Choosing the Future of Learning: Assessing Preferences for Educational GPTs through a Discrete Choice Experiment

• In Event: AERA Roundtable Session Wednesday 11:45 am - Gold 3
  In Roundtable Session: Immersive and Data-Rich STEM Learning Analytics

Abstract

This study employs a discrete choice experiment to investigate user preferences for GPT-based educational tools among 1,705 respondents. Nine attributes were examined: cost, accuracy, explanation depth, prompt ease, complexity management, human-like interaction, subjective norm, speed, and availability. Using a Bayesian D-efficient design, participants completed 12 choice tasks analyzed with panel mixed logit modeling. Answer accuracy and subjective norm emerged as the most influential factors, doubling the importance of cost or speed. Users showed polarized preferences for explanation depth, favoring either superficial or comprehensive responses over moderate detail. Significant demographic heterogeneity emerged, with females and younger users more sensitive to social influence. Findings extend technology acceptance models and provide actionable insights for educational AI development and policy.

Authors

• Belle Li, Purdue University

• Yicheng Mao, Maastricht University