Sequential two-stage d-optimality sensitivity test for binary response data

In order to efficiently extract information about an underlying population based on binary response data (e.g., dead or alive, explode or unexplode), we propose a two-stage D-optimality sensitivity test, which consists of two parts. The first part is a two-stage uniform design used to generate an overlap quickly; the second part conducts the locally D-optimal augmentations to determine optimal follow-up design points. Simulations indicate that the proposed method outperforms the Langlie, Neyer and Dror and Steinberg methods in terms of probability of achieving an overlap and estimation precision. Moreover, the superiority of the proposed method are confirmed by two real applications.