Copyright by Michael John Lustina 2004 - The University of Texas at - repositories lib utexas-2026

The dissertation by Michael John Lustina compares Andrich’s Rating Scale Model (ARSM) and Rost’s Successive Intervals Model (SIM) in the context of Computer Adaptive Testing (CAT). It evaluates both models using two data sets: the Audit of Administrator Communication (ADCOM) and a simulated data set. The study finds that while ARSM provides better estimates of theta, SIM requires fewer items to achieve similar results. The research suggests that the choice between models may depend on the specific research context, with ARSM being preferable for high-stakes situations requiring accuracy, and SIM being more efficient for low-stakes scenarios where fewer items are desired.