The Effect of Fitting a Unidimensional IRT Model to Multidimensional Data in Content-balanced Computerized Adaptive Testing

Book excerpt: This study investigates the effect of fitting a unidimensional irt model to multidimensional data in content-balanced computerized adaptive testing (cat). Unconstrained cat with the maximum information item selection method is chosen as the baseline, and the performances of three content balancing procedures, the constrained cat (ccat), the modified multinomial model (mmm), and the modified constrained cat (mccat), are evaluated in terms of measurement precision, item pool utilization and item exposure control. Three simulation factors are considered: (1) multidimensional structure; (2) ability distribution; and (3) difficulty level of content areas. Simulation results show that overall the content balancing methods are similar to or even better than the maximum information method in terms of measurement precision, especially when the content areas have uneven difficulty levels. However, there is no significant difference in item pool usage and item exposure control. Finally, overall the three content balancing methods perform very similarly, but mmm has the most efficient item pool usage. [The dissertation citations contained here are published with the permission of ProQuest llc. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com/en-US/products/dissertations/individuals.shtml.].