Journal of Sport & Exercise Psychology, 2011, 33, 828-846
© 2011 Human Kinetics, Inc.
Eleanor Quested, Jos A. Bosch, Victoria E. Burns, Jennifer Cumming, Nikos Ntoumanis, and Joan
L. Duda are with the School of Sport and Exercise Sciences, University of Birmingham, Edgbaston,
Basic Psychological Need Satisfaction,
Stress-Related Appraisals, and Dancers’
Cortisol and Anxiety Responses
Eleanor Quested, Jos A. Bosch, Victoria E. Burns,
Jennifer Cumming, Nikos Ntoumanis, and Joan L. Duda
University of Birmingham
Self-determination theory (Deci & Ryan, 2000) posits basic psychological need
satisfaction (BPNS) as essential for optimal functioning and health. Grounded in
this framework, the current study examined the role of BPNS in dancers’ cognitive
appraisals and hormonal and emotional responses to performance stress. Dancers
reported their degree of BPNS 1 month before a solo performance. Threat and
challenge appraisals of the solo were recorded 2 hr before the performance. Salivary
cortisol and anxiety were measured 15 min before, and 15, 30, 45, and 60 min
postperformance. Higher BPNS was associated with lower cortisol responses and
anxiety intensity. Challenge appraisals mediated the association between BPNS
and cortisol. Threat appraisals mediated the BPNS–anxiety intensity relationship.
These findings point to the potential importance of performers’ BPNS for optimal
emotional and hormonal homeostasis in performance conditions.
Keywords: basic psychological needs, stress, threat, challenge, cortisol, anxiety
For elite athletes and performing artists, performances represent the culmi-
nation of many hours dedicated to training and practice. The anticipation and
undertaking of athletic and artistic feats evokes an array of psychological and
biological stress reactions. Indeed, psychosocial stress associated with evalu-
ated performance is considered to be a primary catalyst of hormonal (Bosch et
al., 2009; Dickerson & Kemeny, 2004) as well as emotional fluctuations (Jones,
Meijen, McCarthy, & Sheffield, 2009). These responses may have implications
for the quality of performance as well as be relevant to the long-term health of the
performer (Burns, 2006; Gaab, Rohleder, Nater, & Ehlert, 2005; Swain & Jones,
1996). While the situational predictors of hormonal responses to performance
stress have been well established (Dickerson & Kemeny, 2004; Michaud, Mathe-
son, Kelly, & Anisman, 2008), less is known about psychological differences as
determinants of hormonal fluctuations. This void in the literature seems amiss,
Needs, Stress-Related Appraisals, Cortisol 829
given the importance of maintaining biological and emotional homeostasis for
the performing athlete or artist.
The basic needs theory (Ryan & Deci, 2000), a mini-theory within the self-deter-
mination framework (Deci & Ryan, 1985, 2000), has become a popular approach to
understanding antecedents of healthful, effective and optimal functioning in perfor-
mance domains (Gagne & Blanchard, 2007). The theory advocates satisfaction of
three basic psychological needs (i.e., autonomy, competence and relatedness) to be
a primary determinant of cognitive, emotional, and behavioral responses.
Basic Psychological Needs
Basic needs theorists (Deci & Ryan, 2000; Ryan & Deci, 2000) define the need for
autonomy as feeling that one’s actions are self-directed, self-endorsed and derived
from personal choices, as opposed to external controls or pressures. The extent to
which one feels capable of meeting task demands effectively is inferred by the
term competence (DeCharms, 1968). The need for relatedness is satisfied when
one feels that there is a sense of mutual and meaningful care among those in the
context at hand (Baumeister & Leary, 1995).
Numerous studies undertaken in the physical domain support the role of
basic psychological need satisfaction (BPNS) as a central determinant of athletes’
psychological health (Gagne & Blanchard, 2007). For example, recent evidence
demonstrates that BPNS predicts variability in athletes’ and dancers’ experiences
of positive affect (Adie, Duda, & Ntoumanis, 2008b; Quested & Duda, 2009a,
2010) subjective vitality (Adie et al., 2008b), burnout (Hodge, Lonsdale, & Ng,
2008), and negative affective states (Quested & Duda, 2010). Only a few studies
have considered the role of basic needs as predictors of stress-related emotional
responses (Baard, Deci, & Ryan, 2004; Black & Deci, 2000; Deci et al., 2001).
These studies, undertaken in workplace and academic settings, found the basic needs
to relate negatively to anxiety responses. The aforementioned investigations have
typically relied upon self-reported indicators of optimal functioning. In a recent
study, Bartholomew and colleagues (Bartholomew, Ntoumanis, Ryan, Bosch, &
Thøgersen-Ntoumani, 2011) found psychological need thwarting (defined by these
authors as when individuals perceived their needs to be actively undermined by
others) to predict secretory immunoglobulin A (S-IgA), an immunological protein
recognized to be a marker of psychobiological functioning. However, BPNS was
unrelated to S-IgA (Bartholomew et al., 2011). Very little is known about how
and to what extent BPNS covaries with biological stress responses that may be
determinants of long-term health. Typically, previous work on the BPNS–health
relationship has tended to consider perceived typical health status over a series of
weeks as the outcome variable. The role of BPNS in shaping stress-related apprais-
als and subsequent biological responses is yet to be determined.
