Unpacking Early Mathematical Cognition: A Joint Modeling of Speed and Accuracy Using Nested IRTree Models

• In Event: AERA Roundtable Session Thursday 7:45 am - Gold 4
  In Roundtable Session: From Data to Understanding: Building Fair and Rigorous Math and Science Assessments

Abstract

Assessing early math skills in preschoolers is crucial, yet research in this area is scarce, and traditional assessments often overlook response time (RT). This study addresses this gap by applying a nested Item Response Tree (IRTree) model to jointly analyze the accuracy and speed of preschoolers’ math assessment responses from the LLAMA project. The model conceptualizes problem-solving as a sequential process by first evaluating accuracy, then speed conditional on the outcome, to yield a diagnostic profile of distinct latent traits. Preliminary findings successfully disentangled mathematical accuracy from a unitary processing speed trait. Both item domain and student home language significantly predicted performance, demonstrating the model’s diagnostic utility.

Authors

• Xiaochen Xu, University of California - Davis

• Tony Albano, University of California - Davis