Appendix B-Multiple logistic regression analysis results

Logistic regressions were conducted to determine whether the baseline characteristics can be considered as essential determinants of PWB and SEC. The assumptions of observations being independent and independent variables being linearly related to the log were checked and met. When all baseline characteristics are considered together, they can significantly predict whether or not a participant belong to the low-level groups of PWB (χ2 = 380.831, p = 0.000, df=36) and SEC (χ2 = 353.717, p = 0.000, df = 36). It is worth noticing that the baseline characteristics that significantly predict low-level PWB and SEC are mostly identical to those predicting medium-level PWB and SEC, respectively. Consequently, we present only the regression results for low-level PWB and SEC in the main text and include the complete results in Appendix B.