Test Security and Quality Control for Multistage Tests
Yi-Hsuan Lee, Charles Lewis, and Alina A. von Davier · 2016
As discussed in Yan, Lewis, and von Davier (chapter 1, this volume), a multistage test (MST) is an assessment that differs from a traditional computeradaptive test (CAT) and from a linear test in many respects. It involves modules of items that are preassembled following specific requirements and panels that consist of several stages and at least one module at each stage. It adopts a multistage adaptive design that provides adaptation across test takers at the module level rather than at the item level. In this chapter, items in each multistage module are assumed to be delivered in a fixed ordering as in linear tests. All test takers receive a routing module, which typically contains items from a broad range of difficulty. According to the performance on the routing module, the test takers are routed into one of the second-stage modules with items of different difficulty levels. If the MST has more than two stages, then routing decisions are typically made based on the test takers’ performance on all modules delivered to the point of the decision being made. The routing decision is made on the fly based on predetermined cutscores.