Psychophysiological Stress Reactions
Cortisol has been studied extensively as a putative biological mediator of the links
between the social context, stress, and physical health (Miller, Chen, & Cole, 2009).
Cortisol release regulates homeostatic control via the modulation of metabolic and
830 Quested et al.
immunological processes. However, when cortisol secretion is excessive and/or
sustained bodily systems may be adversely influenced (Miller, Chen, & Zhou, 2007).
Consequently, repeated exposure to situations perceived as excessively stressful
may have metabolic and immunological effects (Raison & Miller, 2003), as well
as long-term physical (Burns, 2006) and psychological (Raedeke & Smith, 2004)
health implications. Given that the health status of dancers is often recognized to
be compromised (Laws, 2005), there is added relevance to examining the role of
BPNS in determining differential responses to stress.
Athletic and artistic events are naturalistic stressors that have been recognized
to elicit dramatic changes in cortisol secretion (Eubank, Collins, Lovell, Dorling,
& Talbot, 1997; Filaire, Sagnol, Ferrand, Maso, & Lac, 2001; Rohleder, Beulen,
Chen, Wolf, & Kirschbaum, 2007). For example, Rohleder and colleagues revealed
that performing in a ballroom dancing competition explained 28% of the variance in
cortisol secretion between control (inactive) and performance days; dancing alone
(as opposed to in a group) explained a further 3% of the variance. These studies
focused on manipulated situational features of the performance or competition (e.g.,
win/loss experiences, dancing alone or in a group). To date, variability in person-
level determinants of cortisol responses has been neglected. Indeed, it has been
argued that future models of stress and metabolic and immunological processes
should consider such person-based variability in perceptions and appraisals as they
may help explain inconsistencies in biological stress responses (Miller et al., 2007).
Predictors of Stress Responses
Cognitive appraisals can shape emotional, physiological and behavioral responses
to stressful situations (Lazarus, 2000; Lazarus & Folkman, 1984). Threat appraisals
represent the construal that the forthcoming event presents danger to the individual’s
well-being or self-esteem. On the contrary, when one appraises the forthcoming
event with a focus on the opportunities for success, growth, learning and mastery,
these reflect challenge-oriented appraisals (Lazarus & Folkman, 1984). Typically,
though not universally, threat appraisals are associated with undesirable anxiety
responses, whereas perceived challenges tend to augment anxiety symptoms in a
more positive manner (Jones et al., 2009). Threat appraisals are also believed to
exaggerate cortisol reactions (Gaab et al., 2005; Jones et al., 2009). For example,
Gaab and colleagues revealed threat appraisals to predict 29% of the variance in
cortisol responses during an evaluative protocol. A recent meta-analysis (Dickerson
& Kemeny, 2004) indicates human cortisol responses are strongest in situations
that pose an evaluative threat.
Despite the recognition that individuals do not appraise and respond to stressful
circumstances in a uniform manner, predictors of this between-person variability
remain underexplored. Basic psychological need satisfaction has been proposed
to impact upon the appraisal process in psychologically demanding situations, by
shaping “the apparent reality of objective events” (Skinner & Edge, 2002, p. 306).
Recent theorizing implies that one’s degree of BPNS may determine whether one
is vulnerable to potentially damaging emotional and physiological responses in
demanding scenarios (Ntoumanis, Edmunds, & Duda, 2009). It seems reasonable
to expect that the extent to which an individual feels competent and able, and in
control and supported (i.e., the degree of BPNS), should determine how he/she
Needs, Stress-Related Appraisals, Cortisol 831
appraises and responds to the demands presented in potentially stressful circum-
stances. High BPNS could determine whether one approaches a performance feeling
in control and perceiving that they have the resources and capacity to tackle the
demand in question. However, with low BPNS, it might be more likely that the
demand is viewed as more daunting and considered a threat to the social or physical
self. However, no evidence exists to support the hypothesized theoretical sequence
linking BPNS, appraisal processes, and hormonal and emotional responses in
stressful circumstances. In sum, a basic needs theory (Ryan & Deci, 2000) driven
approach to examining appraisals of stressful events and predictors of emotional
and hormonal stress responses is warranted. This is relevant to our understanding
of healthy versus health compromising participation in performance domains, such
as the particular case of vocational dance.
