Content uploaded by Maria Meier
Author content
All content in this area was uploaded by Maria Meier on Apr 25, 2022
Content may be subject to copyright.
Validation of an online version of the Trier Social Stress Test in adult
men and women
Maria Meier1*, Kristina Haub1, Marie-Luise Schramm1, Marc Hamma1,
Ulrike U. Bentele1, Stephanie J. Dimitroff1, Raphaela Gärtner1, Bernadette
F. Denk1, Annika B. E. Benz1, Eva Unternaehrer2 & Jens C Pruessner1
1 Department of Psychology, Division of Neuropsychology, University of Konstanz,
Constance, Germany
2 Child and Adolescent Research Department, Psychiatric University Hospitals Basel
(UPK), University of Basel, Switzerland
*Correspondence: Maria Meier (maria.meier@uni-konstanz.de)
Please note that this manuscript is a preprint and has not been peer-reviewed.
Highlights
• The Trier Social Stress Test (TSST) is a well-established standardized protocol
to induce acute psychosocial stress in the laboratory
• We administered the TSST online via videocall (TSST-OL) in healthy adult men
and women
• The TSST-OL significantly increased cortisol, alpha amylase levels as well as
subjective arousal, while it decreased subjective pleasure
• The cortisol response to the TSST-OL was significantly stronger in men as
compared with women
Abstract
The Trier Social Stress Test (TSST) is a reliable and efficient protocol to induce
acute psychosocial stress in the laboratory. If circumstances do not allow in-
person assessments, an online version of the TSST (TSST-OL) could create more
flexible research opportunities. To date, studies have confirmed subjective and
autonomic stress responses to TSST-OL protocols. In this preregistered study
(https://osf.io/u57aj), we focused on the effect of the TSST-OL on cortisol and
alpha amylase levels, and pleasure and arousal ratings. As cortisol stress
reactivity is mediated by sex, we further compared men and women. We
hypothesized significant increases in cortisol, alpha amylase and arousal, and a
decrease in pleasure in response to the TSST-OL. Also, we expected stronger
cortisol responses in males as compared with females, as in the laboratory TSST.
N=48 adults (56% female, meanage=23.02, SD=3.19) participated in the study.
Saliva sampling devices were sent to participants’ home before testing sessions,
during which the experimenter, a mixed-sex panel, and the participant joined a
video call. Participants underwent the TSST-OL and overall provided five saliva
samples for cortisol and alpha amylase detection. Pleasure and arousal ratings
and psychometric questionnaires were also completed online.
As hypothesized, the TSST-OL significantly increased cortisol, alpha amylase,
and arousal levels, while it decreased pleasure. Moreover, cortisol responses were
significantly stronger in males as compared to females. 64% of subjects were
classified as responders (cortisol rise>1.5nmol/l).
The TSST-OL successfully induced psychophysiological stress in adults. Our
protocol offers new possibilities to study stress outside of the laboratory.
Keywords: TSST; TSST-OL; cortisol; alpha amylase; sex
1. Introduction
Environmental threats challenge organism’s homeostasis and an appropriate response to
such stressors is essential for survival. On the physiological level, the autonomic
nervous system (ANS) and the hypothalamic pituitary adrenal (HPA) axis orchestrate
the restoration and maintenance of homeostasis by mediating adaptive cardiovascular
and metabolic processes through their hormonal end products adrenaline and cortisol
(Sapolsky, 2000; Ulrich-Lai & Herman, 2009). The regulation of the ANS and HPA
axis in response to stress is an important determinant of health and disease (Chrousos,
2009) and has therefore been studied extensively in the last decades.
To study the regulation of the acute stress response in humans, various standardized
protocols are in use. Protocols that combine elements of uncontrollability and social-
evaluative threat have been shown to elicit a stronger activation of the HPA axis
(Dickerson & Kemeny, 2004) as compared to solely physiological stressors (e.g., the
Cold Pressor Test; Hines & Brown, 1936). One of the most popular and widely-used
protocols that combines these elements is the Trier Social Stress Test (TSST; Allen et
al., 2017; Kirschbaum et al., 1993). A core component of the TSST is a video-taped
mock job interview, during which the participant presents a free speech and performs a
difficult arithmetic task in front of a mixed-sex panel (Kirschbaum et al., 1993). Several
modifications of the standard TSST have been developed in the last decades, e.g., a
protocol that allows stress induction in children and adolescents (Buske-Kirschbaum et
al., 1997), in a group setting (von Dawans et al., 2011), or in virtual reality (Zimmer et
al., 2019). Yet, to date, these variations rely on the participant visiting the laboratory,
which might not always be feasible, as not all circumstances allow in-person
assessments (e.g., demonstrated in times of contact restriction measures due to the
COVID-19 pandemic). Moreover, subpopulations that have difficulties reaching
research sites can only be studied at great expense, or not at all. This in turn limits the
generalization of results and the spectrum of research questions that can be studied
using the TSST. At the same time, current measurement methods (e.g., determination of
hormones via saliva and stability of some metabolites at room temperature for certain
time periods) allows performing measurements outside of the laboratory. Consequently,
an online version of the TSST could create more flexible research opportunities and
offer new possibilities to study psychosocial stress in a standardized manner outside of
the laboratory (Kirschbaum, 2021).
Previous reports have shown that online variations of the TSST that take place in online
video call settings (TSST online, TSST-OL), can elicit significant subjective stress, and
ANS response in adults (Harvie et al., 2021; Huneke et al., 2021; Reed et al., 2021).
Further, an online adaptation of the TSST for children has already been shown to
significantly increase cortisol, alpha amylase, and subjective stress levels (Gunnar et al.,
2021). However, to date, we are not aware of a study that investigated the effects of the
TSST-OL on the endocrine stress system in an adult sample. While a pilot study in our
lab showed promising results for the efficacy of the TSST-OL in activating the HPA
axis and triggering a cortisol stress response (Meier, Benz, et al., 2021), the small
sample size questioned the generalizability of the results. Further, the sample did not
allow conclusions about the possible effects of biological sex on the cortisol stress
response that is well documented in the literature (Kirschbaum et al., 1999; Kudielka &
Kirschbaum, 2005; Liu et al., 2017).
