Real-Time Automated Feedback on Reflective Writing in Software Engineering Education (supplementary material)

This dataset accompanies a study of a real-time, automated nudges intervention supporting reflective writing in a software engineering project course at the University of Canterbury, New Zealand. The file anonymised-nudges-and-survey-dataset.csv contains one row per student for the full 2025 (treatment) cohort (80 students, 37 variables), identified only by an anonymised code. Raw reflection texts and free-text survey responses are excluded to reduce re-identification risk. Variables comprise: (1) nudge engagement during the treatment period (counts and percentages of nudges received, opened, dismissed, satisfied, and ignored, plus k-means engagement-cluster membership); (2) human-assessed reflection-quality scores for the pre-treatment period (Sprint 1), treatment period (mean of Sprints 2-3), and each post-treatment sprint (Sprints 4-6), coded Low = 0, Medium = 1, High = 2; and (3) 16 post-treatment survey items on a 7-point Likert scale from -3 to +3, present for the 69 students who completed the survey and blank for the remaining 11. The two "too early" / "too late" items (suffixed _R) are reverse-coded so that positive values are favourable.