Study Objectives and Hypotheses
Firstly, it was hypothesized that the dancers’ typical degree of BPNS experienced
in their dance school context over several weeks would significantly predict the
dancers’ state perceptions of the solo performance as a threat and challenge (nega-
tively and positively, respectively) on the day of performance. Given that BPNS
has been proposed to shape adaptive responses in stressful circumstances (Skinner
& Edge, 2002), we hypothesized that BPNS and challenge appraisals would nega-
tively, and threat appraisals would positively, predict the dancers’ salivary cortisol
secretion before and after the solo performance. In line with recent theorizing
(Jones et al., 2009; Ntoumanis et al., 2009), threat and challenge appraisals were
expected to mediate the association between BPNS and the dancers’ anxiety and
Participants and Procedure
Sixty-one (20 male, 41 female, Mage = 19.30 years, SD = 1.74) dancers undertaking
full time training at a leading dance conservatory in London, UK, were recruited
for the study. It was made clear that participants were free to withdraw at any stage
without any implications for their ongoing training. Dancers agreeing to take part
provided informed consent. The duration of the study period spanned 4 weeks. In
Week 1, the dancers completed the control day protocol over two consecutive days.
In Week 4, participants completed the performance day protocol, which included
performing a ballet solo in front of an audience of peers and teachers. The study
was approved by a departmental ethics board at a large UK University.
Control Day. A day before the control condition, dancers completed a
questionnaire packet recording BPNS, demographic information (including age,
weight, height, years of dance experience) and their health status and associated
behaviors over the past 7 days. On the control day, dancers collected five envelopes,
each containing a questionnaire and a saliva collection tube. Instructions specified
that each envelope should be opened on, or as close as possible to, the hour between
13:00 and 17:00. Each time, dancers were asked to provide a saliva sample and to
832 Quested et al.
respond to two questions assessing their state anxiety intensity. At this time dancers
were also asked to provide details of health related behaviors in the preceding
24 hr (first envelope only) and the past hour (all envelopes). During the control
day, dancers completed their usual activities, including practical dance classes.
Dancers were asked to return the saliva samples to the researcher immediately or
to refrigerate them until a convenient time for their return.
Performance Day. On the morning of their solo performance, dancers completed
a questionnaire assessing their perceptions of threat and challenge associated with
the upcoming performance. The performance day protocol for assessing health
status and behaviors, saliva collection and anxiety measurements mimicked that of
the control day, except that the sampling time points differed. Specifically, dancers
were asked to provide saliva samples and complete the anxiety measures 15 min
before their performance, and immediately, 15 min, 30 min, and 1 hr after their
solo finished. Performances took place between 14:00 and 16:00 and lasted for
approximately 2 min.
Cortisol. Unstimulated salivary cortisol samples were collected using Salivettes
(Sarstedt, Nümbrecht, Germany). Dancers were asked to leave the cotton swab under
their tongue for a timed period of 2 min. The chronological time that each sample
was provided and the exact length of time the Salivette remained under the tongue
was recorded. Once returned, Salivettes were refrigerated and subsequently frozen
at –80 °C pending analysis. Following defrosting, saliva samples were centrifuged
at 9400 rpm for 10 min. The samples were analyzed using a commercially available
enzyme-linked immunosorbent assay (IBL International, Hamburg, Germany).
Basic Psychological Need Satisfaction. Three scales tapped need satisfaction
and a composite BPNS score was calculated from the scale means, aligned with
the approach adopted in previous research due to the interdependencies between
the three needs (Standage, Duda, & Ntoumanis, 2005). This accounts for the
tendency of the three needs to have shared variance, and precludes the possibility
of determining the independent contribution of each need as a unique predictor of
the outcome variables. The selected measures were specifically chosen because
their scale reliability and factor structure had been supported in previous research
involving dancers and/or athletes (Quested & Duda, 2009b; Reinboth & Duda,
2006). Satisfaction of the need for competence was measured using the five-item
competence subscale of the Intrinsic Motivation Inventory (McAuley, Duncan, &
Tammen, 1989). Dancers responded to items (e.g., “I am satisfied with my dancing”)
on a Likert scale of 1 (strongly disagree) to 7 (strongly agree). Six items (e.g.,
“I feel free to express my ideas and opinions”) targeting the degree to which the
dancer felt he/she had choice and could make decisions in terms of his/her dance
engagement, assessed satisfaction of the need for autonomy (Deci et al., 2001).
This scale employs a Likert scale ranging from not at all true (1) to very true (7).
The dancers’ need for relatedness was tapped via five items (e.g., “valued”) from
the acceptance subscale of the Need for Relatedness Scale (Richer & Vallerand,
1998) on a Likert series of strongly disagree (1) to strongly agree (5). All items
followed the stem “In this dance school I feel.” Internal consistency of the autonomy
Needs, Stress-Related Appraisals, Cortisol 833
(α = .79), competence (α = .86), and relatedness (α = .91) scales were acceptable.
