Errors in Customer Satisfaction Surveys and Methods to Correct Self-Selection Bias

Abstract

This paper provides an overview of the main types of surveys carried out for customer satisfaction analysis according to the dimension of the firm and to the data collection method. Different errors can be generated by these methods, causing biases in the results. We focus on the self-selection error and we employ three methods to correct the bias associated to it. A simulation underlines the performances of the three methodologies.

Key Words:

• Customer satisfaction surveys
• Heckman two-step procedure
• hierarchical Bayesian approach
• propensity score matching
• self-selection bias

Additional information

Notes on contributors

Nicolini Giovanna

Giovanna Nicolini she has been Associate Professor of Sample Theory at the University of Milano Bicocca, now she is full Professor of Statistics at University of Milan. Her main research areas are the following: probabilistic models, robustness on power statistical tests, tests with balanced errors, permutation tests, sampling techniques, customer satisfaction measures, assessment of the quality of public services.

Dalla Valle Luciana

Luciana Dalla Valle earned a Ph.D. in Statistics from the University of Milano-Bicocca in 2007. Currently, she holds a Post-Doctoral position at the Department of Economics, Business and Statistics of the University of Milan. She is interested in Bayesian statistics, Markov Chain Monte Carlo methods, statistical models for financial risks, copula modeling for financial applications, statistical models for internationalisation and sampling techniques.