Assessing the Robustness of Two Treatments Three Periods Dual Balanced Crossover Designs to Subject Dropout

Crossover designs are very popular designs which are frequently used in clinical and medical research to compare various treatments. in these designs, one treatment is given to each subject for each different time period and response of each subject is taken at the end of each period. when subjects withdraw from the study before they complete their treatment sequence it may cause rather serious problems such as biased results, decrease in the statistical power of study, even a conversion of the design to a disconnected design in which the basic treatment comparisons cannot be performed. in this study, the robustness of two treatments three periods dual balanced designs which enables the estimation of interested treatment comparisons using differences derived from within-subject measurements are evaluated by analyzing the performance measurements obtained by using a optimality criterion and the probability of generating a disconnected design.