A Decade of Progress in Hierarchical Linear Modeling (2015–2025): Methodological Advances, Interdisciplinary Applications, and Open Challenges

Abstract Hierarchical Linear Models (HLM), or multilevel models, provide a core statistical framework for analyzing nested data structures. This review synthesizes methodological developments and interdisciplinary applications of HLM from 2015 to 2025, contextualized by selectively incorporated foundational works. Over the past decade, the framework has expanded substantially, driven by the widespread adoption of hierarchical generalized linear models for non-normal outcomes and by deeper integration with Bayesian inference and structural equation modeling. These developments have enabled the routine use of multilevel structural equation models and latent growth curve models, facilitating the analysis of latent constructs, measurement error, and dynamic processes of change. Contemporary applications span education, psychology, public health, medicine, and longitudinal omics research. Despite this progress, persistent challenges remain regarding statistical power, causal inference, and high-dimensional data. The review concludes by outlining a forward-looking methodological agenda, underscoring the continued evolution and relevance of hierarchical modeling. Keywords: Hierarchical Linear Models (HLM); Multilevel Modeling; Systematic Review; Multilevel Structural Equation Modeling (ML-SEM); Longitudinal Data Analysis; Bayesian Multilevel Modeling; Cross-Classified Models; Causal Inference