Comparison of Covariance-Based and Partial Least Square Structural Equation Modeling Methods under Non-Normal Distribution and Small Sample Size Limitations

Abstract

This study contains a repetition of the data analysis part of a research conducted on building the trust of generation Y customers in B2C websites. In this base study, since the samples size was a limitation of the study, analyses were conducted again by using CB-SEM and PLS-SEM methods separately and the results were compared in small sample size and non-normal distribution. Consistent results were obtained in all of tests of the hypotheses. Finding of this study highlighted the differences between CB-SEM and PLS-SEM and also compared the results obtained different estimation methods of ML, ADF and GLS under non-normal distribution and small sample size limitations.