Impact of Accumulated Error on Item Response Theory Pre-Equating With Mixed Format Tests

摘要

The equating of tests is an essential processes in high-stakes, large-scale testing conducted over multiple forms or administrations. By adjusting for differences in difficulty and placing scores from different administrations of a test on a common scale, equating allows scores from these different forms and administrations to be directly compared to one another (Kolen & Brennan, 2004); as such, the importance of the accuracy of equating is paramount for any assessment program. Due to the increasingly fast turnaround times required of testing companies in the reporting of test scores, many testing programs rely on pre-equating methodologies. Although many testing companies report the use of pre-equating for some of their tests, there is very little research evidence supporting its use. This study seeks to determine the impact of accumulated error from the long-term use of pre-equating in an operational testing program. Results indicate that for some tests, the pre-equating and post-equating produced almost identical results, while in other cases the differences were quite dramatic.