The aim of this study was therefore to validate the efficacy of the TSST-OL in
triggering a cortisol stress response in a sample of adult men and women. We sent study
materials (i.e., saliva sampling devices) to eligible participants via mail and scheduled a
testing session that took place via video call. During the session, participants underwent
the TSST-OL (protocol openly available via: https://osf.io/d3zqk/) and provided five
saliva samples for cortisol and alpha amylase detection as markers of the HPA axis and
the sympathetic branch of the ANS. Further, participants repeatedly rated their current
mood on the dimensions arousal and pleasure. We hypothesized that the TSST-OL
triggers a significant increase in salivary cortisol, alpha amylase, and subjective arousal
while decreasing subjective pleasure (H1). In line with previous reports (Kirschbaum et
al., 1999), we expected that the cortisol stress response was higher in males as
compared with females in the follicular phase (H2). Accordingly, we expected that total
cortisol output during the experiment (as indexed by the area under the curve with
respect to ground, AUCg, Pruessner et al., 2003), and cortisol stress reactivity (as
indexed by the area under the curve with respect to increase, AUCi, Pruessner et al.,
2003) was higher in males as compared with females.
2. Methods
2.1. Preregistration
The hypotheses of this study and the statistical analysis plan were preregistered on Open
Science Framework prior to any human observation of the data (https://osf.io/u57aj;
date of registration: February 9, 2022). This preregistration focused on cortisol as the
main outcome of the study.
2.2. Sample size rational
To estimate our sample size, we conducted a power analysis in G*Power (Faul et al.,
2007) before data collection. The power analysis was based on the interaction
hypothesis (H2), in which we planned to compare the cortisol trajectories (within
subject factor, five timepoints) of two groups (between subject factor, men and women).
We assumed a small (f=.1) to medium (f=.25) effect (mean f=.175) and wanted to
achieve 80% power. In our pilot data (Meier, Benz, et al., 2021), the cortisol values
correlated with r=.63 within subjects on average. Using these estimates, a total sample
of N=32 (16 males and 16 females) was needed. Since the effect size is based on rough
estimates and to account for potential dropouts or exclusions, we planned to test a
minimum of N=20, and a maximum of N=25 participants per group. We stopped
recruiting participants as soon as a minimum of N=20 participants were tested in each
group (males and females).
2.3. Recruitment and exclusion criteria
We recruited participants via the distribution of flyers at the facilities of the University
of Konstanz, the participant database SONA of the University of Konstanz, and
different social media platforms (Instagram, Facebook). Before an invitation to a testing
session, participants filled in an online screening questionnaire on the platform
Qualtrics (duration: approximately 10 minutes). During the screening, participants
reported their assigned sex at birth (male, female, intersex; intersex people were not
invited due to our interest in comparing males to females). We applied the following
exclusion criteria: 1) age < 18 and > 40 years (to ensure all female participants were
pre-menopause), 2) body mass index (BMI) indicating underweight (< 18.5 kg/m2) or
obesity (>30 kg/m2), 3) use of hormonal contraception (including intrauterine device),
lack of (regular) menstrual cycle or current pregnancy in women, 5) smoking more than
five cigarettes per day, 6) working nightshifts, 7) physical or mental illness affecting
HPA axis regulation, 8) medication intake affecting HPA axis regulation (e.g.,
antihistaminic medication) , and 9) depressive symptoms (Beck's Depression Inventory
sum score > 18; Kühner et al., 2007). In addition, participants had to ensure their access
to a stable internet connection, a laptop (or similar device) that allowed for video calls,
and an undisturbed room.
2.4. Experimental procedure
Eligible participants were invited to participate in the study. The study material
(Salivettes to obtain saliva samples for cortisol detection; one piece of Dextro Energy
dextrose; paper) was sent to their home via mail. Testing sessions of women were
scheduled to take place in the early follicular phase of their menstrual cycle (estimation
based on 2-3 last menstrual cycles, invitation at day 1-7 of the next cycle;
Schmalenberger et al., 2021). The testing session took place via the videocall platform
ZoomTM (https://www.zoom.us) at either 3 or 5 p.m. and lasted for approximately 75
minutes. Participants were asked to refrain from smoking, eating, and drinking 2 hours
before the session (except for water and unsweetened tea). They were asked to avoid
exercise on the day of the session, and to stick to their usual sleep routine the night
before testing. The study procedure is depicted in Figure 1.
Participants entered the videocall, were welcomed, completed a technical check (video
and audio quality), and gave written informed consent. Webcams of the experimenter
and participant were activated throughout the experiment. Via the videocall chat, the
experimenter sent a link to an online questionnaire, so that participants could fill in the
questionnaires online. First, a psychophysiological baseline (salivary sample and mood
rating) was assessed. After that, participants consumed one piece of Dextro Energy
(dextrose with blackcurrant flavor; provided with study material) to control blood
glucose levels (Bentele et al., 2021; Meier, Bentele, et al., 2021; von Dawans et al.,
2020; Zänkert et al., 2020). This was followed by a questionnaire period. Then,
participants were exposed to the TSST-OL. After a preparation period, participants
entered a recorded breakout session and performed a free speech task and an arithmetic
task in front of a two member, mixed-sex panel. In the subsequent recovery period,
participants returned to the main videocall session, in which the experimenter was
present, and filled in questionnaires. Overall, five saliva samples for later cortisol
detection, and five concurrent mood ratings (Affect Grid and single item visual analog
scales) were assessed throughout the session. In the end, participants were debriefed.
They received a 15€ Amazon voucher or 1.5h course credits as soon as the Salivettes
were returned using the provided stamped envelopes.
2.5. Tasks and measures
2.5.1. Online Trier Social Stress Test
Acute stress was induced by exposing participants to the Trier Social Stress Test
(TSST), with the main modification being that the task was performed online via
videocall (e.g., on the platform ZoomTM; TSST-OL). Inspired by a TSST study in
children (Gunnar et al., 2021), we translated the standard protocol for adults to an online
setting while maintaining the core components and tasks.
The procedure was introduced by the experimenter just before the start of the TSST-OL.
The participant was told that they will undergo a fictious, videotaped job interview for a
job of choice including a free speech and an arithmetic task which they perform in front
of a two-member, mixed-sex committee. The participant was given 5 minutes to prepare
their free speech in the main session of the videocall. During this preparation, the
participant was allowed to take notes on a white paper provided with the study material.
At the end of the preparation period, they had to fold the paper and put it aside. Then,
the participant was asked to stand up and back away from the camera so that they were
visible from the waist upwards. As soon as audio and video quality from distance were
confirmed, the participant joined a recorded breakout session, in which the mixed-sex
panel awaited them (panel members joined the call separately). The panel members
wore neutral, professional clothes, and sat in front of a neutral background. As in the
standard TSST, they were trained to only show neutral facial expressions. After
ensuring that the screen of the participant was showing the “gallery view” (so that both
panel members were visible to the participant), the panel asked the participant to present
their speech (5 minutes) and perform a mental arithmetic task (5 minutes). At the end of
the task, the recording was stopped, and the participant was redirected to the
experimenter in the main videocall session. A more detailed protocol of the TSST-OL is
available online (https://osf.io/q6tr9/).
