Item Exposure Control in Computer-Adaptive Testing: The Use of Freezing to Augment Stratification

Abstract

Computerized adaptive tests are efficient because of their optimal item selection procedures that target maximally informative items at each estimated ability level. However, operational administration of these optimal CATs results ill the administration of a relatively small subset of items with excessive frequency, while another portion of the item pool is almost unused. This situation both wastes a portion of the available items and is a security risk for testing programs that are available all more than a few scheduled test dates throughout the year. A number of exposure control methods have been developed to reduce this effect. In this study, we investigate the effectiveness of item “freezing” as a means of augmenting tile Stratified-a method for exposure control. A second variation of tile Stratified-a method investigated here concerns use of differing numbers of strata. Using Monte Carlo procedures, we examine these methods under varying conditions of freezing and number of strata, Results are reported in terms of pool usage and test precision, both unconditionally and conditionally on ability.