Determining the Importance of Perceived Quality Attributes on Users Satisfaction: A Latent-Class-Based Mimic Model

Abstract

We propose to use Latent Class-based MIMIC models (LC-MIMIC) for the analysis of determinants of user’s satis- faction in customer surveys. This method overcomes the limitations of the traditional approach based on regression, for this type of analysis. LC-MIMIC models are a kind of Structural Equation Models, composed by a latent cate- gorical endogenous variable, which receives direct effects of exogenous manifest variables. We demonstrate the application of this model to a satisfaction survey applied to 302 users of a helpline, in Chile. The most important determinants of user satisfaction with the service are those associated with the quality of the answers given by the operators. Other factors, such as kindness, schedules or timeouts have little influence on user satisfaction. The pro- posed methodology enables a robust analysis of customer surveys, and improves communicability of results to users;however it requires at least two questions to assess user satisfaction. Empirically, this is the first Chilean study where user’s satisfaction is addressed in the context of helplines. On a methodological ground, the originality of the study is the use of MIMIC models with Latent Classes.