Modeling Item-Level Heterogeneous Treatment Effects With the Explanatory Item Response Model: Leveraging Large-Scale Online Assessments to Pinpoint the Impact of Educational Interventions

This CEPR READS Lab brief presents findings from a study that introduced a new statistical method to analyze how the Model of Reading Engagement (MORE) program improves students’ reading comprehension by examining performance at the individual test-item level. Using data from 7,797 third graders, researchers compared traditional total test scores to item-level analyses across passages requiring different degrees of knowledge transfer—near, mid, and far from the taught content. While overall scores showed no significant differences, students in MORE classrooms performed significantly better on mid-transfer items, indicating successful learning transfer from science to social studies contexts. The study demonstrates the importance of looking beyond overall scores to understand how interventions impact specific types of learning.

Read more about the READS Lab at readslab.org/research.