Dataset on AI-Generated Feedback and Project-Based Learning in Higher Education

These datasets derive from a quasi-experimental study conducted in a university course (“Assessment and Learning”) within a Primary Teacher Education programme. 

The study compared AI-generated feedback (GPT-o4-mini and DeepSeek R1) with expert human teacher feedback in a project-based learning context.

The dataset includes:

Group-level project performance data (PRE and POST scores, adjusted gains, feedback condition, attendance status).

Students’ perception questionnaire data (perceived mastery, positive emotions, negative emotions, satisfaction).

The study examined feedback effectiveness, comparability between AI and human feedback, non-inferiority and equivalence analyses, and students’ affective and cognitive perceptions of feedback.

All data are fully anonymized and comply with ethical research standards.