We were interested in the degree of BPNS afforded normally in the school during
the academic year (i.e., when there is not an immediately imminent performance).
Therefore the measure of BPNS was taken 1 month before performance in an
effort to gauge what was “typical.” Dancers were asked to respond to all items in
relation to their experiences and feelings in their school “over the past few weeks.”
Perceptions of Threat and Challenge. A six-item scale, based on a measure used
in the academic domain (McGregor & Elliot, 2002) and employed in recent sport
research (Nien, 2007), was slightly adapted to measure the dancers’ challenge and
threat appraisals with regard to the upcoming performance. Dancers were instructed
to respond to the stem “When I think about this performance.” Dancers responded
to three threat (e.g., “I view this performance as a threat”) and three challenge (e.g.,
“I look forward to being challenged”) appraisal items on a scale of 1 (not at all true
for me) to 7 (very true for me). The factorial validity and reliability of the long,
10-item (Adie, Duda, & Ntoumanis, 2008a) and shortened, 6-item (Nien, 2007)
versions of this scale were supported in research involving athletes similar in age
to the dancers in the current study. In the present work, internal consistency of the
threat (α = .91) and challenge (α = .72) scales were acceptable.
Anxiety. Dancers responded to two items assessing state anxiety each time a
saliva sample was provided during the performance day. Items were adapted from a
short form self-report assessment previously designed to tap athletes’ state anxiety
(Thomas, Hanton, & Jones, 2002) and all anxiety items were assessed on a 7-point
scale ranging from not at all to extremely. Dancers were asked to indicate their
“thoughts right now” regarding their level of cognitive anxiety (“I feel concerned/
worried/ anxious”) and somatic anxiety (“I feel physically nervous (e.g., tense, fast
heart rate, ‘butterflies’)”), as an indicator of anxiety intensity.
Performance day cortisol responses at –15 min, +15 min and +1 hr were time
matched with the three closest time corresponding control day samples. Version
17 of SPSS was the statistical package used in all non-multilevel analyses. Before
the main analysis, multivariate analysis of variance (MANOVA) determined the
potential influence upon cortisol levels of smoking status, caffeine intake, gender
and dance experience. Differences in cortisol response patterns between the perfor-
mance day and control day were also examined using repeated-measures ANOVA.
Least squares regression analyses determined whether the dancers’ BPNS predicted
their threat and challenge appraisals on the day of performance. Cortisol was also
represented as area under the curve relative to ground (AUCG) which gives a measure
of total cortisol release. This can be calculated using cortisol concentrations and
time with the baseline as zero (Fekedulegn et al., 2007). The AUCG representation
was used to determine an overall effect size in the analyses related to cortisol.
Multilevel modeling (MLM) techniques using MLwiN version 2.11 (Ras-
bash, Charlton, Browne, Healy, & Cameron, 2009) were used to examine whether
BPNS, threat, and challenge appraisals (Level 2 variables) predicted the level and
between-person variability of cortisol and anxiety reactions (Level 1 variables)
during the performance day. Individual cortisol levels at each time point were
834 Quested et al.
used in the multilevel analyses. Cortisol data were centered at +15 min, as this was
found to be the peak in the cortisol when the data were represented graphically.
Anxiety data were centered at –15 min, which was when the data peaked. Center-
ing is an approach used in multilevel modeling to facilitate the interpretation of
the intercept (Singer & Willet, 2003). First, an unconditional growth model was
tested, specifying random effects for the intercept (at both levels), as well as for
the linear and quadratic terms at Level 2 only (Singer & Willet, 2003). This model
examined (a) whether there were significant changes in cortisol responses over
time and (b) whether there was between-person variability in the mean levels of
cortisol at the point of centering (i.e., +15 min), as well as in the rates of change. A
two-level MLM was employed. Individual changes in the dependent variable over
time were modeled at Level 1. Variability in response patterns between participants
was represented at Level 2 (Singer & Willet, 2003). When the pattern of change of
the dependent variables was curvilinear, both a linear and a quadratic term for time
were included as predictors in the equations, in addition to BPNS and/or appraisals
(Dickerson, Mycek, & Zaldivar, 2008; Singer & Willet, 2003). The predictors were
initially treated as random at the between-person level (i.e., Level 2), but when the
variance of a predictor at this level was nonsignificant, analysis proceeded with
the predictors fixed at that level. A series of conditional growth models were run
to examine (a) the direct effects of the dancers’ BPNS on cortisol responses; (b)
the direct effects of the dancers’ threat and challenge appraisals on cortisol levels;
and (c) whether threat and challenge appraisals mediated the association between
BPNS and cortisol levels. The same approach was adopted to analyze the cognitive
and somatic anxiety data.
