Complex sampling designs for the Customer Satisfaction Index estimation

Abstract

In this paper we focus on sampling designs best suited to meeting the needs of Customer Satisfaction (CS) assessment with particular attention being paid to adaptive sampling which may be useful. Complex sampling designs are illustrated in order to build CS indices that may be used for inference purposes. When the phenomenon of satisfaction is rare, adaptive designs can produce gains in efficiency, relative to conventional designs, for estimating the population parameters. For such sampling design, nonlinear estimators may be used to estimate customer satisfaction indices which are generally biased and the variance estimator may not be obtained in a closed-form solution. Delta, jackknfe and bootstrap procedures are introduced in order to reduce bias and estimating variance. The paper ends up with a simulation study in order to estimate the variance of the proposed estimator.