Skip to main content
@shmVirus

publications

Prompted Debugging as a Scaffold for Novice Programmers: Evidence from Lab-Scale Deployment

Sabbir Hosen Mamun†, Priya S. Nair, Lena M. Schreiber† Corresponding author

ACM Technical Symposium on Computer Science Education, pp. 214-221, St. Louis, USA · ACM

Conference2026
PDFCodeSlidesCite

Abstract

Debugging support for beginners often alternates between too little guidance and complete solution disclosure. This study evaluates prompted debugging, a scaffold that asks learners to explain symptoms, localise faults, and predict fixes before receiving automated hints. Across 14 lab sections and 386 submissions, prompted debugging reduced repeated syntax-fix cycles by 27% and increased successful self-repair on semantic bugs. Interaction logs suggest the strongest gains came from prompts that asked students to compare expected and observed program state before editing code.

Cite this work

@inproceedings{Mamun2026Prompted,
  title = {Prompted Debugging as a Scaffold for Novice Programmers: Evidence from Lab-Scale Deployment},
  author = {Sabbir Hosen Mamun and Priya S. Nair and Lena M. Schreiber},
  year = {2026},
  booktitle = {ACM Technical Symposium on Computer Science Education},
  pages = {214-221},
  address = {St. Louis, USA},
  publisher = {ACM},
  series = {SIGCSE TS '26},
  abstract = {Debugging support for beginners often alternates between too little guidance and complete solution disclosure. This study evaluates prompted debugging, a scaffold that asks learners to explain symptoms, localise faults, and predict fixes before receiving automated hints. Across 14 lab sections and 386 submissions, prompted debugging reduced repeated syntax-fix cycles by 27% and increased successful self-repair on semantic bugs. Interaction logs suggest the strongest gains came from prompts that asked students to compare expected and observed program state before editing code.},
}