Designing optimal item pools for computerized adaptive tests with exposure controls

Computerized adaptive testing requires a well-designed item pool containing an appropriate number of items to build an individualized test that matches the examinee's ability level. An optimal item pool can be defined as a pool consisting of appropriate items for each individual test that is capable of reaching the desired level of precision. It also contains well-balanced items that will achieve optimal item usage and lower the cost of item creation. One of the methods to develop an optimal item pool is Reckase's method (2003), which is a Monte Carlo method to determine the properties of an optimal item pool. This study extends the method for designing item pools calibrated with 3PL and applies it to situations where no exposure control, Sympson-Hetter procedure, or a-stratified procedure is imposed to control the item exposure rate. The procedures for designing the item pool and two approaches of simulating test items are presented. The performance of each optimal item pools is evaluated along with the operational item pools.