So sánh CB-SEM và PLS-SEM trong kiểm định mô hình đo lường tính đổi mới của người tiêu dùng tại Thành phố Hồ Chí Minh

Abstract

Structural equation modeling (SEM) has become a popular technique with a dramatic increasing in publications. Two main approaches to estimate structural equation models include covariance-based structural equation modeling (CB-SEM) and partial least squares structural equation modeling (PLS-SEM). The purpose of this study is to compare the testing measurement model results from CB-SEM and PLS-SEM, using a dataset of consumer innovativeness in Ho Chi Minh city. The study shows that consumer innovativeness is measured with four different levels of consumer innovativeness, including innate innovativeness (II), domain-specific innovativeness (DSI), vicarious innovativeness (VI) and behavioral innovativeness (BI). All constructs were found to achieve acceptable reliability, convergent validity, discriminant validity, nomological validity, and model fit criterion. The study also finds that results from CB-SEM and PLS-SEM approaches are likely similar in measurement model and structural model testing; however, PLS-SEM has more advantages in some statistical analysis.