey.
368
BMC Health Serv, 12, 419. doi: 10.1186/1472-6963-12-419
369
Balish, S. M., Rainham, D., Blanchard, C., & McLaren, C. (2014). Correlates of youth sport
370
attrition: A review and future directions. Psychology of Sport and Exercise, 15, 429-
371
439. doi: 10.1016/j.psychsport.2014.04.003
372
Bergman, L. R., Magnusson, D., & El Khouri, B. M. (2003). Studying individual development
373
in an interindividual context: A person-oriented approach. Mahwah, NJ: Lawrence
374
Erlbaum.
375
Coakley, J. (1992). Burnout among adolescent athletes: a personal failure or social problem?
376
Sociology of Sport Journal, 9, 271-285.
377
Cresswell, S. L., & Eklund, R. C. (2007). Athlete burnout: A longitudinal qualitative study.
378
The Sport Psychologist, 21, 1-20.
379
Eklund, R. C., & Cresswell, S. L. (2007). Athlete burnout. In G. Tenenbaum, R. C. Eklund
380
(Eds.), Handbook of sport psychology (3rd Ed.) (pp. 621-641). Hoboken, NJ: John
381
Wiley & Sons, Inc.
382
Eriksson, U.B., Engstrom, L.G., Starrin, B., & Janson, S. (2011). Insecure social relations at
383
work and long-term sickness absence due to burnout and other mental diagnoses. Work
384
38, 319-27.
385
Gould, D., Udry, E., Tuffey, S., & Loehr, J. (1996). Burnout in competitive junior tennis
386
players: I. A quantitative psychological assessment. The Sport Psychologist, 10, 322-
387
340.
388
Gotwals, J. K. (2011). Perfectionism and burnout within intercollegiate sport: A person-
389
oriented approach. The Sport Psychologist, 25, 489-510
390

Running Head: ATHLETE BURNOUT AND
DROPOUT 18
Gustafsson, H., Hassmén, P., Kenttä, G., & Johansson, M. (2008). A qualitative analysis of
391
burnout in elite Swedish athletes. Psychology of Sport and Exercise, 9, 800-816. doi:
392
10.1016/j.psychsport.2007.11.004
393
Gustafsson, H., Kenttä, G., & Hassmen, P. (2011). Athlete burnout: an integrated model and
394
future research directions. International Review of Sport and Exercise Psychology, 4, 3-
395
24. doi: 10.1080/1750984X.2010.541927
396
Gustafsson, H., Kenttä, G., Hassmén, P., & Lundqvist, C. (2007). Prevalence of burnout in
397
competitive adolescent athletes. The Sport Psychologist, 21, 21-37.
398
Hair, J. F., Anderson, R. E., Tatham, R. L., & Black, W. C. (1998). Multivariate data
399
analysis with readings. Englewood Cliffs, NJ: Prentice Hall
400
Hair, J. F., Black, W. C., Babin, B. J., & Anderson, R. E. (2010). Multivariate data analysis
401
(7th ed.). Upper Saddle River, NJ: Pearson Education.
402
Hodge, K., Lonsdale, C., & Ng, J. Y. (2008). Burnout in elite rugby: Relationships with basic
403
psychological needs fulfilment. Journal of Sports Sciences, 26, 835-844.
404
doi:10.1080/02640410701784525
405
Isoard-Gautheur, S., Guillet-Descas, E., & Duda, J. (2013).How to achieve in elite training
406
centers without burning out? An achievement goal perspective. Psychology of Sport
407
and Exercise, 14, 72-83.
408
Isoard-Gautheur, S., Oger, M., Guillet, E. & Martin-Krumm, C(2010). Validation of a French
409
version of the athlete burnout questionnaire. European Journal of Psychological
410
Assessment, 26, 203-211. doi: 10.1027/1015-5759/a000027.
411
Lemyre, P. N., Treasure, D. C., & Roberts, G. C. (2006). Influence of variability in
412
motivation and affect on elite athlete burnout susceptibility. Journal of Sport and
413
Exercise Psychology, 28, 32-48.
414

Running Head: ATHLETE BURNOUT AND
DROPOUT 19
Lonsdale, C., Hodge, K., & Rose, E. A. (2009). Athlete burnout in elite sport: A self-
415
determination perspective. Journal of Sport Science, 27, 785-795. doi:
416
10.1080/02640410902929366
417
Lunn P.D. (2010). The sports and exercise life-course: A survival analysis of recall data from
418
Ireland. Social Science & Medicine, 70, 711-719. doi: 10.1016/j.socscimed.2009.11.006
419
Magnusson, D. (1998). The logic and implications of a person-oriented approach. In R. B.
420
Cairns, L. R. Bergman, & J. Kagan (Eds.), Methods and models for studying the
421
individual: Essays in honor of Marian Radke-Yarrow (pp. 33–64). Thousand Oaks, CA:
422
Sage
423
Maslach, C., Schaufeli, W., & Leiter, M. (2001). Job burnout. Annual Review of Psychology,
424
52, 397-422. doi: 10.1146/annurev.psych.52.1.397
425
Meyers, L.S., Gamst, G., & Guarino, A.J. (2013). Applied multivariate research: Design and
426
interpretation (2nd ed.). Thousand Oaks, CA: Sage.
427
Raedeke, T. D. (1997). Is athlete burnout more than just stress? A sport commitment
428
perspective. Journal of Sport and Exercise Psychology, 19, 396-417.
429
Raedeke, T. D., & Smith, A. L. (2009). The athlete burnout questionnaire manual.
430
Morgantown, West Virginia University: Fitness Information Technology.
431
Röger, U., Rütten, A., Zeimainz, H., & Hill, R. (2010). Quality of talent development
432
systems: results from an international study. European Journal of Sport Society, 7, 7-
433
19.
434
Sarrazin, P., Vallerand, R., Guillet, E., Pelletier, L., & Cury, F. (2002). Motivation and
435
dropout in female handballers: A 21-month prospective study. European Journal of
436
Social Psychology, 32, 395-418.
437
Shirom, A. (2005). Reflections on the study of burnout. Work & Stress, 19, 263-270. doi:
438
10.1080/02678370500376649
439

