4.4. Statistical analysis

Descriptive statistics were computed for both demographic and independent variables (i.e. hospitalisation factors and post-hospitalisation factors). Univariate and multivariate Logistics regression analyses (with all the demographic factors adjusted for and all the independent factors entered, forward method and Conditional selection standard was used) were conducted to test the associations of these variables with probable depressive or probable anxiety. Odds ratio (OR) and 95% CI were reported. Univariate and multivariate linear regression analyses (with all the demographic factors adjusted for and all the independent factors entered, stepwise method was used) were also conducted to test the correlates with PTG. Standardised regression coefficients (β) and 95% confidence interval (CI) were reported. Collinearity diagnosis was performed on the multivariate regression model to clarify the intercorrelations among variables (e.g. perceived discrimination, self-stigma, and perceived affiliate stigma). The variance inflation factor (VIF) was used to detect collinearity between variables (VIF greater than ten means significantly collinearity). The level of statistical significance was 0.05. SPSS for Windows (version 24.0, IBM Corp. Armonk, NY) was used.