Linear and Nonlinear Modeling of Item Position Effects

Authors

Chansuk Kang, University of Nebraska-Lincoln

Date of this Version

Summer 5-21-2014

Citation

Kang, C. (2014). Linear and nonlinear modeling of item position effects (Master's thesis). Available from DigitalCommons @ University of Nebraska - Lincoln.

Comments

A THESIS Presented to the Faculty of The Graduate College at the University of Nebraska In Partial Fulfillment of Requirements For the Degree of Master of Arts, Major: Educational Psychology, Under the Supervision of Professor Anthony D. Albano. Lincoln, Nebraska: August, 2014

Copyright (c) 2014 Chansuk Kang

Abstract

Item parameter invariance is one of the properties of item response theory (IRT) that enables computerized adaptive testing (CAT) for test administration. The possible influence of item position on test performance is one of the severe threats to the property of item parameter invariance within IRT. This study examines how different representations of item position, i.e., using categorical, linear, and quadratic terms, can impact how the relationship between item position and item difficulty is expressed. An explanatory IRT model is formulated for estimating item position effects. The model is demonstrated using data from the Program for International Student Assessment (PISA) 2009 reading data for the U.S., wherein the same items appeared in four different positions across item clusters. Methods of choosing the best model to detect item position effects are discussed as well as preliminary item analysis for the estimation of item position effects.

Adviser: Anthony D. Albano