A Parametric Approach Using Z-Test for Comparing 2 Means to Multi-Group Analysis in Partial Least Square Structural Equation Modeling (PLS-SEM)

Abstract

Aims: Multi-group analysis can be known as modeling the moderator variable since the strength of this method to moderates the influences of exogenous on endogenous variable. Basically, researchers interest to employ such method to extend their study to be more intense and practical. Sampling: This paper work used stratified sampling which is one of the probability sampling. Thus, the parametric method can be applied as set of statistical assumption. Methodology: Thus, multi-group analysis has become convenient to be practice in second generation modeling namely Structural Equation Modeling (SEM). Nevertheless, the limitation of Covariance Based Structural Equation Modeling (CB-SEM) causes the difficulties of researchers to further their studies. Hence, Partial Least Square Structural Equation Modeling (PLS-SEM) introduced to satisfy the necessity of researchers. Truthfully, multi-group analysis is not implement in most of PLS-SEM software. Result: Therefore, this paper intend to demonstrate a parametric approach using z-test to attain the probability level with the help of SmartPls 2.0. Generally, z-test comprised of two types of comparing for each groups namely means and population proportion. However, this paper work apply comparing of means to attain the value of z-scores. Consequently, the aimed of this paper work is success since the implementation of z-test approach to multi-group analysis. Conclusion: In this case, categorical variable which is constituted for 2 groups namely gender group was the one to be moderator variable. The findings reveal that all the exogenous construct that link on endogenous construct fail to meet the required level of family wise error rate.