Analysis of Model Fit and Test Information on Otis-Lennon School Ability Test Using Item Response Theory

Abstract

The Otis-Lennon School Ability Test (OLSAT) is widely used in different academic institutions for various purposes, such as admission tests, performance predictions, and intelligence tests. However, there has yet to be a study on its model fit, item evaluation, and overall test information. This study examined the OLSAT item and the overall test information used by a state university in the admission process through the lens of item response theory utilizing R Studio. Specifically, it seeks to determine the model fit of the data from the university's standardized entrance test and to examine the test information for the standardized entrance test. Comparing the nested models suggests that the more complex model fits the data better (p = 0.000), which is the 2 Parameter Logistic Model. Some items' p-values suggest local dependence, but it can be tolerated upon examination of some of the actual items. The index of difficulty ranges from -51.33198812 to 11.61952106, while the discrimination index ranges from -0.04759777 to 1.16897332 suggesting that some items have a very low discrimination index and some have a high difficulty index. In conclusion, the test provides more information about the middle portion of the ability scale, which is suitable for admission purposes of the test.