Usefulness of Neural Network as Pre-run to Structural Equation Modeling Analysis: Case of Hotels

Abstract

Judging by a lack of analytics approach such as the use of artificial neural networks (ANNs) in
social science and management research, this study demonstrates the complementary advantage of ANNs
to the traditional theory-driven hypothesis testing. A valid 151 dataset, involving the participations of the
hotel management and employees, both front and back offices, across chain and non-chain hotels, of
below 3 stars, 3 to 4 stars, is used in ANNs simulation and structural equation modeling (SEM)
computation. The empirical tests reveal the usefulness of neural network as pre-run to structural equation
modeling analysis. As a closure, numerous scopes of implications are also presented, highlighting some
important theoretical characteristics revealed from the SEM, such as towards the use of the “complex
adaptive system” concept.