Data Cleaning
Cognitive Tasks. Data cleaning for the executive functioning
tasks will follow individual task recommendations (e.g., in the
flexibility task, trials with response times below 200ms and trials
following errors will be eliminated; Friedman et al., 2008).
Physiological Data. Heart rate, HRV, and electrodermal activity
will be analyzed using Mindware Technologies HRV Analysis Software
(V3.2.11.). We will select the multi-events option to analyze heart rate
between for each phase (see Figure 1): Baseline (between baseline and T1
questionnaires), reactivity (between MAST onset and T3 questionnaires),
and recovery (between recovery onset and T4 questionnaires).
Heart Rate and HRV. The Mindware Technologies HRV
software allows for the calculation of R-R intervals. The Very Low
Frequency (VLF) Band (Hz) will be set between 0.003 and 0.040, the Low
Frequency (LF) Band (Hz) between 0.04 and 0.12, and the High
Frequency/RSA Band (Hz) between 0.12 and 0.40 (see Task Force
recommendations, 1996). The minimum and maximum heart rates will be set
at 40 and 200bpms, respectively. Each of the three phases will be
divided into 60-second epochs, where any segments that are less than 30
seconds will not be deemed fit for analysis. In addition, any segment
for which the respiration power (in volts2, as
recorded through the respiration belt) falls outside the recommended LF
and HF Bands will not be considered for analysis (Berntson et al.,
1997). Each segment will be screened for peak artifacts (e.g.,
incorrectly identified peaks, missing heartbeats, or unusable peaks due
to movement), which, on a case-by-case basis, will be deleted. Depending
on the within-software recommendations, a mid-beat will be inserted to
correct for these deleted artifacts. Following sufficient agreement
between the two raters, the first author will carry out the remaining
data visualization and modification. We will report the average number
of normal R peaks, as well as the mean number of deleted segments. We
will average heart rate and HRV values over each respective segments for
each phase.
Skin Conductance Levels. Electrodermal activity will be
analyzed using the Mindware Technologies EDA software, version 3.2.9.,
similarly to heart rate. The Event Related Skin Conductance Minimum
Response Latency will be set at one second and the maximum at three
seconds from the event (Levinson & Edelberg, 1985). The threshold level
in µSiemens to detect a skin conductance response will be set at 0.05
(SPRAHCEM, 2012). A constant rolling filter of 500 data points per block
will be applied to smooth the EDA signal and we will perform a visual
inspection of the segments. Artifacts such as subject movement and
respiration will be identified and edited out (e.g., using the spline
function of the software). We will calculate mean skin conductance
levels for the three phases.
Data Analysis
All data analyses will be conducted on RStudio and SPSS. Data and
scripts will be made available to readers upon request.
Screening and Removal of Outliers. We will check for normality
in distribution for all predictor and outcome variables. We will examine
the standardized residuals from the regression analyses using the Q-Q
plots. Any significant outliers will be removed from the analyses and
the models will be reanalyzed without them.
Stress Induction Verification. For all stress indicators
(physiological and subjective), we will use multilevel models with
condition (MAST vs. placebo-MAST) and time (baseline, instructions,
MAST, and recovery levels) as categorical variables, along with their
interaction term to verify the efficacy of the stress induction. The
potential influence of age and gender will be considered in the analyses
(e.g., Shapiro et al., 1996).