Partial least squares structural equation modeling (PLS-SEM) in the AI Era: Innovative methodological guide and framework for business research

Abstract

Partial Least Squares Structural Equation Modeling (PLS-SEM) serves as a comprehensive methodological framework, critically addressing theoretical underpinnings, rigorous analytical approaches, and state-of-the-art modeling techniques vital for contemporary business research. The methodological discussion includes detailed exploration of reflective and formative measurement models, structural model specification, reliability, and validity assessments, alongside advanced analytical methods such as Confirmatory Tetrad Analysis (CTA-PLS) and Importance-Performance Matrix Analysis (IPMA). Advanced algorithms including bootstrapping and blindfolding procedures are elaborated, emphasizing predictive relevance and methodological precision. Partial Least Squares Structural Equation Modeling further offers robust analytical capabilities to evaluate modern AI-driven innovations, facilitating sophisticated assessment of user trust, perceived accuracy, and satisfaction with recommender systems, voice assistants, autonomous vehicles, AI-driven healthcare diagnostics, personalized educational platforms, and fraud detection technologies. Ethical considerations, reporting best practices, computational tools (SmartPLS, SEMinR), and Explainable AI (XAI) integration enhance the comprehensive nature of this framework. Furthermore, integration of cutting-edge analytical approaches such as moderation, mediation, Multi-Group Analysis (MGA), nonlinear modeling, machine learning integration, and quantum computing potential positions PLS-SEM as indispensable for contemporary business and technology research, ultimately promoting actionable scholarly insights and ensuring maximum methodological impact.