Multistage Test Reliability Estimated via Classical Test Theory

Multistage testing is a procedure designed to provide the benefits of adaptive testing without the problems that arise from testing each test taker with a possibly different set of test items. Instead of selecting the items for each test taker individually, it uses preassembled test modules. The testing procedure is divided into stages. In the first stage, there is only one module; all test takers taking that form of the test are tested with the same set of items. In the second stage, there are two or more modules that differ systematically in difficulty. Each test taker is assigned to a second-stage module on the basis of his or her performance on the first stage. If there is a third stage, it is adaptive in the same way as the second stage. A form of the test consists of a set of items for each module, the decision rules for tracking test takers to different modules (usually, a set of cut scores), and a scoring procedure that yields a reported score for each possible set of item responses on each possible combination of test modules.