Creating a Resource-Based Cognitive Model for Predicting Mathematical Item Difficulty

• In Event: Poster Session 10
  In Poster Session: Cognition and Assessment SIG Poster Session

Abstract

To produce greater cognitive complexity in mathematical achievement test items, features are often added to the basic mathematical operations, such as contextualization with word problems or extra subgoals. To isolate sources of cognitive complexity, a global cognitive model of processing stages has been developed (Embretson & Daniel, 2008; Daniel & Embretson, 2010) and recently applied to score middle-school standards-based, end-of-the-year assessment exams from a Midwestern state. The current research proposes a new model structured by cognitive resources for working memory to predict item difficulty. Previous literature revealed many potential variables suitable for consideration in the model. Based on exploratory correlations with item difficulty, specific variables were selected for regression analysis for inclusion in the model.

Authors

• Kristin Morrison, Georgia Institute of Technology

• Susan Embretson, Georgia Institute of Technology