2.5.2. Salivary Cortisol and Alpha Amylase
Five saliva samples for detection of free cortisol (nmol/l) and alpha amylase (U/ml)
were collected using Salivettes (Sarstedt, Nümbrecht, Germany; Gröschl et al., 2008).
The biochemical analysis took place in the biochemical laboratory of the Department of
Neuropsychology of the University of Konstanz. Participants were asked to store the
samples in the fridge until they shipped them back to the laboratory, where they were
stored at -20°C until analysis. For cortisol analysis, samples were analyzed in duplicates
using a commercially available competitive enzyme immunosorbent assay (Cortisol
Saliva ELISA, RE-52611, IBL International GmbH, Hamburg, Germany). Thawed
samples were centrifuged at 2500g for 10 minutes. For alpha amylase analysis, samples
were thawed a second time and analyzed in duplicates using a commercially available
liquid phase enzymatic assay (alpha-Amylase Saliva Assay, RE-80111, IBL
International GmbH, Hamburg, Germany). In five samples, the amount of saliva was
too small for analysis, and they were thereby excluded. Furthermore, 24 samples
exceeded the upper detection limit, so they were re-analyzed after dilution. All inter-
and intra-assay coefficients of variation were in the acceptable range.
2.5.3. Subjective arousal and pleasure
Concurrently to taking saliva samples, participants rated their current mood using the
Affect Grid (Russell et al., 1989). The Affect Grid is a single item scale that is rated on
a 9x9 grid. The grid spans the two dimensions displeasure/pleasure, and
sleepiness/arousal. Values on each dimension range from 1 to 9, with higher values
indicating higher arousal, or higher pleasure respectively.
2.5.4. Questionnaires
Participants filled in several questionnaires during the experimental session. The
questionnaire data are not part of the current preregistered hypotheses but are used to
describe the sample. We used the sum score of the Beck’s Depression Inventory II
(Kühner et al., 2007) to index self-reported depressive symptoms. Self-esteem was
assessed using the sum score of the Rosenberg Self-Esteem Scale (Rosenberg, 2002).
Self-reported perceived stress was measured using the sum score of the 10 item
Perceived Stress Scale (Klein et al., 2016). Childhood trauma was measured using the
sum score of the Childhood Trauma Questionnaire (Bernstein et al., 2003). A complete
list of the questionnaires assessed can be found on the OSF website related to this
project (https://osf.io/d3zqk/).
2.6. Participants
Overall, N=48 adults (56.00% females, meanage=23.02, SD=3.19) participated in the
study. Since prior exposure to the TSST can lead to habituation of the stress response,
we excluded participants that reported that they had been exposed to any variation of
the TSST within the last 4 months (Kexel et al., 2021). This applied to one person who
took part in another TSST study 4 days before the testing session. The sample analyzed
in the following thus comprised n=47 adults (55.00% females, meanage=23.11,
SD=3.16).
2.7. Data processing
The cortisol, alpha amylase, and subjective arousal and pleasure data were screened for
missing values. Missing data were imputed using the mean of the respective group
(males, females) at the respective timepoint. We defined outliers in the cortisol data as
values that exceed the mean of the group (male, female) by more than 3 standard
deviations (SD). To decrease the impact of such values on our results, cortisol and alpha
amylase values were winsorized across groups, so that outliers were replaced with
values that were equivalent to 3SD above the respective group mean (applied to 3
cortisol and 3 alpha amylase values). Using the winsorized cortisol data, we calculated
the area under the curve with respect to ground (AUCg) across the complete time course
of the study as an index of total cortisol output during the experiment, and the area
under the curve with respect to increase (AUCi) across the complete time course of the
study as an index of cortisol stress reactivity (Pruessner et al., 2003). Analogously,
AUCg and AUCi of alpha amylase, subjective arousal and pleasure levels were
computed. Since the cortisol data lacked normality (Shapiro-Wilk test: W=0.787,
p<.001), we transformed the values using Box Cox transformation as recommended for
longitudinal endocrine data (Miller & Plessow, 2013).
To compare cortisol responder rates of the TSST-OL to other published studies in the
field, we calculated the percentage change in cortisol values from baseline (-15min) to
expected peak (+30min) concentrations (baseline-to-peak-increase in %) and defined
cortisol non-responders as participants with a baseline-to-peak-increase of < 15.5%, or
1.5nmol/l (Miller et al., 2013), or cortisol stress reactivity < 0 respectively.
2.8. Statistical analysis
Analyses were conducted using R version 4.0.3 (R Core Team, 2019), RStudio version
1.4.1106 (RStudio Team, 2016), and nlme (Pinheiro et al., 2018). Graphs were created
using ggplot2 (Wickham, 2016) and patchwork (Pedersen, 2019). The level of
significance was set to alpha=.05.
For descriptive purposes, we compared demographic and personality characteristics of
males and females using t-tests and Chi squared tests.
To test whether the TSST-OL triggered a significant cortisol stress response, we
modeled cortisol changes over time using a growth curve approach within a multilevel
modeling framework. By doing so we could consider individual differences in cortisol
baseline (random intercepts) and cortisol trajectories over time (random slopes) in our
model (Curran et al., 2010). We modeled a linear, quadratic, and cubic fixed effect of
time. Further, since repeated measures of cortisol are usually correlated (r~.63) we
added a first-order autoregressive covariance structure (AR1). We used a stepwise
approach to build the models and compared the overall model fit of the nested models
using the log-likelihood ratio and evaluated the final model including all random and
fixed effects. In case of model convergence problems, we simplified the complexity of
the random effect structure (e.g., by excluding higher order random slopes).
In addition to the growth model, we used Bonferroni corrected post-hoc t-test to
conduct pairwise comparisons of the five timepoints (e.g., to contrast the baseline at -
15min and the expected post-stress peak at +30min). To allow for comparison between
the magnitude of the cortisol stress response to the TSST-OL and the standard TSST,
we calculated the effect size of cortisol change from baseline to peak (Goodman et al.,
2017).
