2.3. Statistics

To test the normality of the distributions, the Kolmogorov–Smirnov test was calculated. For numerical variables (age, PAQ-A raw scores), the means and standard deviations were calculated, while frequencies and percentages were reported for the remaining variables.

The t-test for independent samples was used to identify differences between groups in parametric variables (PAQ-A raw scores, age). The χ² test was used to identify differences between groups in ordinal variables, and to test the associations between ordinal variables (i.e., gender and PAQ-A observed at a categorical scale (sufficient vs. insufficient PAL)). Changes in PAQ-A raw scores during the study course were evaluated by a t-test for dependent samples.

Binary logistic regression was used to estimate relationships between studied predictors observed at the first-testing wave and dichotomized PAQ-A criteria (e.g., insufficient vs. sufficient PAL, please see previous text for details), and the odds ratio (OR) and the corresponding 95% confidence intervals (95% CI) were reported. Since preliminary values indicated a strong influence of gender on PAL, the logistic regressions were controlled for gender as a covariate. Finally, gender was observed as an effect modifier, and all variables were additionally checked for correlations in logistic regressions stratified for gender. Hosmer Lemeshow test (HL) was calculated to test the model fit.

Statistical significance was set at 95%, and all analyses were done by software Statistical ver. 13.5 (Tibco Inc., Palo Alto, CA, USA).