*3.4. Predictors of Anxiety and Depression*

Correlation analysis showed that anxiety was positively associated with exposure to COVID-19 (r = 0.25, *p* < 0.001) and the perceived coronavirus impact (PCI) on students' well-being (r = 0.16, *p* < 0.001). Hierarchical multiple linear regression was conducted separately for anxiety and depression as an explanatory variable and sex, exposure to COVID-19, the impact of COVID-19 on well-being, and physical activity (PA) as predictor variables. Both regression models were tested in a preliminary analysis to ensure that they met the assumptions of residual normality, linearity, homoscedasticity, and non-collinearity (using VIF < 10). All criteria were met. As presented in Table 6, all variables were found to be significant predictors of anxiety. Sex alone explained about 3% of anxiety variability, F(1, 1510) = 48.41, *p* < 0.001. The negative correlation found between sex and anxiety indicated that females (coded as "0") presented higher anxiety levels than males (coded as "1"). When exposure to COVID-19 was included in the regression in the second step, the model explained 9% of anxiety variability, F(2, 1509) = 73.23, *p* < 0.001. The change in variance explained was 6%, F(1, 1509) = 95.04, *p* < 0.001. The third model of regression included the impact of COVID-19 on students' well-being. The variability explained significantly increased to 13% (F(3, 1508) = 74.50, *p* < 0.001), with 4% of change in variance explained (F(1, 1508) = 70.30, *p* < 0.001). The fourth step of regression analysis included PA, which significantly changed the variance explained to 14% (F(4, 1507) = 61.92, *p* < 0.001). The variability explained changed significantly by about 1% (F(1, 1507) = 21.21, *p* < 0.001). The negative association between PA and anxiety indicates that those university students who spent less than 150 min doing PA every week during the COVID-19 pandemic experienced higher levels of anxiety than their counterparts who engaged in more than 150 min PA per week.



\*\*\* *p* < 0.001.

Depression was found to be positively associated with exposure to COVID-19 (r = 0.23, *p* < 0.001) and PCI (r = 0.23, *p* < 0.001). Hierarchical multiple regression analysis showed that all variables included in the model were significant predictors of depression (Table 7). Similar to the previous analysis, sex explained about 3% of depression variability (F(1, 1510) = 54.01, *p* < 0.001), with higher depression among female students than male. In the second step, gender and exposure to COVID-19 explained for 8% of anxiety variability, F(2, 1509) = 70.46, *p* < 0.001. The change in variance explained equaled 5%, F(1, 1509) = 83.95, *p* < 0.001. When PCI was included in the third regression model, the percentage of variability explained significantly increased to 12% (F(3, 1508) = 71.29, *p* < 0.001), with 4% of change in variance explained (F(1, 1508) = 66.81, *p* < 0.001). The fourth model of regression included PA, which significantly changed the variance explained to 15% (F(4, 1507) = 66.64, *p* < 0.001). The variability explained changed significantly by about 3% (F(1, 1507) = 46.26, *p* < 0.001). In comparison with the hierarchical regression model for anxiety, PA seems more important for depression than for anxiety.


**Table 7.** Hierarchical regression results for depression.

#### **4. Discussion**