To test whether the cortisol stress response to the TSST-OL was significantly higher in
males as compared to females, we added the independent variable group (male, female)
to our growth curve and evaluated the main effect of group, and the interaction effect of
group by time in the final model. We followed up with the calculation of Bonferroni
corrected post-hoc t-tests to test whether the cortisol values after stress onset are higher
in males as compared to females. Further, we compared total cortisol output (AUGg)
and cortisol stress reactivity (AUCi) of males and females by using t-tests.
Complementing the preregistration, the same analyses were conducted using alpha
amylase, subjective arousal, and pleasure as outcome variables. Further, we
exploratively compared whether cortisol stress responders were distributed differently
among men and women using Chi squared tests.
3. Results
To maintain the clarity and brevity of this report, we summarize the core results in the
following. The detailed results (including results of nested model comparisons etc.) can
be found in the supplemental material and recalculated using the available analysis
script and data (see https://osf.io/d3zqk/). Descriptive statistics of the groups (males,
females) are summarized in Table 1.
3.1. Cortisol stress response
The inclusion of random intercepts, random slopes, a quadratic, and cubic trend of time
as well as both, a main effect of sex, and the sex by time interactions led to significant
increases in model fit of the growth curve. Coefficients of the final model can be
retrieved from Table 2. We found that cortisol levels changed significantly throughout
the experiment, with the time trend being best described by a cubic effect (see Figure
2A). Bonferroni corrected post-hoc t-test confirmed that the TSST-OL led to a
significant increase in cortisol levels from levels before stress to post stress levels, with
the peak being reached +30min after stressor onset, which is in line with previous
standard TSST studies (Kudielka & Kirschbaum, 2005). The effect size of cortisol
change from baseline to peak was d=1.08, which is comparable to effect sizes reported
for the standard TSST (cf. d’=.925 in Goodman et al., 2017).
The cortisol response to the TSST-OL was significantly higher in males as compared
with females, which was reflected in a significant group by quadratic time interaction
effect (see Table 2), and significantly higher peak levels in males (mean=14.39,
SD=9.61) as compared with females (mean=5.83, SD=4.65) at timepoint +30min. The
results of all Bonferroni corrected post-hoc t-test and the mean cortisol values per group
at each timepoint are reported in the supplemental material.
We found significantly higher total cortisol output (AUGg) in males (mean=615.63,
SD=376.05) as compared with females (mean=266.89, SD=186.91), t(27.91)=-3.88,
p<.001, d=-1.22, and higher cortisol stress reactivity (AUCi) in males (mean=346.06,
SD=317.85) as compared with females (mean=71.42, SD=149.09), t(27.06)=-3.65,
p=.001, d=-1.15.
Counting cortisol responder rate based on the baseline-to-peak-increase in % (Miller et
al., 2013), all participants were rated as responders. In contrast, when counted based on
the 1.5nmol/l criterium (Miller et al., 2013), 64% of participants were responders, with
no significant difference between men and women, Chi2(1)=3.57, p=.059. A
comparable picture emerged when using positive cortisol stress reactivity (AUCi>0) as
responder criterium (79% responders, no significant difference between men and
women, Chi2(1)=1.99, p=.158.
3.2. Alpha amylase stress response
The inclusion of random intercepts, and a quadratic, and cubic trend of time led to
significant increases in model fit of the growth curve. Neither the main effect of group
(males, females) nor the group by time interactions led to significant improvements of
the model. Coefficients of the final model can be retrieved from Table 3. Alpha amylase
levels changed significantly over the course of the experiment (see Figure 2B), with
Bonferroni corrected post-hoc t-test confirming that alpha amylase levels increased
significantly in response to the TSST-OL. Peak levels could be observed +15min after
stressor onset, which is in line with previous standard TSST studies (Nater & Rohleder,
2009). The effect size of alpha amylase change from baseline to peak was d=0.46.
The response did not significantly differ between males and females (no significant
group*time interaction effect in the growth curve approach, no difference in total alpha
amylase output or alpha amylase stress reactivity). The results of all Bonferroni
corrected post-hoc t-test and the mean alpha amylase values per group at each timepoint
are reported in the supplemental material.
3.3. Subjective arousal and pleasure ratings
Modeling changes in subjective arousal over time, only the inclusion of a quadratic time
trend led to significant increases in model fit. Arousal levels increased in response to the
TSST-OL and decreased thereafter, with the highest ratings being observed +15min
after stressor onset (see Figure 2C). The response did not significantly differ between
males and females (no significant group*time interaction effect in the growth curve
approach, no difference in AUCg or AUCi). Results of the model comparisons and
coefficients of the final model can be retrieved from the supplemental information.
Correspondingly, when looking at subjective pleasure ratings, only the inclusion of the
correlation structure led to a significant increase in model fit. Results of the model
comparisons and coefficients of the final model can be retrieved from the supplemental
information. Pleasure ratings decreased in response to the TSST-OL and increased in
the recovery phase, with the lowest ratings being observed directly after stressor
cessation (see Figure 2D). Males and females did not significantly differ in their ratings
(no significant group*time interaction effect in the growth curve approach, no
difference in AUCg or AUCi).
4. Discussion
The results showed that our TSST-OL protocol successfully triggered a cortisol, alpha
amylase, and subjective stress response in adults. The cortisol stress response was
higher in males as compared with females in the follicular phase (Kirschbaum et al.,
1999; Kudielka & Kirschbaum, 2005; Liu et al., 2017), but alpha amylase stress
responsivity (van Stegeren et al., 2008), and arousal and pleasure ratings did not differ
between sexes. Based on effect size measures, the cortisol response elicited by the
TSST-OL (d=1.08) was comparable to the standard TSST in adults (cf. d’=.925 in
Goodman et al., 2017). Based on the 1.5nmol/l criterium (Miller et al., 2013), 64% of
subjects were classified as responders, with no significant difference between sexes.
Compared to the standard TSST, the TSST-OL might thus be slightly less effective
(e.g., responder rate of > 70% reported in Kudielka & Kirschbaum, 2005; yet we must
note that responder rate criteria are inconsistently used and reported, which complicates
comparability between studies). Overall, our results are in line with and expand
previous studies that showed significant subjective and autonomic stress responses to
online TSST protocols (Harvie et al., 2021; Huneke et al., 2021; Reed et al., 2021). By
confirming a significant increase in cortisol, we could moreover ensure an activation of
the main endocrine stress system, the HPA axis.
