There are some developments in IRT models for ipsative tests. Among them, the Rasch ipsative model appears the most promising. To facilitate its applications, in this study we developed computerized adaptive testing algorithms under this model, and conducted simulations to evaluate their performance. The simulation results showed that using the Fisher information criterion for item selection could achieve precise latent traits estimates; and the longer the test, the more precise the estimates would be. In addition, a procedure was proposed to control statement exposure within examinees and the results showed that the procedure could control the statement exposure successfully and yield comparably accurate latent trait estimates.
|Publication status||Published - Apr 2014|