English abstract
This study aims to investigate the impact of sample size on estimation precision of item parameters and person ability for Science subject test using Rasch model. In order to achieve the objective of the study, 12 samples with different sample size of 200, 500, 1000 and 1500 were randomly selected by SPSS from the main sample that is composed of 8484 students from grade 7 in Albatinah North region. The estimation of item difficulty, person ability, and test information function were carried by Multilog 7.0 using Marginal Maximum Likelihood method. To evaluate the estimation precision, Root Mean Square Error (RMSE) was used to compare between the estimated parameters from the main sample and estimated parameters from the other 12 data. The result of the study revealed that the estimation precision of the three parameters increases as sample size increases. In addition, the study indicated that a sample size of 500 is sufficient to precisely estimate item difficulty parameter, person ability and test information function with Rasch model.
