GenAI Learning Outcomes Meta-Analysis: 36 Studies and 7,229 Participants

A meta-analysis synthesizing findings from 36 empirical studies on the impact of generative artificial intelligence in higher education. The study aggregates 132 effect sizes from 7,229 participants, analyzing learning outcomes using the LOTG framework and examining seven moderators. It was authored by Changxin Fan and last updated in May 2026.

Use Cases

• Evaluating the overall effect size of GenAI on student learning based on the reported g = 0.499
• Comparing the impact of GenAI across different learning outcome categories, such as understanding versus dispositions
• Investigating the moderating role of teaching methods, such as collaborative or blended learning, on GenAI's effectiveness

Strengths

• Aggregates data from 36 empirical studies, providing a synthesized view
• Includes analysis of 7,229 participants and 132 effect sizes
• Examines seven study-level moderators to contextualize findings

Limitations

• Dataset is a 40.6 KB document; the underlying raw data or detailed study-level records are not provided
• Column-level documentation is absent; field semantics must be inferred from the document text
• Row count for any potential underlying data tables is unknown

Provenance

Source

figshare

Collection Method

Three-level meta-analysis of published empirical studies

Freshness

Last updated 2026-05-15 05:27:54