CMStatistics 2022: Start Registration
View Submission - CMStatistics
B0868
Title: Strategies to construct directly optimal and near-optimal symmetric paired choice experiments for main effects models Authors:  Abdulrahman Sultan S Alamri - RMIT University (Australia) [presenting]
Abstract: Discrete choice experiments (DCEs) are increasingly used for identifying the underlying influences on an individual's choice behaviour in various fields, e.g., health resources, marketing, transport, economics, and the list goes on. Choosing the DCE design plays an essential role in defining which effects are observable. For paired choice experiments, we present the optimal form of the information matrix for attributes at two levels and main effects models. Moreover, we apply globally D-optimal designs to construct DCEs and address some identification issues by suitably modifying the constructions of $D$-optimal designs. For this situation, we cover the part where some practitioners somehow may need to use choice sets that are of size other than zero modulo $4$, i.e. $N \not \equiv 0 mod 4$. Furthermore, as against the existing efficient designs, our designs have higher D-efficiencies for the same number of choice pairs. Also, our design techniques can be extended to be applied to include situations where attributes of DCEs have a higher number of levels with sufficiently small sample sizes.