A set of multilevel models were tested to evaluate mediation with respect to
each of the three dependent variables (Krull & MacKinnon, 2001). The change in
the magnitude of the direct X to Y path (c) with the inclusion of the mediators (i.e.,
path c′) provides information regarding the total mediation effect (see Figure 1). The
magnitude and significance of each mediating effect (BaBb) was also determined
(Krull & MacKinnon, 1999).
Figure 1 — Mediation paths tested in MLM analyses (adapted from Krull & MacKin-
non, 2001). The predictor X denotes BPNS. The mediators (M) were threat and challenge.
The dependent variable Y represents cortisol, somatic anxiety intensity, cognitive anxiety
intensity, which were separately analyzed.
Needs, Stress-Related Appraisals, Cortisol 835
Data Preparation and Screening
Data were examined for distribution (skewness and kurtosis), Mahalanobis dis-
tances, multicollinearity and violations of standard univariate and multivariate
analysis assumptions (Tabachnick & Fidell, 2001). No severe deviations from
normality were observed, except in the case of the cortisol data distribution.
Log transformations (log 10 x + 1) were performed to improve the distribution
of the positively skewed cortisol data, and one univariate outlier was removed.
Actual physiological cortisol levels are provided in the Tables and in Figure 2.
Logged data were used in all multilevel analyses to avoid violating statistical
MANOVA analyses revealed that there were no differences in cortisol responses
on account of smoking status (ps > .23), caffeine consumption (ps > .43), or gender
(ps > .22). Therefore these variables were not included in the models tested. There
were significant differences between dancers in their first and second year of dance
training in cortisol responses at all time points in the performance day and the
time-matched –15 min control day data point (ps < .05), as well as in challenge
appraisals (p = .02). Therefore, dance experience was controlled in all analyses
when challenge or cortisol were the dependent variables.
Cortisol Responses: Control Day vs. Performance Day
Figure 2 reveals differences in the patterning of cortisol between the two condi-
tions. Specifically, cortisol responses in the control day maintained a flat profile.
On the contrary, the performance day showed a marked increase, peaking 15 min
postperformance and then gradually decreasing. Repeated-measures ANOVA
revealed a main effect for time, F(2, 31) = 4.01, p = .02, η2 = .11; a main effect
for condition that approached statistical significance, F(2, 31) = 3.27, p = .08, η2
= .09; and a significant condition × time interaction, F(2, 31) = 4.17, p = .02, η2
= .12, indicating that cortisol secretion during the performance day increased at a
greater rate than in the control day.
Descriptive Statistics and Correlations
Table 1 presents the descriptive statistics for the study variables. Overall, dancers
reported modest BPNS and were more likely to appraise their upcoming perfor-
mance as a challenge than as a threat. Mean anxiety scores averaged across time
indicating that the dancers experienced modest cognitive and somatic anxiety symp-
toms. It is noteworthy that standard deviations of all self-reported data were small.
Basic psychological need satisfaction was significantly related to threat, cognitive
anxiety and somatic anxiety in the hypothesized directions. Figure 2 presents the
cortisol patterning of dancers with low and high BPNS. Threat appraisals positively
related to the intensity of the cognitive anxiety symptoms reported.
836 Quested et al.
Figure 2 — Mean cortisol values during the control day (all) and performance day condi-
tions (median spilt by high and low BPNS).
Regression Analyses: Predicting Threat
and Challenge Appraisals (Ba Path)
Basic psychological need satisfaction negatively predicted the dancers’ threat
appraisals (B = –.80, p < .001, R2 = .09) and positively predicted the dancers’
challenge appraisals (B = .29, p < .01, R2 = .04).
Multilevel Regression Analyses: Predicting Cortisol
The dancers’ cortisol response peaked 15 min after the end of their performance
(see Figure 2). As this peak was of interest in our study, time was centered at +15
min when predicting cortisol responses. The results showed significant linear (B =
0.03, p < .001) and quadratic (B = –0.01, p < .001) changes in cortisol responses
over time. The between-person variability associated with the linear and quadratic
slopes was negligible and not significant. This finding was reinforced via a visual
inspection of the data. This suggests there to be no between-person variability in
Table 1 Descriptive Statistics and Bivariate Correlations Among the Study Variables
Note. AUCG = area under the curve relative to ground. CA = cognitive anxiety; SA = somatic anxiety (means are averaged across time points).
a p < .06, *p < .05, **p < .01.
838 Quested et al.
the patterns of cortisol secretion over time. Further, the difference in the deviance
statistic when the linear and quadratic components of time were fixed (iterative
generalized least squares (IGLS) deviance = –243.25) versus random (IGLS devi-
ance = –234.99) was nonsignificant (difference = 8.26, df = 5, p = .14). Therefore,
analysis proceeded with the two terms for time fixed at Level 2.