Running Head: ATHLETE BURNOUT AND
DROPOUT 20
Smith, R. E. (1986). Toward a cognitive-affective model of athletic burnout. Journal of Sport
440
Psychology, 8, 36-50.
441
Taris, T. (2006). Is there a relationship between burnout and objective performance? A
442
critical review of 16 studies. Work & Stress, 20, 316-334. doi:
443
10.1080/02678370601065893.
444
Taris, T., Le Blanc, P., Schaufeli, W., & Schreurs, P. (2005). Are there causal relationships
445
between the dimensions of the Maslach Burnout Inventory? A review and two
446
longitudinal tests. Work & Stress, 19, 238-255. doi:10.1080/02678370500270453
447
Vallerand, R. J. (1997). Toward a hierarchical model of intrinsic and extrinsic motivation. In
448
M. P. Zanna (Ed.), Advances in experimental social psychology, Vol. 29 (pp. 271-360).
449
San Diego, CA: Academic Press. doi:10.1016/S0065-2601(08)60019-2
450
451

Running Head: ATHLETE BURNOUT AND
DROPOUT 21
Table 1.
452
Mean (SD) Values of the Three Dimensions of Burnout for the Three Clusters of Participants.
453
Cluster 1:
“lower
burnout”
Cluster 2:
“higher
exhaustion”
Cluster 3:
“higher
burnout”
p
Reduced sense of
accomplishment
2.29 (.03)
2.43 (.04)a
3.16 (.04) a, b
<.001
Emotional and physical
exhaustion
1.97 (.03)
3.09 (.03) a
2.78 (.07) a, b
<.001
Sport devaluation
1.42 (.03)
1.41 (.03)
2.49 (.06) a, b
<.001
n
168 (36.60%)
174 (37.91%)
117 (25.49%)
<.001
Note. a Different from cluster 1 using the Games-Howell test; b Different from cluster 2 using
454
the Games-Howell test
455

Running Head: ATHLETE BURNOUT AND
DROPOUT 22
Table 2.
456
Multinomial Logistic Regression for Continued Participation vs. Dropout (n = 458)
457
OR
95% CI
p
LL
UL
Logit. “Continued participation” vs. “Dropout” (reference outcome)
Cluster 1: “lower burnout”
2.21
1.32
3.68
.002
Cluster 2: “higher exhaustion”
2.41
1.44
4.02
.001
Cluster 3: “higher burnout”
(Reference)
458
459

Running Head: ATHLETE BURNOUT AND
DROPOUT 23
Table 3.
460
Multinomial Logistic Regression for Players’ Status (n = 458)
461
OR
95% CI
p
LL
UL
Logit 1. “Regional” level vs. “Dropout” (reference outcome)
Cluster 1: “lower burnout”
1.73
0.94
3.171
.07
Cluster 2: “higher exhaustion”
1.88
1.02
3.44
.04
Cluster 3: “higher burnout”
(Reference)
Logit 2. “National” level vs. “Dropout” (reference outcome)
Cluster 1: “lower burnout”
2.13
1.09
4.16
.03
Cluster 2: “higher exhaustion”
2.29
1.18
4.48
.01
Cluster 3: “higher burnout”
(Reference)
Logit 3. “Professional” level vs. “Dropout” (reference outcome)
Cluster 1: “lower burnout”
3.60
1.69
7.66
.001
Cluster 2: “higher exhaustion”
3.98
1.87
8.43
.000
Cluster 3: “higher burnout”
(Reference)
Note. Model χ² (6) = 17.93; p < .01
462
463

Running Head: ATHLETE BURNOUT AND
DROPOUT 24
464
i
The data in the present study strongly suggest a multilevel structure. As a result, before the
central analysis, multilevel analysis was conducted to ensure that being in one of the elite
training centers was not a variable that influenced the athletes’ responses. Our results indicate
that enrollment in an elite training center was not a significant level of analysis (only 1.89% of
the explained variance). Thus, we did not use multilevel analysis in our data analysis.
ii
As the nature of the analytic strategy (i.e., cluster analysis with continuous variables) may
raise some concerns about how the analysis could impact the findings, we have conducted a
logistic regression with a three-way interaction and compared the results with those of the
logistic regression with cluster. However, the three-way interactions coefficients did not
reached significance (i.e., .08 < p < .38). However in order to compare the shape of the results
with those obtained with cluster analysis, we have plot these interactions. Inspection of the
plots helps in part to support the results obtained with cluster analysis. Indeed, when individuals
have higher exhaustion, sport devaluation and reduced sense of accomplishment (i.e., which
correspond to the "higher burnout” profile in the cluster analysis), they are less likely to
continue vs. dropout, to play at the regional level vs. dropout, to play at the national level vs.
dropout, and to play at the professional level vs. dropout, than individuals with lower
exhaustion, sport devaluation and reduced sense of accomplishment (i.e., which correspond to
the "lower burnout” profile in the cluster analysis) and individuals with higher exhaustion, and
lower sport devaluation and reduced sense of accomplishment (i.e., which correspond to the
"higher exhaustion” profile in the cluster analysis).