Work

Optimal Stratification of Item Pools in α-Stratified Computerized Adaptive Testing

Year: 2003

Type: article

Abstract: A method based on 0-1 linear programming (LP) is presented to stratify an item pool optimally for use in α-stratified adaptive testing. Because the 0-1 LP model belongs to the subclass of models with a network flow structure, efficient solutions are possible. The method is applied to a previous item pool from the computerized adaptive testing (CAT) version of the Graduate Record Exams (GRE) Quantitative Test. The results indicate that the new method performs well in practical situations. It improves item exposure control, reduces the mean squared error in the θ estimates, and increases test reliability. Index terms: computerized adaptive testing, α-stratified adaptive testing, item pool stratification, 0-1 linear programming

Source: Applied Psychological Measurement

DOI: 10.1177/0146621603027004002

Authors: Hua-Hua Chang, Wim J. van der Linden

Institutions: The University of Texas at Austin, University of Twente

Language: English

Topic: Psychometric Methodologies and Testing

Subfield: Management Science and Operations Research

Field: Decision Sciences

Domain: Social Sciences

Open Access status: green

Funders: National Science Foundation, Spencer Foundation, Center for Advanced Study in the Behavioral Sciences, Stanford University, Center for Advanced Study, University of Illinois at Urbana-Champaign

Award: Improving Computerized Adaptive Testing In the United States