The sex-dimorphic pattern in cortisol stress responses is consistent with previous studies
investigating stress-induced changes in free, biologically active cortisol. It might be
related to basal gonadal hormone differences, i.e. estradiol and testosterone levels, that
have been linked to HPA axis responsivity in animals and humans (Kirschbaum et al.,
1999; Kudielka & Kirschbaum, 2005). Yet, besides hormonal differences, other factors
could have contributed to the sex dimorphic pattern. For example, women are
particularly sensitive to stress paradigms involving social rejection, as compared with
achievement challenges (Kudielka & Kirschbaum, 2005; Stroud et al., 2002). If judges
in the TSST-OL are not as salient as compared with the standard TSST, because of
limited exposure through a computer screen for example, this could particularly affect
women’s cortisol stress responses. Indeed, a panel out-version of the TSST, in which
the judges are sitting behind a one-way mirror (Andrews et al., 2007; Juster et al., 2012;
Lupien et al., 1997; Marin et al., 2012; Raymond et al., 2019; Wadiwalla et al., 2010),
seems to maximize sex-specific effects by decreasing heterosexual women’s cortisol
response as compared to the standard TSST (Juster et al., 2015). As such, the decreased
responsivity of women as compared with men in our study could either be due to
gonadal hormone differences (Kudielka & Kirschbaum, 2005), or effects of sex and
gender attribution on the perception of the stressor (Juster et al., 2015; Stroud et al.,
2002). While our data do not serve to answer which factor weighs particularly strong,
one strength of our study is that we controlled for the match of sex and gender
identification (all participants reported to be cis) and tested all female participants in the
early follicular phase of their menstrual cycle. Future studies should elaborate on the
question of whether the TSST-OL is equally effective in eliciting a cortisol stress
response in different population subgroups as compared to the standard TSST or
variants of it.
Overall, we can conclude that the TSST-OL successfully triggered an acute
psychophysiological stress response in adults. While shipping the study material (e.g.,
Salivettes) to participants before the testing session involves an extra but manageable
amount of planning and financial effort, the TSST-OL protocol offers new opportunities
to study acute psychosocial stress without the need of bringing participants to the
laboratory.
A limitation of the TSST-OL procedure is that participants need to have access to a
stable internet connection, an undisturbed room, and a laptop with a webcam to conduct
the testing session. Along that line, we cannot assume that all populations are equally
familiar with the use of laptops and video conferencing platforms, or equally willing to
(at least partly) share their living conditions by turning on the webcam during the
testing session. These factors can on the one hand lead to a sampling bias that impacts
the generalizability of results. On the other hand, they can complicate running the
testing session with a direct impact on data quality or usability. If it is unclear whether
participants can join without any obstacles due to unfamiliarity with the remote setup, it
might be useful to conduct an introductory session before the actual experiment, during
which the handling of the video conferencing platform is explained (Gunnar et al.,
2021). Overall, however, we believe that the possibility of conducting acute stress
research remotely can enrich the spectrum of research questions being investigated in
the future at large.
While this validation of the TSST-OL protocol showed that it can induce acute
psychosocial stress, we need to consider some limitations of our study when interpreting
the results. First, the recruited participants were young adults, which limits
generalizability regarding the findings and regarding ease of conducting the remote
protocol, as this age group is probably very familiar with video calls. Second, we did
not ask participants for the size of the laptop screen they used to conduct the study (e.g.,
seen in Gunnar et al., 2021). Related to this, although we asked participants to run the
video conference in gallery mode, we did not check this by, for example, letting
participants show a captured cell phone image (Gunnar et al., 2021). These variables
could impact the visibility and therefore salience of the panel during the stressor, and
consequently, as discussed above, the cortisol stress response of women in particular
(Kudielka & Kirschbaum, 2005; Stroud et al., 2002). Future studies should thus test the
applicability of the TSST-OL in various populations and age groups, and, if possible,
control for variables that could impact the presence of the judges during the stressor.
Taken together, we demonstrated that delivering the TSST in an online environment via
videocall can induce robust cortisol and alpha amylase stress responses in adult men and
women. This remote version offers several opportunities that, in our opinion, remedy
the cost of organising the shipping of study material. For example, while we previously
had problems recruiting eligible male participants in on-site stress studies (Bentele et
al., 2021; Meier, Bentele, et al., 2021), we were able to draw on a larger reach using the
TSST-OL (indeed, almost 50% of the sample were not present locally at the time of the
study), and thus achieve a balanced male-female ratio. Further, as already discussed by
colleagues (Gunnar et al., 2021), we could work independently of room availability at
the faculty, all people involved in testing could join the session from home, and the
panel could even work on parallel sessions as they only joined the experiment for 20
minutes. This overall highly increased flexibility and efficiency of testing sessions.
While possible sampling bias might need to be considered, this is an issue that also
plays a role in on-site studies. As such we conclude that the TSST-OL offers exciting
new and more flexible opportunities to study stress in the digital age, where online
meetings are becoming the norm rather than the exception in remote work
environments, and the TSST-OL may open new avenues to more fully examine stress in
the digital work context.
Funding
This study was conducted with internal research funds of the University of Konstanz
(AFF ICE-TIP) granted to JCP.
Conflict of Interest Declaration
The authors declare to have no conflict of interest.
Preregistration, open material, data, and scripts
Preregistration, materials (the protocol of the TSST-OL), and analysis script, as well as
data of this study and of a pilot study (Meier, Benz, et al., 2021) are available online:
https://osf.io/d3zqk/.
CRediT author contribution
Maria Meier: Conceptualization, Methodology, Formal analysis, Investigation, Data
Curation, Writing - Original Draft, Visualization;
Kristina Haub: Methodology, Investigation, Data Curation, Writing - Review &
Editing;
Marie-Luise Schramm: Methodology, Investigation, Data Curation, Writing - Review
& Editing;
Marc Hamma: Methodology, Investigation, Data Curation, Writing - Review &
Editing;
Ulrike U. Bentele: Writing - Review & Editing;
Stephanie J. Dimitroff: Writing - Review & Editing;
Raphaela Gärtner: Writing - Review & Editing;
Bernadette F. Denk: Writing - Review & Editing;
Annika B. E. Benz: Conceptualization, Writing - Review & Editing;
Eva Unternaehrer: Writing - Review & Editing;
Jens C. Pruessner: Conceptualization, Methodology, Resources, Writing - Review &
Editing, Funding acquisition;
References
Allen, A. P., Kennedy, P. J., Dockray, S., Cryan, J. F., Dinan, T. G., & Clarke, G.
(2017). The Trier Social Stress Test: Principles and practice. Neurobiology of
Stress, 6, 113–126. https://doi.org/10.1016/j.ynstr.2016.11.001
Andrews, J., Wadiwalla, M., Juster, R. P., Lord, C., Lupien, S. J., & Pruessner, J. C.
