METHOD OF SCORE EQUALITY AND SAMPLE SIZE

Abstract

This study is aimed to obtain information of different score variance result of equating linear method and equipercentile method for sample size 200, 400, and 800 in Ujian Akhir Sekolah Berstandar Nasional (UASBN). The research is important in considering the test device of UASBN shaped packages of different tests. Scores obtained from different packages can not be directly inferred the existence of differences in ability between them, because the difficulty level of the package used influencing these differrences. To overcome the differences are doing through equating. The method used is an experiment of two variables, equating method and the number of respondents. The experiments are not conducted during the learning process, but conducted after the score and the pattern of the answers obtained through UASBN. The population examinee UASBN SD/MI 2008/2009 for IPA subject matter at East Jakarta. Sampling uses random replacement technique. The hypothesis is tested using similarity variance. The results with α = 0,05 shows: (1) score variance equipercentile method (σ²_ekp200) is not different to score variance linear method (σ²_lin200) for the sample size 200, (2) score variance equipercentile method (σ²_ekp400) is not different to score variance equating linear method (σ²_lin400) for the sample size 400, and (3) score variance equipercentile method (σ²_ekp800) is not different to score variance equating linear method (σ²_lin800) for the sample size 800.