Using Fine-grained Programming Trace Data to Inform Disciplinary Models of Self-Regulated Learning in Computing Education

对于本科生来说，计算机科学(CS)是一个越来越重要的话题；它是他们学习编程和开发软件的地方。然而，对于许多学生来说，这也是一门具有挑战性的学科。决定学生在现代CS课堂上取得成功的一个关键因素是他们是否以及如何使用自我调节学习(SRL)技能，如计划、目标设定和调整学习策略以应对新的挑战。这个项目的目的是帮助开发理论来解释学生如何参与自我调节，特别是在CS课堂上，以及这如何影响他们的学习结果。这一特定于CS的SRL模型对于帮助科学家开发改善学生学习的有效干预措施(例如，教学生如何设定有效目标)至关重要，因为它将帮助计算机科学教育工作者和研究人员预测学生可能对这些干预措施做出的反应。为了促进理论的发展，本项目将开发在CS课堂上学习SRL的新方法。目前，详细研究SRL的最好方法之一是让学生在工作和学习时大声思考，这样科学家就可以了解学生正在使用哪些SRL技能和策略，以及他们使用这些技能和策略的原因。然而，这是一个耗时的过程，而且在真正的课堂上使用是不切实际的。为了应对这一挑战，该项目将有声思考的数据与学生在使用技术学习时产生的日志数据配对，以验证如何将这些日志数据用作学生SRL的衡量标准。能够依赖这些自动记录的数据将使科学家能够在真实的CS课堂上大规模地研究SRL，并有大量不同的学习者群体。这些研究还有助于确定CS课程的SRL过程与其他STEM课程的SRL过程有何相似或不同之处。这个项目的结果将帮助CS教育研究人员开发更有效的学习干预措施，特别是针对CS中代表性不足的学生。本项目将通过验证和分析学生在真实本科课堂环境中与编程工具交互的细粒度跟踪数据，在计算机科学(CS)教育中应用、测试和阐述自我调节学习(SRL)模型。SRL包括学生通过计划、监控、控制和反思他们学到了什么和如何学习来深思熟虑地追求学术目标。这些行为是学生在许多领域取得成功的关键决定因素，包括CS。CS中基于学科的SRL模型仍处于初级阶段，在很大程度上没有经过严格的测试，反映了可以扩展到更大人群的数据收集和验证的尖端实践。SRCSL模式的开发需要从真实的CS教室大规模收集经过验证的SRL踪迹。为此，该项目首先将使用实验室研究来收集编程工具的数字跟踪数据和学生完成真实学习活动时的有声思考协议(TAP)数据，对TAP进行编码以识别SRL事件。研究人员将使这两个数据源保持一致，将数字记录的事件映射到它们反映的口头SRL事件，然后使用编码的TAP数据来验证关于SRL处理数字跟踪数据所指示的内容的推断。其次，研究人员将从两门CS课程收集大量跟踪数据样本(N SRL 2,000)，并应用经过验证的映射来从课堂上发生的数字跟踪数据中识别≈事件。这将提供关于SRL的顺序性、偶然性和动态性的分析，为SRCSL的经验支持的初始模型提供信息。该模型还将假设SRL行为如何预测学生的认知和非认知课程结果，这些关系如何在不同的课程背景下变化，以及它们如何受到学生个人特征的调节。该项目得到了NSF的EDU核心研究(ECR)计划的支持。ECR计划强调基础STEM教育研究，以产生该领域的基础知识。这一奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，认为值得支持。

Computer science (CS) is an increasingly important topic for undergraduate students; it is where they learn to program and develop software. However, it is also a challenging subject for many students to learn. A key determinant of students' success in modern CS classrooms is whether and how they use self-regulated learning (SRL) skills, such as planning, goal setting, and adapting their learning strategies to meet new challenges. The aim of this project is to help develop theory to explain how students engage in self-regulation, specifically in CS classrooms, and how this impacts their learning outcomes. This CS-specific model of SRL is critical to help scientists develop effective interventions for improving student learning (e.g., teaching students how to set effective goals) because it will help computer science educators and researchers predict how students might respond to those interventions. To advance theory, this project will develop new methods for studying SRL in CS classrooms. Currently, one of the best ways to study SRL in detail is to have students think out loud as they work and learn, so scientists can understand what SRL skills and strategies students are using and why they are using them. However, this is a time-consuming process, and it is impractical to use in real classrooms. To address this challenge, this project will pair think-aloud data with the log data students produce when learning with technology, to verify how to use this log data as a measure of students’ SRL. Being able to rely on these automatically logged data will allow scientists to study SRL in authentic CS classrooms, at scale, with large, diverse groups of learners. These studies may also help determine how SRL processes in CS courses are similar or distinct from those in other STEM courses. The results of this project will help CS education researchers to develop more effective learning interventions, especially for students who are underrepresented in CS.This project will apply, test, and elaborate a model of self-regulated learning (SRL) in computer science (CS) education (i.e., SRCSL) by validating and analyzing fine-grained trace data from students' interactions with programming tools in authentic undergraduate classroom settings. SRL encompasses students' thoughtful pursuit of academic goals by planning, monitoring, controlling, and reflecting upon what and how they learn. These behaviors are a key determinant of students' success in many domains, including CS. Discipline-based models of SRL in CS are still nascent and largely untested in rigorous ways that reflect cutting-edge practices for data collection and validation that can scale to larger populations. The development of a SRCSL model requires large-scale collection of validated SRL traces from authentic CS classrooms. To do so, first the project will use laboratory studies to collect both digital trace data from programming tools and think-aloud protocol (TAP) data from students as they complete authentic learning activities, coding TAPs to identify SRL events. Researchers will align these two data sources, mapping digitally logged events to the verbal SRL events that they reflect, and then use the coded TAP data to validate inferences about what SRL processes the digital trace data indicate. Second, researchers will collect a large sample (N ≈ 2,000) of trace data from two CS courses and apply the validated mapping to identify SRL events, from digital trace data, as they occur in the classroom. This will afford analyses regarding the sequential, contingent, and dynamic nature of SRL, informing an empirically-supported initial model of SRCSL. The proposed model will also posit how SRL behaviors predict students' cognitive and non-cognitive course outcomes, how these relationships vary across course contexts, and how they are moderated by students' personal characteristics. This project is supported by NSF's EDU Core Research (ECR) program. The ECR program emphasizes fundamental STEM education research that generates foundational knowledge in the field. Investments are made in critical areas that are essential, broad and enduring: STEM learning and STEM learning environments, broadening participation in STEM, and STEM workforce development.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.