Optimal Reduced Size Choice Sets with Overlapping Attributes

Abstract

Discrete choice experiments are used when choice alternatives can be described in terms of attributes. The objective is to infer the value that respondents attach to attribute levels. Respondents are presented sets of profiles based on attributes specified at certain levels and asked to select the profile they consider best. When the number of attributes or attribute levels becomes large, the profiles in a single choice set may be too numerous for respondents to make precise decisions. One strategy for reducing the size of choice sets is the sub-setting of attributes. However, the optimality of these reduced size choice sets has not been examined in the literature. We examine the optimality of reduced size choice sets for 2^n experiments using information per profile (IPP) as the optimality criteria. We propose a new approach for calculating the IPP of designs obtained by dividing attributes into two or more subsets with one, two, and in general, r overlapping attributes, and compare the IPP of the reduced size designs with the original full designs. Next we examine the IPP of choice designs based on 3^n factorial experiments. We calculate the IPP of reduced size designs obtained by sub-setting attributes in 3^n plans and compare them to the original full designs.