Computational thinking (CT), as one of the key skills in the twenty-first century, has been integrated into educational programming as an important learning goal. This study aims to explore CT processes involved in pair programming with the support of visual flow design. Thirty freshmen participated, working in pairs to solve two programming problems. Their discourses were recorded, transcribed, and coded based on a CT framework encompassing cognitive, practical, and social perspectives. Both quantitative and qualitative methods were applied to analyze the data. In particular, Epistemic Network Analysis (ENA) was applied to explore the patterns of their CT processes. The findings revealed that social perspectives emerged the most frequently in all pairs’ discourses. The high-level groups (HLGs) focused more on practical and social perspectives whereas the low-level groups (LLGs) emphasized more on cognitive perspectives. The ENA networks revealed that social perspectives mostly centered around cognitive perspectives for all pairs with CT process patterns in HLGs crossing the three perspectives more frequently. In addition, HLGs exhibited a more complicated and developmental trend in solving the two problems, while LLGs displayed a relatively similar CT pattern. The current study provides insights into the design and implementation of collaborative learning activities in educational programming.

计算思维（CT）作为二十一世纪的关键技能之一，已经作为重要的学习目标融入到教育规划中。本研究旨在探索在可视化流程设计支持下结对编程所涉及的 CT 流程。三十名新生参与其中，两人一组解决两个编程问题。他们的话语是基于涵盖认知、实践和社会观点的 CT 框架进行记录、转录和编码的。应用定量和定性方法来分析数据。特别是，认知网络分析（ENA）被用来探索他们的 CT 过程的模式。调查结果显示，社会观点在所有配对的谈话中出现得最频繁​​。高级别群体（HLG）更注重实践和社会观点，而低级别群体（LLG）更强调认知观点。ENA 网络显示，对于所有 HLG 中具有 CT 过程模式的配对，社会视角主要集中在认知视角，更频繁地跨越这三个视角。此外，HLGs在解决这两个问题上表现出更加复杂和发展的趋势，而LLGs则表现出相对相似的CT模式。当前的研究提供了对教育规划中协作学习活动的设计和实施的见解。