TITLE Practical Considerations about Expected A Posteriori Estimation in Adaptive Testing: Adaptive A Priori, Adaptive Correction for Bias, and Adaptive Integration Interval. PUB DATE 2002-04-07

In a computerized adaptive test (CAT), it would be desirable to obtain an acceptable precision of the proficiency level estimate using an optimal number of items. Decreasing the number of items is accompanied, however, by a certain degree of bias when the true proficiency level differs significantly from the a priori estimate. G. Raiche (2000) has suggested that it is possible to reduce the bias, and even the standard error of the estimate, by applying to each provisional estimation one of a combination of these strategies: (1) the adaptive correction for bias proposed by R. Bock and R. Mislevy (1982); (2) adaptive a priori estimate; and (3) adaptive integration interval. A simulation study was conducted to explore these approaches. One thousand administrations of a CAT were simulated for varying proficiency levels. Expected a priori estimation of a unidimensional Rasch model with 40 quadrature points was used for provisional and final estimates. Simulation results demonstrate that bias can be reduced more effectively b