Propuesta metodológica mediante diseños Box-Behnken para mejorar el rendimiento del análisis conjunto en estudios experimentales de mercado

Abstract

Conjoint analysis is a technique used to study consumer preferences in market research. One of the most important issues is to determine the choice set which respondents must assess; usually factorial designs to estimate part-worth factors have been used. But, if the researcher is also interested in estimating two or more factor interactions, factorial designs require such a large number of alternatives that makes their evaluation very difficult, leading respondents to not use compensatory criteria. Using Box-Behnken designs in blocks reduce the cognitive effort made by respondents and, at the same time, it allows fitting a quadratic model. This paper illustrates, through an experiment, the properties and how to use Box-Behnken designs in market research. Results showed a better performance of these models when compared with standard factorial designs.