Post-Hoc Comparison Tests for Odds Ratios

The null hypothesis of homogeneity of odds ratio across the strata is testedby several tests. When the null hypothesis is rejected, means at least one ofthe odds ratios signicantly diers from others. Post hoc tests are used afterthe null hypothesis of equality of groups is rejected. Those tests aim to revealthe true dierences between groups. In this paper, we propose post-hoc pro-cedures that control familywise type-I error. These procedures provide thehomogeneous subsets that appear in the same homogeneous subset that arenot signicantly dierent. The suggested procedures are applied to severalCOVID-19 data sets.