CMStatistics 2021: Start Registration
View Submission - CMStatistics
B0652
Title: Optimal two-level designs under model uncertainty Authors:  Steven Gilmour - KCL (United Kingdom) [presenting]
Pi-Wen Tsai - National Taiwan Normal University (Taiwan)
Abstract: Two-level designs are widely used for screening experiments where the goal is to identify a few active factors which have major effects. Most work on two-level designs is based on the effect hierarchy assumption that lower-order effects are of more importance than higher-order effects, so the focus is on two-level designs with level balance and pairwise orthogonality. We apply the model-robust $Q_B$ criterion for the selection of optimal two-level designs by incorporating experimenters' prior knowledge on the importance of each effect into the optimality criterion. We find a smooth relationship between the choice of designs and the experimenters' prior beliefs. Additionally, we provide a coordinate exchange algorithm for the construction of $Q_B$-optimal designs without the restrictions of level-balance and pairwise orthogonality.