Context-Specific Instruction: A Longitudinal Study on Debugging Skill Acquisition and Retention for Novice Programmers

2025-09-26

AI Summary

This study addresses the acquisition and retention of debugging skills among novice programmers. We propose and empirically validate a context-specific instructional approach that embeds concrete debugging steps directly within authentic programming problems. Over an eight-week longitudinal experiment, we compared four instructional conditions using multimodal evaluation: quantitative performance metrics (debugging accuracy and time-on-task), psychometric self-reports (perceived stress, task difficulty, and satisfaction), and qualitative feedback. Results show that learners in the context-specific guidance condition achieved 80% accuracy in Week 1 and sustained this performance through Week 3, with average debugging times of 13–15 minutes—significantly outperforming the control group. They also reported lower stress and higher satisfaction. By anchoring instruction in realistic problem contexts, this method effectively bridges the gap between abstract pedagogy and authentic debugging practice. It offers a scalable, transferable intervention paradigm for cultivating high-fidelity debugging competencies in introductory programming education.

Technology Category

• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning
• Reasoning under Uncertainty: Probabilistic Programming
• Data Mining & Knowledge Management: Knowledge Acquisition from the Web

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• User Modeling, Personalization and Recommendation: Practical large-scale studies of user experience

Abstract

Bug localization is a critical skill, yet novices often lack systematic approaches. Prior work tested abstract guidelines and general concrete steps; the impact of context-specific instruction is unclear. We ran an eight-week longitudinal study with four conditions: no instruction (G1), abstract guidelines (G2), concrete steps (G3), and our context-specific instruction that pairs concrete bug-localization steps with problem-specific details (G4). Forty-four undergraduates participated; 41 completed all five sessions (S1-S5). Each session included 2-3 debugging tasks to identify the minimal code element containing a seeded logical fault. We measured correctness (binary), time to completion, self-perceived scores (stress, difficulty, satisfaction, and strategy adherence). G4 achieved higher correctness and shorter time to completion: it reached 80% correctness after one session (vs. 20-44% for other groups) and maintained 80% after three weeks, outperforming all groups (p < 0.05); its time to completion stabilized at 13-15 minutes in S1, whereas other groups took 2-3 sessions to stabilize at 22-27 minutes. Qualitative responses showed lower stress and higher satisfaction in G4, with participants internalizing strategies via contextual examples. We conclude that context-specific instruction yields faster skill acquisition and stronger retention than abstract guidelines or context-agnostic steps. Even 1-2 sessions produced significant gains, while extended practice optimized and stabilized performance. Integrating contextual examples with abstract principles may bridge theory-practice gaps in bug-localization education and provide a more equitable path for novices.

Problem

• Teaching systematic debugging skills to novice programmers
• Comparing context-specific instruction with abstract guidelines
• Measuring skill acquisition and retention in debugging

Innovation

• Context-specific instruction pairs concrete steps with problem details
• It achieves faster debugging skill acquisition and stronger retention
• Integrates contextual examples with abstract principles for novices