(2007). Effects of manipulating the amount of social-evaluative threat on the
cortisol stress response in young healthy men. Behavioral Neuroscience, 121(5),
871–876. https://doi.org/10.1037/0735-7044.121.5.871
Bentele, U. U., Meier, M., Benz, A. B. E., Denk, B. F., Dimitroff, S. J., Pruessner, J. C.,
& Unternaehrer, E. (2021). The impact of maternal care and blood glucose
availability on the cortisol stress response in fasted women. Journal of Neural
Transmission. https://doi.org/10.1007/s00702-021-02350-y
Bernstein, D. P., Stein, J. A., Newcomb, M. D., Walker, E., Pogge, D., Ahluvalia, T.,
Stokes, J., Handelsman, L., Medrano, M., Desmond, D., & Zule, W. (2003).
Development and validation of a brief screening version of the Childhood
Trauma Questionnaire. Child Abuse & Neglect, 27(2), 169–190.
https://doi.org/10.1016/S0145-2134(02)00541-0
Buske-Kirschbaum, A., Jobst, S., Wustmans, A., Kirschbaum, C., Rauh, W., &
Hellhammer, D. (1997). Attenuated free cortisol response to psychosocial stress
in children with atopic dermatitis. Psychosomatic Medicine, 59(4), 419–426.
https://doi.org/10.1097/00006842-199707000-00012
Chrousos, G. P. (2009). Stress and disorders of the stress system. Nature Reviews
Endocrinology, 5(7), 374–381. https://doi.org/10.1038/nrendo.2009.106
Curran, P. J., Obeidat, K., & Losardo, D. (2010). Twelve Frequently Asked Questions
About Growth Curve Modeling. Journal of Cognition and Development, 11(2),
121–136. https://doi.org/10.1080/15248371003699969
Dickerson, S. S., & Kemeny, M. E. (2004). Acute Stressors and Cortisol Responses: A
Theoretical Integration and Synthesis of Laboratory Research. Psychological
Bulletin, 130(3), 355–391. https://doi.org/10.1037/0033-2909.130.3.355
Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: A flexible
statistical power analysis program for the social, behavioral, and biomedical
sciences. Behavior Research Methods, 39(2), 175–191.
https://doi.org/10.3758/BF03193146
Goodman, W. K., Janson, J., & Wolf, J. M. (2017). Meta-analytical assessment of the
effects of protocol variations on cortisol responses to the Trier Social Stress
Test. Psychoneuroendocrinology, 80, 26–35.
https://doi.org/10.1016/j.psyneuen.2017.02.030
Gröschl, M., Köhler, H., Topf, H.-G., Rupprecht, T., & Rauh, M. (2008). Evaluation of
saliva collection devices for the analysis of steroids, peptides and therapeutic
drugs. Journal of Pharmaceutical and Biomedical Analysis, 47(3), 478–486.
https://doi.org/10.1016/j.jpba.2008.01.033
Gunnar, M. R., Reid, B. M., Donzella, B., Miller, Z. R., Gardow, S., Tsakonas, N. C.,
Thomas, K. M., DeJoseph, M., & Bendezú, J. J. (2021). Validation of an online
version of the Trier Social Stress Test in a study of adolescents.
Psychoneuroendocrinology, 125, 105111.
https://doi.org/10.1016/j.psyneuen.2020.105111
Harvie, H. M. K., Jain, B., Nelson, B. W., Knight, E. L., Roos, L. E., & Giuliano, R. J.
(2021). Induction of acute stress through an internet-delivered Trier Social
Stress Test as assessed by photoplethysmography on a smartphone. Stress,
24(6), 1023–1032. https://doi.org/10.1080/10253890.2021.1995714
Hines, E. A., & Brown, G. E. (1936). The cold pressor test for measuring the reactibility
of the blood pressure: Data concerning 571 normal and hypertensive subjects.
American Heart Journal, 11(1), 1–9. https://doi.org/10.1016/S0002-
8703(36)90370-8
Huneke, N. T. M., Rowlatt, H., Hyde, J., McEwan, A., Maryan, L., Baldwin, D., &
Garner, M. (2021). A Modified Trier Social Stress Test to Investigate Social
Anxiety using Videoconferencing Software: A Proof-of-Concept study [Preprint].
PsyArXiv. https://doi.org/10.31234/osf.io/q5rbj
Juster, R.-P., Hatzenbuehler, M. L., Mendrek, A., Pfaus, J. G., Smith, N. G., Johnson, P.
J., Lefebvre-Louis, J.-P., Raymond, C., Marin, M.-F., Sindi, S., Lupien, S. J., &
Pruessner, J. C. (2015). Sexual Orientation Modulates Endocrine Stress
Reactivity. Biological Psychiatry, 77(7), 668–676.
https://doi.org/10.1016/j.biopsych.2014.08.013
Juster, R.-P., Perna, A., Marin, M.-F., Sindi, S., & Lupien, S. J. (2012). Timing is
everything: Anticipatory stress dynamics among cortisol and blood pressure
reactivity and recovery in healthy adults. Stress, 15(6), 569–577.
https://doi.org/10.3109/10253890.2012.661494
Kexel, A.-K., Kluwe-Schiavon, B., Visentini, M., Soravia, L. M., Kirschbaum, C., &
Quednow, B. B. (2021). Stability and test-retest reliability of different hormonal
stress markers upon exposure to psychosocial stress at a 4-month interval.
Psychoneuroendocrinology, 132, 105342.
https://doi.org/10.1016/j.psyneuen.2021.105342
Kirschbaum, C. (2021). Why we need an online version of the Trier Social Stress Test.
Psychoneuroendocrinology, 125, 105129.
https://doi.org/10.1016/j.psyneuen.2021.105129
Kirschbaum, C., Kudielka, B. M., Gaab, J., Schommer, N. C., & Hellhammer, D. H.
(1999). Impact of Gender, Menstrual Cycle Phase, and Oral Contraceptives on
the Activity of the Hypothalamus-Pituitary-Adrenal Axis: Psychosomatic
Medicine, 61(2), 154–162. https://doi.org/10.1097/00006842-199903000-00006
Kirschbaum, C., Pirke, K.-M., & Hellhammer, D. H. (1993). The ’Trier Social Stress
Test’—A Tool for Investigating Psychobiological Stress Responses in a
Laboratory Setting. Neuropsychobiology, 28, 76–81.
