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

Abstract

Software architecture determines success or failure in the domain of software development and design. As a system evolves, software architecture erodes. This phenomenon is called architectural erosion. Several studies that focused on various approaches to the problem of architectural erosion have been conducted. As a direct consequence of this, the metrics strategy has emerged as the most widely used solution for architectural erosion. However, providing a comprehensive perception of the elements required to evaluate the phenomenon of architectural erosion with an acceptable level of quality is a challenge. The primary goals of this research, which drew from the prior literature about identifying architectural erosion, were to (1) determine whether various adopted measures approaches determine architectural erosion in order to develop a formal model, and (2) evaluate the construct reliability and construct validity of the model. This research presents a model based on the chosen measures approaches for identifying architectural degradation. This model can be used as the cornerstone for a formal definition of general approaches and adopted metrics. Data was collected from 130 software engineering professionals with experience in architecture erosion and software metrics via a questionnaire-based survey. Structural equation Modelling (SEM) was used to analyse construct reliability, construct validity, and research hypotheses. The results demonstrate a substantial association between all metrics approach classes and architectural erosion, except for architectural complexity and architectural technical debt. Both researchers and practitioners can significantly benefit from this model’s empirical assessment and evaluation, which includes a valuable information in this context.