Assumptions for Structure Equation Modeling (SEM), Normalityof Data Distribution Analysis & Model Fit Measures

Abstract

Structure Equation Modeling (SEM) is a well-known research technique. Before proceed further in data analysis, the researcher describes the fundamentals of Structure Equation Modeling (SEM), as well as its modeling criteria, assumptions, and concepts. The researcher uses Structure Equation Modeling (SEM) to make assumptions about normality, missing data, and sampling errors measurement. In evaluating the model’s fit with the data, Confirmatory Factor Analysis (CFA) starts with a model that anticipates the existence of a predetermined number of latent factors as well as the indicator variables that each factor will load on. Firstly, Normality Test. The “Skewness and Kurtosis” scores of the assessment model, Confirmatory Factor Analysis (CFA), range from -2 to +2. Independent Variables adopted are Self-Efficacy (SE), Perceived Benefits (PB), Behavioural Beliefs (BB), Mediating Variable is Consumer Innovativeness (CI), and Dependent Variable is Health Protective Behaviours (HPB). This study used a total sample size of 400 respondents of private healthcare customers, indicating that the data was normally distributed and satisfied Structure Equation Modeling (SEM)’s normality predictions. Secondly, missing data check.
Missing data would jeopardise the statistical analysis in later part and the result might not be able to represent the idea from population. As a result, those questionnaires with more than 30% missing value would b eliminated and excluded from analysis to prevent such phenomena happened. Thirdly, measurement and sampling errors. Minimising sampling error was done by using suitable sample size. A widely used minimum sample size estimation method in PLS-SEM is the “10-times rule” method, which builds on the assumption that the sample size should be greater than 10 times the maximum number of inner or outer model links pointing at any latent variable in the model. Lastly, model fit measures. The study model meets all of the fit indices in general [1].