Rasch Analysis Properties of a Chemistry Test for Form Four Students

摘要

Testing is without a doubt an indispensable tool used to measure how successful students in teaching and learning activities carried out by their teachers. Multiple choice test (MCT) is one of the most widely applied tools for estimating the students' ability in certain subjects. In the process of test assessment, teachers need to assess the items, in terms of difficulty and test the students' ability to appropriately respond to the given items. However, many teachers are superficial in their assessment and only rely on raw test scores as the final measure of their students' ability. Rasch measurement enables the researcher to simultaneously calibrate the students' ability and item difficulty into one common scale. This study aimed to ascertain the Rasch analysis properties (item fit and unidimensionality) of a 50-item Multiple Choice Chemistry Test. The test was administered to a group of 70 Form Four students that study chemistry in school. The chemistry test was developed by the researcher with the help of three chemistry teachers with the experience of over 10 years. The data were analysed using Bond & Fox Steps 2. The person reliability was 0.69 and separation was 1.50. The item separation was 3.55 and the reliability was 0.93. The mean for the person (-0.04) and item (0.00) were less than 0.50, indicating that the test was on-target. The infit and outfit mean squares ranged from 0.68 to 1.36, meaning that none of the items was misfits. All the values of PTMEA Corr. were positive, ranging from 0.05 to 0.55. The raw variances shown by the 50 items were less than 40% and the first contrast had the Eigenvalue units of less than five. Thus, the items in the test were unidimensional. In conclusion, the Chemistry Test showed good Rasch analysis properties. The invariance test (Differential Item Functioning) can be carried out to test the item bias against gender or other demographic information.