CMStatistics 2017: Start Registration
View Submission - CMStatistics
Title: Application of transition models for higher-order crossover designs with binary responses Authors:  Takuma Kurosawa - Tokyo University of Science (Japan) [presenting]
Asanao Shimokawa - Tokyo University of Science (Japan)
Etsuo Miyaoka - Tokyo University of Science (Japan)
Abstract: Higher-order crossover designs are considered, which include more than two sequences or more than two periods or both. The simplest crossover design, AB/BA design without baselines, has some disadvantages that test for an effect of carryover or direct by period interaction lacks power and it alias the effect of carryover, direct by period interaction and group difference each other. We able to use higher-order crossover designs to avoid these undesirable characteristics. These designs, however, tend to be complicated and be larger designs, which needs a lot of sequences and periods that is why needs much more subjects and long terms like orthogonal Latin square designs. To address this problem, we apply transition model to higher-order crossover designs with binary responses and test efficiency of the model through some simulation studies. Moreover, we show the results of applying the model to actual data.