In the conditional model, and supporting the main study hypotheses, BPNS (B
= –0.07, p < .05) significantly predicted cortisol responses. In a separate model, the
association between challenge and cortisol was significant (B = –0.04, p < .05) but
cortisol was not significantly predicted by threat (B = 0.01, p = .52). Accordingly,
threat was no longer included in the mediation analysis. Table 2 presents the results
of the mediation analysis. When BPNS and challenge were simultaneously included
as predictors, the BPNS coefficient (Bc) was reduced and became nonsignificant
(Bc′), suggesting evidence of a mediating effect. However, the indirect effect (BaBb)
via challenge was not significant.
Table 2 Multilevel Estimates of the Bb, Bc, and Bc′ Coefficients for
the Tests of Mediation
YBc (SE)Bc′ (SE)Bb (SE)
Note. Bc denotes the direct path between the independent variable and the dependent variable, without
(Bc) and with (Bc′) the mediator included in the model. In the case of cortisol, the mediator was chal-
lenge. For somatic and cognitive anxiety, the mediator was threat. Bb represents the path between the
mediator and the dependent variable. The multilevel analyses were undertaken on the logged cortisol
data collected from each dancer on multiple occasions during the performance day.
*p < .05, **p < .01, ***p < .001.
Basic psychological need satisfaction accounted for 11%, and threat and chal-
lenge contributed a further 5% to the variance in cortisol AUCG. A threat × linear
time interaction term significantly predicted cortisol responses (B = 0.004, p <
.05), indicating that the cortisol responses of those with higher threat appraisals
increased at a greater rate than those with lower threat appraisals. No other interac-
tion terms were significant.
Multilevel Regression Analyses: Predicting Anxiety
The dancers’ somatic and cognitive anxiety intensities peaked 15 min before per-
formance, decreased immediately following performance and the curves reached
a plateau 15 min after the performance finished. Therefore, time was centered
at –15 min in the models predicting anxiety responses. Interactions between the
independent variables and linear and quadratic times were examined as predictors
of all anxiety variables. No interaction terms were significant.
Needs, Stress-Related Appraisals, Cortisol 839
Somatic Anxiety Intensity. In an unconditional growth model, both linear (B
= –1.06, p < .001) and quadratic (B = 0.09, p < .001) changes in the intensity of
somatic anxiety were found. The between-person variances associated with the
linear and quadratic terms were nonsignificant (p > .05). In addition, there was
no significant difference (difference = 6.14, df = 5, p = .34) between the deviance
statistic when the linear and quadratic components of time were specified as random
(IGLS deviance = 886.88), versus when they were treated as fixed (IGLS deviance
= 892.58). Similarly, we found no between-person variability in BPNS, challenge
and threat appraisals. Thus, all predictors were treated as fixed at Level 2.
Results indicated BPNS to significantly and negatively predict somatic anxi-
ety intensity (B = –0.59, p = .02). In a separate model, when threat and challenge
were entered as predictors, threat (B = 0.21, p = .025) significantly and positively
predicted somatic anxiety intensity, but the relationship between somatic anxiety
and challenge (B = 0.07, p = .64) was not significant. The multilevel estimates
from the mediation analyses are presented in Table 2. When BPNS and threat
were simultaneously included as predictors of somatic anxiety intensity, the BPNS
coefficient reduced and became nonsignificant, indicative of a mediating effect via
threat. The indirect effect via threat was significant (BaBb = –0.16, p < .05). BPNS
accounted for 11% of the variance in mean somatic anxiety intensity, with threat
and challenge appraisals contributing a further 1%.
Cognitive Anxiety Intensity. Both linear (B = –1.00, p < .001) and quadratic (B =
0.09, p < .001) changes in cognitive anxiety intensity were found. The difference in
the deviance statistic between a model in which the linear and quadratic components
of time were specified as random (IGLS deviance = 878.14), and a model where
both were fixed (IGLS deviance = 908.23) was significant (difference = 30.10, df
= 5, p < .001). This means that the trajectories were not the same for all dancers.
Therefore the linear and quadratic components of the model were modeled as
random. As there was no significant between-person variance in BPNS, challenge
and threat appraisals these predictors were fixed.
BPNS significantly and negatively predicted cognitive anxiety intensity (B =
–0.70, p < .01). When threat and challenge were entered as predictors, threat sig-
nificantly and positively predicted cognitive anxiety intensity (B = 0.38, p < .001),
but there was no significant relationship between cognitive anxiety and challenge
(B = 0.02, p = .88). With the inclusion of threat, the BPNS coefficient reduced and
became nonsignificant. A significant mediating effect via threat was evident (BaBb
= –0.30, p < .01). BPNS predicted 19% of the variance in mean cognitive anxiety
intensity and an additional 9% was contributed by threat and challenge appraisals.
Basic psychological need satisfaction is considered to be a necessity for optimal
functioning and the experience of well-being (Deci & Ryan, 2000; Ryan, 1995). This
study demonstrated the relevance of BPNS and cognitive appraisals of a stressful
event to dancers’ state hormonal and emotional stress responses. Findings were
largely supportive of the study hypotheses.