Klein, E. M., Brähler, E., Dreier, M., Reinecke, L., Müller, K. W., Schmutzer, G.,
Wölfling, K., & Beutel, M. E. (2016). The German version of the Perceived
Stress Scale – psychometric characteristics in a representative German
community sample. BMC Psychiatry, 16(1), 159.
https://doi.org/10.1186/s12888-016-0875-9
Kudielka, B. M., & Kirschbaum, C. (2005). Sex differences in HPA axis responses to
stress: A review. Biological Psychology, 69(1), 113–132.
https://doi.org/10.1016/j.biopsycho.2004.11.009
Kühner, C., Bürger, C., Keller, F., & Hautzinger, M. (2007). Reliabilität und Validität
des revidierten Beck-Depressionsinventars (BDI-II): Befunde aus
deutschsprachigen Stichproben. Nervenarzt, 78, 651–656.
Liu, J. J. W., Ein, N., Peck, K., Huang, V., Pruessner, J. C., & Vickers, K. (2017). Sex
differences in salivary cortisol reactivity to the Trier Social Stress Test (TSST):
A meta-analysis. Psychoneuroendocrinology, 82, 26–37.
https://doi.org/10.1016/j.psyneuen.2017.04.007
Lupien, S. J., Gaudreau, S., Tchiteya, B. M., Maheu, F., Sharma, S., Nair, N. P. V.,
Hauger, R. L., McEwen, B. S., & Meaney, M. J. (1997). Stress-Induced
Declarative Memory Impairment in Healthy Elderly Subjects: Relationship to
Cortisol Reactivity 1. The Journal of Clinical Endocrinology & Metabolism,
82(7), 2070–2075. https://doi.org/10.1210/jcem.82.7.4075
Marin, M.-F., Morin-Major, J.-K., Schramek, T. E., Beaupré, A., Perna, A., Juster, R.-
P., & Lupien, S. J. (2012). There Is No News Like Bad News: Women Are More
Remembering and Stress Reactive after Reading Real Negative News than Men.
PLoS ONE, 7(10), e47189. https://doi.org/10.1371/journal.pone.0047189
Meier, M., Bentele, U. U., Benz, A. B. E., Denk, B., Dimitroff, S., Pruessner, J. C., &
Unternaehrer, E. (2021). Effects of psychological, sensory, and metabolic
energy prime manipulation on the acute endocrine stress response in fasted
women. Psychoneuroendocrinology, 134, 105452.
https://doi.org/10.1016/j.psyneuen.2021.105452
Meier, M., Benz, A., Bentele, U. U., Dimitroff, S. J., Denk, B., Unternaehrer, E., &
Pruessner, J. C. (2021). Validation of an online version of the Trier Social Stress
Test– preliminary results of a pilot study in adults. Psychoneuroendocrinology,
131, 105498. https://doi.org/10.1016/j.psyneuen.2021.105498
Miller, R., & Plessow, F. (2013). Transformation techniques for cross-sectional and
longitudinal endocrine data: Application to salivary cortisol concentrations.
Psychoneuroendocrinology, 38(6), 941–946.
https://doi.org/10.1016/j.psyneuen.2012.09.013
Miller, R., Plessow, F., Kirschbaum, C., & Stalder, T. (2013). Classification Criteria for
Distinguishing Cortisol Responders From Nonresponders to Psychosocial
Stress: Evaluation of Salivary Cortisol Pulse Detection in Panel Designs.
Psychosomatic Medicine, 75(9), 832–840.
https://doi.org/10.1097/PSY.0000000000000002
Nater, U. M., & Rohleder, N. (2009). Salivary alpha-amylase as a non-invasive
biomarker for the sympathetic nervous system: Current state of research.
Psychoneuroendocrinology, 34(4), 486–496.
https://doi.org/10.1016/j.psyneuen.2009.01.014
Pedersen, T. L. (2019). patchwork: The Composer of Plots (1.0.0.) [R].
Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., & R Core Team. (2018). nlme: Linear
and Nonlinear Mixed Effects Models. (3.1-137) [R].
Pruessner, J. C., Kirschbaum, C., Meinlschmid, G., & Hellhammer, D. H. (2003). Two
formulas for computation of the area under the curve represent measures of total
hormone concentration versus time-dependent change.
Psychoneuroendocrinology, 28(7), 916–931. https://doi.org/10.1016/S0306-
4530(02)00108-7
R Core Team. (2019). R: A language and environment for statistical computing. R
Foundation for Statistical Computing [R].
Raymond, C., Marin, M.-F., Juster, R.-P., & Lupien, S. J. (2019). Should we suppress
or reappraise our stress?: The moderating role of reappraisal on cortisol
reactivity and recovery in healthy adults. Anxiety, Stress, & Coping, 32(3), 286–
297. https://doi.org/10.1080/10615806.2019.1596676
Reed, S. C., Oliva, A., Gomez, S. G., & Evans, S. M. (2021). A novel remote TSST
procedure reliably increases stress reactivity in cannabis users: A pilot study.
Experimental and Clinical Psychopharmacology.
https://doi.org/10.1037/pha0000528
Rosenberg, M. (2002). Self-Esteem Scale. Zusammenstellung sozialwissenschaftlicher
Items und Skalen (ZIS). https://doi.org/10.6102/ZIS46
RStudio Team. (2016). RStudio: Integrated Development for R. [R].
Russell, J. A., Weiss, A., & Mendelsohn, G. A. (1989). Affect Grid: A single-item scale
of pleasure and arousal. Journal of Personality and Social Psychology, 57(3),
493–502. https://doi.org/10.1037/0022-3514.57.3.493
Sapolsky, R. M. (2000). Stress Hormones: Good and Bad. Neurobiology of Disease,
7(5), 540–542. https://doi.org/10.1006/nbdi.2000.0350
Schmalenberger, K. M., Tauseef, H. A., Barone, J. C., Owens, S. A., Lieberman, L.,
Jarczok, M. N., Girdler, S. S., Kiesner, J., Ditzen, B., & Eisenlohr-Moul, T. A.
(2021). How to study the menstrual cycle: Practical tools and recommendations.
