Adaptive programming practice often relies on fixed libraries of worked examples and practice problems, which require substantial authoring effort and may not correspond well to the logical errors and partial solutions students produce while writing code. As a result, students may receive learning content that does not directly address the concepts they are working to understand, while instructors must either invest additional effort in expanding content libraries or accept a coarse level of personalization. We present an approach for knowledge-component (KC) guided educational content generation using pattern-based KCs extracted from student code. Given a problem statement and student submissions, our pipeline extracts recurring structural KC patterns from students' code through AST-based analysis and uses them to condition a generative model. In this study, we apply this approach to worked example generation, and compare baseline and KC-conditioned outputs through expert evaluation. Results suggest that KC-conditioned generation improves topical focus and relevance to students' underlying logical errors, providing evidence that KC-based steering of generative models can support personalized learning at scale.

A significant portion of student programming submissions in CS1 learning environments are uncompilable, limiting their use in student modeling and downstream knowledge tracing. Traditional modeling pipelines often exclude these cases, discarding observations of student learning. This study investigates automated program repair as a strategy to recover uncompilable code while preserving students' structural intent for use in student modeling. Within this framework, we assess large language models (LLMs) as repair agents under high- and low-context prompting conditions. Repairs were evaluated for compilability, edit distance, and preservation of students' original structure and logic. While all models produced compilable repairs, they differed in how well they preserve students' control flow and code structure, affecting their pedagogical utility. By recovering uncompilable submissions, this work enables richer and more comprehensive analyses of learners' coding processes and development over time.

Prior research has explored the impact of various demographic and psychological factors in introductory computer science courses (CS1), but there remains a gap in understanding how course engagement patterns relate to students' performance and how these behaviors are influenced by demographic characteristics. This study investigates the relationships between student characteristics, course behaviors, and performance in a CS1 course taught using a flipped classroom model. We analyzed data from 410 engineering students, examining factors including gender, prior programming experience (PPE), grade point average (GPA), and learning self-efficacy (LSE). The study focused on four key course behaviors and attitudes: engagement with pre-class recorded lectures, self-reported class attendance, perceived availability of support, and perceived quality of in-class activities, exploring how these related to exam performance and student characteristics. Analysis indicated significant positive correlations between exam performance and three factors: students' engagement with pre-class recorded lectures, perceived availability of support, and perceived quality of in-class activities. Self-reported class attendance did not relate to exam scores. Female students reported significantly higher engagement with pre-class recorded lectures compared to male students, but lower perceptions of available support. Students with higher GPAs reported lower engagement with pre-class lectures, yet indicated higher perceptions of available support and found in-class activities more helpful. These findings highlight important behavioral and perceptual differences in how students engage with CS1 flipped classrooms. Understanding these patterns could enhance learning environments that better support all students, particularly those who may feel less included in computing education.

Clear, well-chosen names for variables and functions significantly enhance code readability and maintainability. In computer science education, teaching students to select appropriate identifiers is a critical task, especially in CS1. This study explores how large language models (LLMs) could assist in teaching this skill. While prior research has explored the use of LLMs in programming education, their precision and consistency in teaching code quality, particularly identifier selection, remains largely unexplored. For this purpose, this study investigated how well different LLMs can detect and report misleading identifiers. In a dataset of 33 code samples, we manually labeled misleading identifiers. On this dataset, we then tested five different LLMs on their ability to detect these misleading identifiers, measuring the overall accuracy, precision, recall, and f-score. Results revealed that the most successful model, GPT-4o, was able to correctly detect most of the manually flagged misleading variable names. However, it also tended to flag issues with variable identifiers in cases where the human evaluators would not, and refined prompting was not able to discourage this behavior.

Conversational AI (C-AI), like OpenAI's ChatGPT [38] or Google's Gemini [1], has seen a surge in development in recent years, driven by advancements in large language models. C-AI has the unique capability to instantaneously communicate with others using vast and contextual knowledge, providing personalized assistance tailored to individual needs. While the specific applications and advantages of conversational technologies are still being explored, prior research has noted the potential for conversational agents to serve in pedagogical settings, such as teaching agents, collaborative partners, or motivational tools [13,30]. The successful development and implementation of pedagogical C-AI relies on an understanding of learners' perceptions, trust, and overall acceptance of C-AI. There is a need for a comprehensive understanding of the factors influencing learners' trust and acceptance of this emerging technology.

Within the context of computing education, we refer to outcome expectancy as students' self-estimated performance in a learning environment. We believe this construct has the potential to serve as a proxy for a broad range of motivational constructs capable of influencing students' engagement and persistence in a course. While substantial prior research has delved into other motivational factors like students' self-efficacy, there remains a need for further exploration into the nuances of students' intrinsic belief constructs in relation to their learning behavior, engagement, and performance. This paper studies the impact of factors like GPA, self-efficacy, and other identified constructs on students' outcome expectancy. We also study the group-based differences in students' performance based on outcome expectancy. For this purpose, we analyzed the data of four hundred and ten engineering students enrolled in an introductory programming course (CS1). An exploratory factor analysis of surveyed questions was performed, identifying factors linked to students' self-efficacy, attitude toward learning, and perceptions about programming. We found that students' outcome expectancy significantly differed based on each of these identified factors and GPA. Furthermore, we also found that students' performance in the course significantly differed based on their outcome expectancy. We believe that such an analysis will provide CS educators with a better grounding to understand the underlying belief constructs that influence students' participation, persistence, and performance.

Designing interactive virtual learning environments with effective and engaging design elements is crucial in enhancing learners' motivation, engagement, and learning outcomes. By prioritizing a positive user experience that curates cognitive load, interactions within virtual learning environments may effectively promote voluntary and formative engagement. This analysis focuses on investigating the affordances of having an opportunity to immediately reattempt an incorrectly answered question and additionally have access to video solutions on students' voluntary engagement with mastery learning activities. These mastery learning activities were provided in the form of quizzes through a virtual learning environment, YANTRA EDU. This application was developed to facilitate mastery learning, where learners have the opportunity to engage with the sequential practice of various concepts in an introductory programming (CS1) course.