Mixed Effects Modeling in Conjoint Analysis with Multivariate Normal Responses

• Activity Number: 227
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Government Statistics Section
• Author(s): Tanita Cronje*+ and Frans H.J. Kanfer and Sollie Millard and Mohammad Arashi
• Companies: University of Pretoria and University of Pretoria and University of Pretoria and Shahrood University
• Keywords: Conjoint analysis ; Multivariate normal distribution ; Mixed effect models

Abstract

In this decision-driven era, it has become vital for modelers to efficiently model consumer choices and preferences (from a marketing perspective for instance). Conjoint analysis is a known method which has been used to perform such analyses. A mixed effects model is proposed to perform a conjoint analysis with normal responses, illustrated by an application of modeling respondent's preferences to different industrial detergents. The proposed model allows for predicting how observed attributes (which describes a product in terms of its characteristics and features) of decision makers and choice options, influence decisions. Inference regarding the parameters of the proposed model with a normal distribution is discussed in the mixed effect conjoint setting. Extensions of this model, regarding Bayesian prior selection are also discussed.