Determining Test Length Precision for Economics Testing: The Implementation of IRT Model for Classroom Assessment

Abstract

The Item Response Theory (IRT) has been used extensively in the test analysis to produce the estimation of item parameters and the ability of tester. However, test length and sample size can affect parameter estimates based on the IRT approach. The purpose of this study was to analyze the effect of sample size and test length on the stability of parameters estimation. This study was conducted using two stages of simulation and real testing in economics for senior high school students. The data were analyzed using the WINGEN for data generation and the BILOG for item parameter estimation and ability parameter estimation. The sample size were varied from small to large samples, while the length test were short tests (20 items) and long tests (> 20 items). The results showed that 1) sample size had an effect on the stability of the item parameters, where the increasing of the sample size impact on the raise of item parameter estimation, 2) the test length effect on the stability of the ability parameters, the rise of number of items effect on the increases of ability parameter estimation. Rasch models (1-PL IRT) can be applied by the teacher in conducting classroom assessment with test length 20 and has sufficient reliability.