Ensuring the Future of Computerized Adaptive Testing
Bernard P. Veldkamp · 2012
Capitalization on chance is a huge problem in computerized adaptive testing (CAT) when Fisher information is used to select the items. Maximizing Fisher information tends to favor items with positive estimation errors in the discrimination parameter and negative estimation errors in the guessing parameter. As a result, information in the resulting tests is overestimated and measurement precision is lower than expected. Since reduction of test length is one of the most important selling points of CAT, this is a serious threat to both the validity and viability of this test administration mode. In this chapter, robust test assembly is presented as an alternative method that accounts for uncertainty in the item parameters during test assembly