The Impact of Gender, Academic Performance, and Programming Ability on Computational Thinking: Epistemic Network Analysis Based on Programming Tests

ABSTRACT Background Previous Epistemic Network Analysis (ENA) studies examined CT under various instructional strategies, performance levels and scaffolding. However, limited work addresses how gender, academic performance and programming ability shape CT. Objectives This study aims to investigate how gender, academic performance and programming ability influence students' computational thinking using ENA. Methods This study employed retrospective think‐aloud protocols to capture students' CT ( N = 486), encoded them into quantitative data, and applied ENA for analysis. Results and Conclusions Whether the differences lie in gender, academic performance, or programming ability, disparities in CT network structures were consistently reflected in the connection strength of the ‘concepts–practices–perspectives’ triad and in the fluency of transfer, explanation, and reflection processes. The findings revealed that male and female students' epistemic networks were complementary, characterised respectively by interactive‐constructive and self‐constructive orientations. For academic performance and programming ability, high‐level groups exhibited more complex, richly connected, and well‐balanced networks. These results highlight the need for differentiated instruction tailored to distinct groups, encouraging heterogeneous pairing to foster the complementarity of group advantages. This study contributes to the ongoing discussion of CT and provides actionable insights for the design of personalised CT instruction.