Toward Improving Software Functionality through Refactoring Operations: An Experimental Study

Abstract

Refactoring is widely used to improve internal software quality; however, its impact on external functionality remains insufficiently explored. This study investigates how different refactoring operations influence software functionality through a controlled experimental analysis. A set of ten commonly used refactoring operations was applied to the jEdit system. Software functionality was quantitatively evaluated using a composite metric derived from cohesion, polymorphism, interface size, design size, and inheritance hierarchy. The selected ten refactoring operations were performed 453 times across ten independent experiments in jEdit. The results reveal that refactoring operations do not have uniform effects on functionality. Operations that enhance encapsulation and modular distribution significantly improve functionality, with Encapsulate Field achieving the highest increase. In contrast, operations that reduce abstraction, such as Inline Method and Inline Class, negatively impact functionality. Additionally, some operations show no measurable effect, indicating limitations in metric sensitivity. These findings demonstrate that refactoring should not be assumed to universally improve software functionality. Instead, its impact depends on the nature and context of the operations being applied. The study provides empirical evidence and practical guidance for selecting refactoring operations that effectively enhance functional quality while avoiding potential degradation.