Statistical Analysis
Statistical analysis was performed using IBM SPSS (28.0; IBM Corp,
2017), MATLAB (version R2020a, MathWorks), and JASP (Version 0.16; JASP
Team, 2021).
To assess if the severity of misophonia predicts the prevalence of
mimicry, we employed binomial logistic regression with mimicking
(yes/no) as a response variable and severity of misophonia (range 0 to
15) as the explanatory variable. We also explored whether the tendency
to mimic depends on any particular sound category. To assess this, we
used a multivariable logistic regression with mimicking (yes/no) as a
response variable and the distress experienced on 7 different sound
categories (people eating, repetitive tapping, rustling, nasal sounds,
throat sounds, consonant/vowels, environmental sounds) as explanatory
variables.
Misophonia scores for different sound categories tend to be correlated:
somebody triggered by eating sounds could also be triggered by
nasal/throat sounds and other related sounds as well. A multivariate
analysis using principal component analysis (PCA) can take into
consideration the correlation among variables and aims to find common
latent variable/s underlying the measured variables. We ran PCA with
promax rotation on the data consisting of 8 variables (distress on 7
sound categories and mimicry) from 676 participants. Component loadings
> 0.5 were considered meaningful and significant.