Some Uses of Rasch Models Parameters in Customer Satisfaction Data Analysis

Abstract

Rasch model is becoming an interesting and performed tool in customer satisfaction studies. Through the Rasch analysis it is possible to obtain independent measures for both customers and items devoted to collect customers’ opinions about quality. The aim of this paper is to use these parameters to perform further analysis: item parameters and thresholds parameters allows us to check items and responses categories coherence; estimated person parameters are instead used as dependent variables in multilevel models to study relations between customers’ satisfaction and other variables included in a hierarchical structure. To illustrate the proposed method an application with students’ satisfaction data is carried out.

Key Words:

• Customer satisfaction
• multilevel models
• Rasch models

Additional information

Notes on contributors

Laura Pagani

Laura Pagani is an Associate Professor of Statistics in the Department of Statistics at the University of Udine. She graduated in Political Sciences at the University of Milan, and received her PhD in Methodological Statistics from the University of Trento. Her research interest include model for categorical data analysis (log linear model for ordinal categorical variables, logit models), Event History Analysis for the analysis of historical demographic data, Survival Models in discrete time, Multilevel Models, data reduction models for categorical variables (Nonlinear Principal Component Analysis), Rasch models, Hierarchical Generalized Linear Rasch Models with a special focus on application in educational, health and customer satisfaction.

Maria Chiara Zanarotti

Maria Chiara Zanarotti is an Associate Professor of Statistics in the Department of Statistical Sciences at Catholic University of Milan. After university studies in Political Sciences at University of Milan, received a doctorate in Methodological Statistics from University of Trento. Her research interests include categorical data analysis, (Logit and Probit models), dynamic ordinal models, switching models, hidden Markov models. She also took care of statistics education, theories of learning and teaching statistics. Her current research interests are services quality measure, Rasch models and multilevel analysis.