An Empirical Analysis of Approach‐Based Metrics Model for Architectural Erosion Detection

ABSTRACT Context Software architecture plays a crucial role in the success or failure of software development and design. Over time, architectural degradation—known as architectural erosion—can compromise system quality and maintainability. Metrics‐based approaches have become a prevalent method for identifying and addressing this phenomenon. Objective This study aims to (1) investigate whether various metrics‐based approaches can effectively determine architectural erosion and support the development of a formal model, and (2) assess the construct reliability and validity of the proposed model. Methods A formal model was constructed based on selected metrics approaches derived from the literature. A questionnaire‐based survey was conducted with 130 software engineering professionals experienced in architectural erosion and metrics. Structural Equation Modelling (SEM) was employed to evaluate the model's construct reliability, construct validity, and research hypotheses. Results The analysis revealed significant relationships between most classifications of metrics‐based approaches and architectural erosion. However, no substantial relationships were found for the classifications of architectural complexity and architectural technical debt. Conclusion The proposed model offers a valuable foundation for the formal definition and evaluation of metrics‐based approaches to architectural erosion. Its empirical validation provides meaningful insights for both researchers and practitioners aiming to assess or mitigate architectural erosion in software systems.