Missing data are frequently encountered in the statistical analysis of randomized experiments. I propose statistical methods that can be used to analyse randomized experiments with a non-ignorable missing binary outcome where the missing data mechanism may depend on the unobserved values of the...