Using Rasch Modeling to Validate AI Scoring Models

April 2025

Authors


Steven McGee

The Learning Partnership

Yaurie Hwang

The Learning Partnership

Willow Kelleigh

The Learning Partnership

This study explores the development and validation of AI scoring models for assessments related to the Exploring Computer Science curriculum. Using BERT, AI models were trained to replicate human scorers and validated using Rasch modeling. Results showed that the AI models approximated human scorers, offering opportunities for timely, formative feedback.

Suggested Citation

McGee, S., Hwang, Y., & Kelleigh, W. (2025, April 26). Using Rasch Modeling to Validate AI Scoring Models [Poster]. National Council for Measurement in Education. Denver, CO. https://doi.org/10.51420/Conf.2025.4