Psychoneuroendocrinology, 123, 104895.
https://doi.org/10.1016/j.psyneuen.2020.104895
Stroud, L. R., Salovey, P., & Epel, E. S. (2002). Sex differences in stress responses:
Social rejection versus achievement stress. Biological Psychiatry, 52(4), 318–
327. https://doi.org/10.1016/S0006-3223(02)01333-1
Ulrich-Lai, Y. M., & Herman, J. P. (2009). Neural regulation of endocrine and
autonomic stress responses. Nature Reviews Neuroscience, 10(6), 397–409.
https://doi.org/10.1038/nrn2647
van Stegeren, A. H., Wolf, O. T., & Kindt, M. (2008). Salivary alpha amylase and
cortisol responses to different stress tasks: Impact of sex. International Journal
of Psychophysiology, 69(1), 33–40.
https://doi.org/10.1016/j.ijpsycho.2008.02.008
von Dawans, B., Kirschbaum, C., & Heinrichs, M. (2011). The Trier Social Stress Test
for Groups (TSST-G): A new research tool for controlled simultaneous social
stress exposure in a group format. Psychoneuroendocrinology, 36(4), 514–522.
https://doi.org/10.1016/j.psyneuen.2010.08.004
von Dawans, B., Zimmer, P., & Domes, G. (2020). Effects of glucose intake on stress
reactivity in young, healthy men. Psychoneuroendocrinology, 105062.
https://doi.org/10.1016/j.psyneuen.2020.105062
Wadiwalla, M., Andrews, J., Lai, B., Buss, C., Lupien, S. J., & Pruessner, J. C. (2010).
Effects of manipulating the amount of social-evaluative threat on the cortisol
stress response in young healthy women. Stress, 13(3), 214–220.
https://doi.org/10.3109/10253890903277561
Wickham, H. (2016). ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag
New York.
Zänkert, S., Kudielka, B. M., & Wüst, S. (2020). Effect of sugar administration on
cortisol responses to acute psychosocial stress. Psychoneuroendocrinology, 115,
104607. https://doi.org/10.1016/j.psyneuen.2020.104607
Zimmer, P., Buttlar, B., Halbeisen, G., Walther, E., & Domes, G. (2019). Virtually
stressed? A refined virtual reality adaptation of the Trier Social Stress Test
(TSST) induces robust endocrine responses. Psychoneuroendocrinology, 101,
186–192. https://doi.org/10.1016/j.psyneuen.2018.11.010
Figure 1. Study procedure. Sessions took place via videocall. Participants received a
dextrose load after baseline (0min). TSST = Trier Social Stress Test.
Figure 2. Changes in (A) cortisol, (B) alpha amylase, (C) subjective arousal, and (D)
subjective pleasure in males (blue circles) and females (yellow squares) over the course
of the experiment. Shaded area=TSST-OL speech and mental arithmetic task.
Eligibility Screening
Introduction
Questionaires incl.
Food Craving Task (pre)
TSST online -Preparation
TSST online - Spech
TSST online -Maths
Questionaires incl.
Food Craving Task (post)
Questionaires
015 30 45 55
Salivary Cortisol
Time in minutes
TSST judges
present
Questionnaires
Questionnaires
Questionnaires
Salivary Cortisol
and mood
+45
+30+150-15
Table 1. Descriptive characteristics of the sample.
Females
(n=26)
Males
(n=21)
p-value
age
21.65±2.06
24.90±3.40
p<.001
BMIa
21.86±2.09
24.11±3.51
p=.015
depressivenessb
5.08±3.38
4.76±4.53
p=.793
childhood traumac
1.44±1.16
1.33±1.06
p=.747
chronic stress leveld
31.32±3.35
31.71±4.27
p=.733
self-esteeme
31.77±4.97
34.29±4.22
p=.070
cortisol baseline (nmol/l)f
3.26±2.28
4.49±2.73
p=.105
alpha amylase baseline
(U/ml)f
176.4±111.96
201.86±120.17
p=.471
Note. If not otherwise specified, an independent t-test comparing groups was calculated to test whether
groups differed in respect to the listed variables. In these cases, data is expressed as mean±standard
deviation.
aBMI=body mass index, bindexed by Beck’s Depression Inventory II sum score, cindexed by Childhood
Trauma Questionnaire sum score, dindexed by Perceived Stress Scale sum score, eindexed by Rosebnerg Self-
Esteem Scale sum score, fbased on raw values.
Table 2. Coefficients of the growth curve model (number of observations: 235, number of participants: 47)
predicting cortisol changes over time by group (females, males).
Unconditional
Conditional
Fixed effects
Estimate
SE
Estimate
SE
Baseline level, β0
1.99***
0.17
1.46***
0.20
Time linear, β1.1
4.32***
0.95
2.55*
1.22
Time quadratic, β1.2
-3.54***
0.71
-1.95*
0.89
Time cubic, β1.3
-2.23***
0.60
-2.48**
0.82
Group, β2
-
-
1.20***
0.30
Time linear by Group, β3.1
-
-
3.97*
1.83
Time quadratic by Group, β3.2
-
-
-3.56**
1.33
Time cubic by Group, β3.3
-
-
0.56
1.22
Random effects
SD
covariance
baseline-slope
SD
covariance
baseline-slope
Variance baseline level, b0i
1.14
-
.98
-
Variance slope linear, b1.1i
5.92
.34
5.58
.21
Variance slope quadratic, b1.2i
4.07
-.90
3.69
-.87
Variance slope cubic, b1.3i
3.23
.18
3.23
.15
Residual, εti
0.37
-
0.37
-
Note. The unconditional growth model does not include the main effect of group and group*time interaction
terms, while the conditional model does. Please note that the model was calculated using Box Cox transformed
values, which is why coefficients do not represent nmol/l. SE=standard error. SD=standard deviation * p<.05,
** p<.01, *** p<.001.
Table 3. Coefficients of the growth curve model (number of observations: 235, number of participants: 47)
predicting alpha amylase changes over time by group (females, males).
Unconditional
Conditional
Fixed effects
Estimate
SE
Estimate
SE
Baseline level, β0
192.48***
14.77
179.56***
19.82
Time linear, β1.1
-1.15
78.61
-23.28
106.54
Time quadratic, β1.2
-218.48**
69.49
-209.94*
94.16
Time cubic, β1.3
-137.00*
62.61
-116.65
84.85
Group, β2
-
-
28.93
29.65
Time linear by Group, β3.1
-
-
49.76
159.38
Time quadratic by Group, β3.2
-
-
-19.11
140.87
Time cubic by Group, β3.3
-
-
-45.55
126.94
Random effects
SD
covariance
baseline-slope
SD
covariance
baseline-slope
Variance baseline level, b0i
96.44
-
95.36
-
Variance slope linear, b1.1i
323.11
.26
322.85
.25
Variance slope quadratic, b1.2i
204.92
-.73
204.56
-.73
Residual, εti
62.07
-
62.03
-
Note. The unconditional growth model does not include the main effect of group and group*time interaction
terms, while the conditional model does. SE=standard error. SD=standard deviation * p<.05, ** p<.01, ***
p<.001.
