Optimal experimental design for hyperparameter estimation in hierarchical linear models with application to marketing.
by
Angela Dean
Statistics Department, The Ohio State University, Columbus, Ohio, USA
Coauthors: Qing Liu and Greg Allenby, Statistics and Marketing Departments, The Ohio State University

Marketing, and business in general, requires an understanding of when effect sizes are expected to be large and when they are expected to be small. Gaining an understanding of the contexts in which consumers are sensitive to promotional offers, and to other variables such as price, is an important aspect of merchandising. Hierarchical models are today being used successfully to estimate the importance of product attributes in the presence of subject heterogeneity. In this talk, experimental designs for the efficient estimation of the hyperparameters in a hierarchical linear model will be discussed and illustrated through a study of the "level effect" in conjoint analysis. The level effect is the phenomenon, observed in many psychological and marketing studies, that the importance of a factor as perceived by a respondent increases with the number of levels presented to that respondent.

Date received: June 14, 2006