The results indicate BPNS experienced in dance school to be associated with
the way in which dancers perceive an upcoming salient performance; that is, BPNS
840 Quested et al.
predicted the dancers’ threat and challenge appraisals. Previous authors have pointed
to the potential for BPNS to enable individuals to appraise potentially stressful
situations in a more positive light (Ntoumanis et al., 2009; Skinner & Edge, 2002).
Our findings are largely supportive of this theorizing and consistent with the over-
arching tenets of basic needs theory (Deci & Ryan, 2000).
The cognitive appraisal process is understood to be largely driven by an
appraisal of the demand, and an evaluation of the resources available to tackle the
forthcoming demand (Blascovich, Mendes, Tomaka, Salomon, & Seery, 2003). The
findings indicate that BPNS may foster valuable psychological resources that can
be accessed and applied to promote facilitative pre-performance appraisals. For
example, a dancer who typically perceives herself to possess high dance competence
(i.e., high satisfaction of the basic need for competence) might be more likely to
appraise that she has the physical and technical skills to perform the ballet solo.
If the dancer also generally feels supported, a sense of volition and that she is in
control with regard to her dance training, she would more likely feel in command,
respected and cared for and capable when publicly executing dance skills. With
these psychological resources available, the construal that an opportunity to perform
in front of others is more of a challenge is perhaps unsurprising. On the contrary,
when a dancer feels less competent, connected with others and autonomous in his/
her behaviors, personal resources are more likely to be perceived as inadequate. In
this case, it is understandable that performances are more likely to elicit a sense of
pressure and apprehension, associated with threat appraisals. Our findings point to
the importance of promoting BPNS in dance training if dancers are to feel equipped
with the psychological resources to handle evaluative performance-related demands.
BPNS, Cognitive Appraisals and Cortisol Secretion
Salivary cortisol showed a sharp increase immediately following the solo per-
formance. It is unlikely that this increase is due to the physical strains of the
performance. Evidence suggests that exercise must be intense (> 70% VO2max)
and exceed 40 min in duration to result in significant increases in cortisol secre-
tion (Jacks, Sowash, Anning, McGloughlin, & Andres, 2002). While it has been
recognized that the aerobic demands of ballet can reach 80% VO2max (Schantz
& Åstrand, 1984), the dancers’ solo performances were between 1 and 3 min in
duration. Rohleder and colleagues (2007) established that the dancers’ cortisol
levels before and after a simulated nonpublic and noncompetitive dance perfor-
mance were significantly lower than at equivalent times during an authentic public
and competitive performance. In the former condition, cortisol profiles mimicked
those of an inactive day (Rohleder et al., 2007). Collectively, these points provide
evidence that the dancers’ elevated cortisol profiles over the study time period were
psychologically, rather than physiologically, stimulated.
Previous studies have failed to establish a relationship between person level
variables and cortisol fluctuations induced by independent episodes of acute stress
(Pruessner et al., 1997). The need for further theoretically driven hypotheses regard-
ing the role of cognitive, behavioral and emotional mechanisms in this hormonal
response has been emphasized (Miller et al., 2007). In line with the tenets of self-
determination theory (Deci & Ryan, 2000), the current study indicates that BPNS
may be a central psychological mechanism in this biological response. Character-
istics of the stressor, such as perceived controllability, threats to self-esteem and
Needs, Stress-Related Appraisals, Cortisol 841
threats to the social self, have previously been linked with cortisol output (Dickerson
& Kemeny, 2004). Thus, satisfaction of the needs for autonomy, competence, and
relatedness could be psychological differences between individuals of relevance
to the potential biological impact of stressful performance situations.
The present findings suggest that dancers with low BPNS may be particularly
at risk for prolonged or repeated cortisol elevation and, consequently, physiologi-
cal and psychological deregulation (Michaud et al., 2008). Indeed, cortisol is a
catabolic hormone that has been associated with health risks of specific concern
in the dance population, such as degradation of bone mass (Newman & Halmi,
1989), disturbed eating (McLean, Barr, & Prior, 2001), and impaired tissue repair
(Bosch, Engeland, Cacioppo, & Marucha, 2007). Given that cortisol levels fol-
lowing dance performance can remain above baseline for 6 to 12 hr (Rohleder
et al., 2007), longitudinal research is required to delineate the role of BPNS in
safeguarding against compromised allostasis and health. For example, it would be
valuable to examine whether BPNS predicts the extent to which cortisol responses
to performance stress habituate or adapt over time.
