This paper outlines the methodological approach to Structural Equation Modelling (SEM) model selection employed in a quantitative study investigating cognitive biases in workplace decision-making in Singapore. Building upon a comprehensive systematic literature review (Ohms 2025g), which identified key research gaps and prominent cognitive biases, and a subsequent paper detailing the research framework and hypothesis development (Ohms 2025f), this study aimed to holistically assess the complex interplay between biases, decision-making stages, and moderating factors. The robust methodological design (Ohms 2025e) and data validation procedures (Ohms 2025a), including outlier detection, normality assessment, and reliability testing, provided the essential foundation for advanced statistical analysis. This paper specifies, estimates, and systematically refines the SEM model through a backward elimination strategy. It elaborates on the choice of a robust estimator, the interpretation of various model fit indices (e.g., Chi-square, CFI, TLI, RMSEA, AIC, BIC), and the iterative selection of the most simplified yet well-fitting model (Model 15). By thoroughly documenting the rigorous SEM model selection methodology, this paper contributes to establishing a strong analytical basis for subsequent empirical investigations into direct and moderated relationships, ensuring the reliability and validity of the holistic insights derived from the research.
Benjamin Ohms (Tue,) studied this question.