Saturday, February 25
PS3 Poster Session 3 and Continental Breakfast Sat, Feb 25, 8:00 AM - 9:15 AM
Conference Center AB

Optimal Experimental Designs for Mixed Categorical and Continuous Responses (303467)

Ming-Hung Jason Kao, Arizona State University 
*Soohyun Kim, Arizona State University 

Keywords: Complete class, Convex optimization, Design efficiency, Generalized linear model

The study concerns statistical planning for experiments where the bivariate response consists of possibly correlated categorical and continuous variables. Many experiments in engineering, medical studies and other fields have such mixed responses. For example, efficacy and toxicity are simultaneously observed to provide complementary information on the effect of a drug in clinical studies. Although several statistical methods have been proposed for jointly analyzing these two types of variables, a way to design such experiments remains rather unclear. In this study, we aim to find optimal experimental designs for mixed categorical and continuous responses. Our study is expected to allow experimenters to collect the most informative data from this type of experiments. In particular, we develop some results to significantly reduce the number of candidate designs, and implement an algorithm for nonlinear optimization to search for optimal designs. The optimality of obtained designs is verified by a general equivalence theorem.