Subsample analysis in structural equation modelling example from customer satisfaction modelling in banking industry

Abstract

This paper stresses the importance of sub-sample analysis in the context of Structural Equation Modelling. Based on original survey on 459 banking customers, structural model of latent variables pricing policy, individual approach to customer and customer satisfaction was investigated with respect to sex groups. We've found that pattern of females' answers is inconsistent with economic theory-driven expectations and differ a lot from answers obtained from male subsample. Combining both subsets resulted in expected behaviour pattern -this was the problem we've decided to analyse. We show that although overall model exhibit good fit characteristics, results cannot be generalised on both sexes. This is because of different values of estimated coefficients and their corresponding standard errors on subgroups.