PacisClassPass Mock Corpus: A Synthetic Classroom Engagement Dataset for Reproducibility

Synthetic classroom-engagement corpus supporting the study Agentic AI in Higher Education: Building a Custom Classroom Engagement Platform Through AI-Assisted Software Engineering (Pacis, 2026, under review at Educational Technology & Society). The corpus contains 91 files (1.3 MB uncompressed) generated deterministically with seed = 42, comprising 565 mock users, 78 quiz items, and 3,485 mock quiz responses distributed across baseline, small-cohort, and geofence scenarios. Purpose. This dataset lets reviewers and readers reproduce every numerical claim in §4 and §7 of the paper without access to any live classroom system. It is paired with the PacisClassPass Load Generation and Verification Bundle (separate Zenodo deposit) which consumes the corpus and reproduces the reported latency and throughput figures. Contents. mock-data/R0-baseline/ — baseline scenario (users, quiz items, quiz responses, location pings, selfie events, session manifest). mock-data/R1-small-cohort/ — small-cohort scenario. mock-data/R10-geofence-70/ — geofence variant at 70 m radius; additional geofence scenarios follow the same pattern. Each scenario directory contains manifest.json, users.jsonl, quiz-items.jsonl, quiz-responses.jsonl, location-pings.jsonl, selfie-events.jsonl, and session.json. Regeneration. The corpus is reproducible from source using the companion software deposit: node generate-mock-dataset.mjs --seed 42 Safety and ethics. No human-subjects data were collected. All records are synthetic and generated by seeded pseudorandom functions. The study is IRB-exempt under BYU–Hawaii policy, and the paired software refuses to execute unless the environment variable EVAL_MODE=mock is set. Recommended citation. Pacis, A. (2026). PacisClassPass mock corpus: A synthetic classroom engagement dataset for reproducibility (Version 1.0.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.21317666