AI in Higher Education: Experimental Insights into Motivational and Cognitive Barriers

How to cite

Rezaei, Sarah; Westbrock, Bastian. 2025. "AI in Higher Education: Experimental Insights into Motivational and Cognitive Barriers." AEA RCT Registry. AEARCTR-0015659. https://doi.org/10.1257/rct.15659-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Sarah Rezaei* , Bastian Westbrock

* Primary investigator

Registry ID
AEARCTR-0015659
DOI
https://doi.org/10.1257/rct.15659-1.0
First registered
Last updated
Status
In development
Trial dates
2025-03-20 – 2030-06-30
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/15659

Abstract

This project explores how generative AI can support personalized learning and whether it affects student motivation and performance gaps. While AI has the potential to boost learning, many students remain skeptical, especially with creative tasks. These attitudes may differ by background or task type.

The study uses Randomized controlled trials in the field. Participants complete tasks with or without AI support. The goal is to measure how AI affects their performance and motivation in future tasks.

This research will also examine barriers to using AI effectively. It aims to determine whether these barriers vary by gender, SES level, and learning subjects. The results will help design educational tools that make AI more helpful and fair for all students.

Intervention

Undergraduate students are randomly assigned different versions of a coursework assignment. Some versions include GenAI-generated answer suggestions below numerical questions. The intervention is the provision of these AI-generated suggestions, enabling a comparison between students who receive this support and those who do not.

Experimental design

This randomized controlled trial involves undergraduate students who are assigned different versions of a coursework assignment. Some versions include AI-generated answer suggestions. All students report their expected mark, and those with GenAI support also rate its helpfulness. Coursework data will be matched with administrative records on student background and prior academic performance to analyze outcomes and heterogeneity.

Design details

Randomization method
Randomization is based on the last digit of each student’s university-assigned student number, which is effectively random. This approach assigns students to one of four coursework versions in a predetermined, non-manipulable manner.
Randomization unit
The unit of randomization is the individual student. Each student is assigned to a coursework version based on the last digit of their student number, which serves as a randomization mechanism at the individual level.
Clusters
Aproximately 1000 individual students (1000 clusters at the individual level).
Observations
Approximately 1000 students (one observation per student).
Arms
Approximately 500 students in the control group (no AI-generated suggestions) and 500 students in the treatment group (with AI-generated suggestions), based on random assignment via the last digit of the student number.
Primary outcomes
The primary outcomes are (1) students’ performance on the individual coursework assignment, measured by the total mark awarded (out of 100), and (2) students’ motivation, measured through self-reported responses on a short post-assignment survey. These outcomes will be used to evaluate the effect of AI-generated answer suggestions on both academic achievement and student motivation.

Cited by

No citation data available for this registration yet.

Keywords