Cognitive Models in Educational Assessment

• Activity Number: 475
• Type: Invited
• Date/Time: Wednesday, August 6, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Social Statistics Section
• Author(s): Thanh Pham*+ and Yi-Hsin Chen Chen
• Companies: University of South Florida and University of South Florida
• Keywords: Cognitive models ; Educational assessments ; Psychometrical diagnostic models ; Mathematics achievement test

Abstract

As the issue of accountability takes center stage in education in recent years, there has been a rise in standardized tests as tools for educational assessment. These tests, which are developed and scored based on traditional psychometric models (e.g., CTT or IRT), give only an aggregated estimate of students' ability. However, achievement tests can provide instructors with information about students' present learning and about the conditions of instruction. By incorporating processes of cognition in modeling procedures (i.e., cognitive models), psychometrically diagnostic models can be applied to extract additional information from item responses to reach this goal. This paper introduces various cognitive models extracted from TIMSS-2007 mathematics achievement test items and applies two psychometrical diagnostic models (i.e., the LLTM and DINA models) along with these cognitive models to students' item responses. Results presented demonstrate that well-defined cognitive models with appropriate psychometrical models can provide detailed and rich portfolio for purposes of item selection, test development, and construct validation as well as learning and instruction diagnosis.