Mixed-Methods Study of AI-Mediated Collaborative Lesson Design

A 2026 mixed-methods quasi-experimental study investigates ChatGPT as a peer-like feedback provider in collaborative learning. The research involved 102 teachers in training organized into 21 groups, analyzing their adaptation to AI interaction and performance patterns. It reports a Post-Withdrawal Sustained Performance effect and tensions in peer-to-peer idea-building.

Use Cases

• Analyze qualitative evidence of tensions, including reduced peer-to-peer idea-building and creativity concerns, from group interactions.
• Examine the Post-Withdrawal Sustained Performance (PWSP) effect pattern for technology-related knowledge and artifact quality.
• Study the progression of how student-teachers refined their queries to the model and evaluated its suggestions across the session.
• Investigate the counterbalanced crossover structure for knowledge assessment and sequential two-task design for artifact production.

Strengths

• Study design includes 102 teachers in training (M_age = 38.87, SD = —8.01) organized into 21 groups.
• Employs a mixed-methods approach combining quasi-experimental, quantitative, and qualitative evidence.
• Data is recent, last updated in March 2026.

Limitations

• Dataset is a single 142.1 KB PDF research article, not a structured data table with rows and columns for direct analysis.
• Sample is specific to 102 teachers in training, limiting generalizability to other populations.
• No raw interaction logs, survey responses, or coded data are provided as machine-readable files.

Provenance

Source

Daniele Agostini via figshare.

Collection Method

Mixed-methods quasi-experimental study with a counterbalanced crossover and sequential two-task design.

Time Range

Study conducted within a single session; specific dates not provided.

Freshness

Last updated 2026-03-25.

Geography

null