2.2.2. Statistical Analyses

The statistical methods used can be separated into two parts. The first part was an assessment of the BSMAS and consisted of the following; (i) frequencies and descriptive statistics of key variables, (ii) test of distribution normality, Keiser–Mayer–Olkin test for sampling adequacy and Bartlett’s test of sphericity, (iii) principal component analysis (PCA) to inspect potential deviation from unidimensionality of the BSMAS, (iv) confirmatory factor analysis (CFA) using robust diagonally weighted least squares (DWLS) estimation procedure [36] due to ordinal data and deviations from normality of distributions, (v) internal consistency and construct reliability, assessed by Cronbach’s alpha, (vi) Spearman correlation coefficient ρ for evaluating criterion validity, (vii) Mann–Whitney test to compare results across genders with respective effect sizes (Cohen’s d), calculated from the values of U, and (viii) Cohen’s h to establish the magnitude of an effect between two proportions was calculated by application of the Excel formula: Cohen’s h = 2 × (ASIN(SQRT(p1)) − ASIN(SQRT(p2))).

The second part was dedicated to the testing of the structural equation model (SEM), constructed to assess the potential protective role of resilience. All latent constructs included in the model were subjected to CFA and reliability assessment. The DWLS estimation procedure was applied for both—measurement and structural models. The only model modification procedure was constraining residual error terms within a construct (latent variable) following the suggestions of the JAMOVI 2.3.12 software [37]. The magnitude of indirect effects was calculated by the multiplication of values of direct path coefficients as suggested by Kline [38].

All the statistical procedures were performed using the JAMOVI software, version 2.3.12 [37] and by Psychometrica, an online tool for the calculation of effect sizes [39].