Decoding Structural Equation Modeling: Insights on Data Assumptions, Normality, and Model Fit in Advancing Digital Marketing Strategies

Abstract

The study utilizes structural equation modeling to examine issues related to normality, missing data, and sampling errors in digital marketing engagement research. The primary focus is on exploring relationships between self-esteem, social comparison, social interactions, perceived social support, and psychological well-being, with perceived social support as a mediating factor. Confirmatory factor analysis is applied to evaluate model fit using data from 400 social media users. Skewness and Kurtosis values are assessed to ensure normality, with scores kept within the acceptable range of -2 to +2. Questionnaires with over 30% missing values are excluded to maintain data quality, and the “10-times rule” is used to ensure adequate sample size and reduce sampling errors. Results confirm a normal distribution and indicate that the model aligns with SEM assumptions, meeting all fit indices. The research offers insights into SEM's application in digital marketing and suggests future studies should investigate advanced modeling techniques for further exploration.