Findings support our hypotheses regarding the role of challenge, but not threat,
appraisals in the BPNS– cortisol relationship. Threat and challenge appraisals are
considered to induce differential physiological mechanisms (Blascovich et al.,
2003). Specifically, hypothalamic-pituitary-adrenal axis activity is understood to
be triggered by perceptions of threat, but unstimulated in challenge states (Jones et
al., 2009). In the current study, challenge appraisals were negatively associated with
cortisol secretion. On the surface, this result may appear contradictory to previous
work. In explicating the present findings, however, it is notable that only two danc-
ers scored below the midpoint range in the scale assessing challenge appraisals.
Therefore we are not in a position to argue that low challenge appraisals stimulated
the release of cortisol. Rather, it could be the case that challenge appraisals have the
propensity to suppress cortisol secretion (Dienstbier, 1989; Eubank et al., 1997).
Findings from the mediation analysis partially support the hypothesis that BPNS
may abate hormonal stress responses via the fostering of challenge appraisals as
there was some evidence of a mediating effect. However, the indirect effect was
not significant. Thus, these suggestions are tentative.
Contrary to our expectations, threat appraisals did not significantly relate to
the dancers’ cortisol secretion. An examination of the mean scores suggests that
on the whole, the dancers did not consider the performance to be threatening (M
= 2.51 SD = 1.69); only five of the dancers scored 5 or above on the 1 (low) to 7
(high) threat scale. Nevertheless, the direction of the beta coefficient supports the
expected association between threat appraisals and cortisol secretion. In addition,
the findings of the interaction analyses also indicate that cortisol was secreted at
a higher rate when dancers perceived the performance as more threatening. The
direct effect of BPNS upon cortisol secretion suggests other mechanisms besides
stress-related appraisals may also mediate the BPNS–cortisol association.
BPNS, Cognitive Appraisals and Anxiety Responses
Our hypotheses were supported with regard to the association between BPNS
and the intensity of the dancers’ cognitive and somatic anxiety. Threat appraisals
significantly predicted the intensity of the dancers’ cognitive and somatic anxiety
responses, and significantly mediated the impact of BPNS upon the intensity of
842 Quested et al. Download full-text
reported state cognitive and somatic anxiety. Collectively, these findings point to
the relevance of BPNS for performers. Even in the midst of physically and psy-
chologically demanding situations, BPNS may enhance resilience to maladaptive
cognitive appraisals, and as a consequence, related emotional responses may be
The nonsignificant paths between challenge appraisals and the intensity of
the dancers’ somatic and cognitive anxiety did not support our hypothesis. It has
been proposed that challenge states can elicit intense negative as well as positive
emotions (Jones et al., 2009). Thus, when viewed as challenging, there might still
be the potential for public dance performances to elicit unease and uncertainty.
Even if the dancer felt competent, cared for and autonomous (i.e., high in BPNS),
an important performance may nonetheless be unsettling and associated with
worry, not necessarily through fear of failure, but perhaps reflecting the value and
importance the dancer places on tackling this challenge. Positive emotions, such
as excitement and exhilaration associated with challenge appraisals (Folkman,
2008), may also instigate undesirable somatic reactions. While some dancers may
have the self-regulatory skills to manipulate their physiological arousal, for others,
somatic responses may become disruptive to the ideal physical performance state.
This complex association may explain the null finding between the dancers’ chal-
lenge appraisals and the level of their somatic and cognitive anxiety symptoms. It
is also possible that challenge appraisals are predominantly associated with positive
emotions, which were not analyzed in the current study.
Conclusions and Future Directions
This study indicated that higher BPNS and high challenge appraisals, coupled with
low threat appraisals, may result in a physiological and emotional state that can
be advantageous for dancers’ healthful functioning and performance. The explo-
ration of the interrelationships between dancers’ BPNS, cognitive appraisals and
cortisol/anxiety responses in an ecologically valid setting was a notable strength
of the current study. This enabled the examination of naturally occurring, rather
than laboratory manipulated, emotional and hormonal responses. With respect to
the latter, significant differences in cortisol secretion were noted between a typical
school day for the vocational dancers (i.e., the control day) and the performance
day. In future research, it could also be informative to examine the psychophysi-
ological correlates of state BPNS immediately before, during, and perhaps right
after performance. State BPNS could predict the dancers’ response variables over
and above the BPNS experienced during the normal teaching term and this pos-
sibility is worthy of research attention.
While challenge appraisals were more salient in the relationship between BPNS
and cortisol reactions, threat was the principal mechanism in the BPNS–anxiety
association. The expected association between threat and cortisol secretion may
not have emerged on account of the dancers’ low perceptions of threat in the cur-
rent study. Nevertheless, the observed association between threat and state anxiety
indicates that the threshold at which threat appraisals stimulate anxiety responses
may be lower than that required to trigger cortisol secretion. The role of BPNS
in the stimulation of other anticipatory physiological stress reactions (e.g., sym-
pathetic adrenomedullary activity) considered to be differentially associated with
performance stress (Jones et al., 2009) warrants research attention.