Permutation Test, Non-parametric, and Confidence Set Approaches to Multi Group Analysis for Comparing 2 Groups Using Partial Least Square Structural Equation Modeling (Pls-Sem)

Abstract

Partial Least Square Structural Equation Modeling (PLS-SEM) is becoming more prominent as an alternative to Covariance Based Structural Equation Modeling (CB-SEM) because the technique employ is much comfortable. Thereby, this research paper intend to present guide on how to carry on the Partial Least Square based on Multi-Group Analysis (PLS-MGA) using categorical variable. In particular, the discussion of PLS-MGA comprises of three approaches namely permutation test, non-parametric test, and non-parametric confidence set interval. The three approaches are established as non-parametric test in which no statistical assumption of normality is assumed. Thus, this paper is aimed at determining which approach is more appropriate to apply so as to present the guide for readers. Moreover, the practice of Square Multiple Correlation (R2) also has been sustained to identify the importance and performance of each exogenous constructs applied. Once executed three approaches on the same data, two approaches namely permutation test and non-parametric test suggest all of these exogenous constructs applied cannot be moderated via gender group between exogenous and endogenous constructs. In addition, the capability of R2 is proved can be extended to determine the importance and performance of independent variables. This paper is an attempt to show how the three approaches namely permutation test, non-parametric test, and non-parametric confidence set interval is achieved.