MULTIGROUP GENERALIZED STRUCTURED COMPONENT ANALYSIS METHOD ON COMPONENT-BASED MULTIGROUP SEM

Abstract

Structural Equation Modeling (SEM) is a statistical modeling that combines factor analysis, regression analysis, and path analysis that aims to examine the relationships between variables in a model simultaneously. Generalized Structured Component Analysis (GSCA) is a strong component-based SEM method because it is not limited by parametric assumptions. One extension of the GSCA for sample data that is divided into several sub-sample groups is known as the multigroup GSCA. The aim of this research is to apply the multigroup GSCA method to the data of factors that influence supplement consumption behavior which is reviewed by gender, and to test the significance of these gender variables. The results of this study are that in the multigroup GSCA model constructed, all indicators have a positive and significant influence on each latent. Furthermore, it was found that the latent variables that most significantly influence a person's behavior in taking supplements are the norm (other people's opinion) in the female group and control (the level of opportunities and resources encountered) in the male group. The AFIT value obtained at 0.527 indicates that the model is able to explain about 52.7$\%$ of the variation of the data. The significance test of the gender variable using the t test shows that gender does not affect the causality relationship between attitudes and intentions, norm and intentions, control and intention, as well as control and behavior. This represents that in both the women's and men's groups, the four pathways have the same effect. Conversely, gender influences the causal relationship between attitude and behavior as well as intention and behavior, meaning that each of these pathways has a significantly different effect on the two groups.