Segmenting the heterogeneity of tourist preferences using a latent class model combined with the EM algorithm

Abstract

An important component of conjoint analysis is market segmentation where the main objective is to address the heterogeneity of consumer preferences. Latent class methodology is one of the several conjoint segmentation procedures that overcome the limitations of aggregate analysis and a-priori segmentation. The main benefit of Latent class models is that they simultaneously estimate market segment membership and parameter estimates for each derived market segment. In this paper we present two latent class models. The first model is a latent class metric model using mixtures of multivariate conditional normal distributions to analyze rating data. The second is a latent class multinomial logit model used to analyze choice data. The EM algorithm is employed to maximize the likelihood in both models. The application focuses on tourists’ preference and choice behaviour when assessing package tours. A number of demographic and product related explanatory variables are used to generate segments that are accessible and actionable. A Monte Carlo study is also presented in this paper. This study examines how the number of hypothetical subjects, number of specified segments and number of predictors affect the performance of the latent class metric conjoint model with respect to parameter recovery and segment membership recovery.