ISEDJ

Abstract: The identification of at-risk students early in introductory programming courses is critical to their success. Timely intervention requires assessment before substantial code has been written, and good and bad habits are formed. This study asserts that the use of natural language writing prompts can be used as a diagnostic tool, based on the SOLO taxonomy of cognitive development. The relationship between natural language metrics (Flesch Reading Ease, Flesch-Kincaid Grade Level, Gunning FOG Index) and code maintainability (McCabe’s Essential Complexity) by 29 novice student programmers completing a Java assignment was evaluated. Statistical results show significant differences in all three natural language metrics between students who produced maintainable and unmaintainable code, with the strongest predictability demonstrated by the FOG Index (FOG Index: p=.086, CI90= [-3.4, -.20]). These results were interpreted through the SOLO taxonomy, suggesting that students performing at the Multistructural level produce both disconnected writing (high complexity scores) and unstructured code (high essential complexity), while students performing at the Relational level produce coherent structures in both domains. Further, this study provides implementation guidelines for instructors to manage writing prompts, interpret results, and design impactful interventions, resulting in a low-cost, scalable approach for early student assessment.

Download this article: ISEDJ - V24 N3 Page 4.pdf

Recommended Citation: Clark, J.D., Kinnett, S.J., (2026). Writing Prompts to Identify At-Risk Students in Introductory Programming Courses. Information Systems Education Journal 24(3) pp 4-15. https://doi.org/10.62273/RRRD8805