An Application of Data Fusion to Conjoint Analysis Using Test-Retest Format

Abstract

While conjoint analysis has become a major technique to analyze consumers` preference , one of the major problems in conjoint analysis is that as the number of attributes and attribute levels increases, the number of product profiles increases multiplicatively. As a way to solve this problem, we applied data fusion approach to the estimation of conjoint part-worths. While previous research applied fusion approach to conjoint analysis in a post hoc manner by deleting and recovering the known value, we carried out two conjoint experiments in a test-retest format. One involved a full set of attributes and the other a partial set of attributes (with one or two attributes deleted). Results show that estimated values of conjoint part-worths (via fusion) were closed to the actual value (estimated from a fullest of attributes).