This paper presents a robust approach for modeling customer satisfaction in services. This approach represents an extension of the method proposed by Tinoco and Ribeiro (2007), through the inclusion of robust statistics that can adequately represent the causal relations involving customer satisfaction in a service environment, even in the presence of outliers (inconsistent respondents). These robust procedures allow one to identify inconsistent respondents and to assign a weight according to the agreement of each questionnaire based on the average of the observations. The approach proposed in this work is demonstrated by building a general model of customer satisfaction in services. For this purpose, twelve segments of services, chosen to represent different service types with respect to tangibility, customer contact and customer involvement degree, were assessed. The robust approach confirmed the most significant relationships between determinants of satisfaction obtained from the traditional approach proposed by Tinoco and Ribeiro (2007) and also incorporated a further important relationship that would not be considered from the traditional technique.
Robust modeling; Customer satisfaction; Services
