On the Statistical Inference from Survival Experiments with Two Types of Failure