Data and scripts for: AI and teacher feedback in L2 argumentative writing

This archive contains anonymized data, analysis scripts, research instruments, and static analysis reports for a three-condition quasi-experimental study of L2 argumentative writing. The study compares AI feedback, teacher feedback, and self-revision conditions across revision depth, Toulmin-based argumentation quality, seven CAF measures, L2 proficiency moderation, and dual-outcome mediation.
The package includes anonymized source CSV files, analysis-ready CSV files, Python scripts, analysis reports, coding documentation, task and rubric materials, and an ethics/data-release statement. It excludes original essays, raw feedback dialogues, interview data, identity mapping files, author-identifying submission materials, API tokens, and local runtime artifacts.