Applying Item Response Theory In Validating An Indonesian Language Profeciency Test

The Indonesian language test is a measurement of competence in understanding the discourse of written and oral media. The understanding of examining the dimensions of the test needs to be analyzed to look for factors that can explain the relationship or correlation between the various independent indicators observed. This study investigated the application of Item Response Theory (IRT) in analyzing packages of Indonesian test for foreign speakers. The test consists of 3 competencies that measure the ability in listening, reading, and grammar. The analysis was carried out through two stages of testing, namely (1) testing the validation and estimation of test reliability, and (2) measuring the ability to examine with the Rasch Model to determine the level of difficulty of the item. The stages of testing are testing content validation through the Aiken index, testing the instrument dimensions, confirmatory factor analysis, KR-20 reliability estimation, estimation of alpha reliability from Cronbach, and estimation of reliability through item information function. The difficulty level of the test item is measured by IRT Model package of the freeware R. The results of the study indicate that the Indonesian language test consists of 3 dimensions that correlate to explain latent factors. Based on test validation testing, the Indonesian language test instrument measures the competency according to the test construction. Test reliability is estimated by three methods showing quite high results. Based on the reliability estimation the test shows that the instrument has the consistency of the test results indicated by a coefficient above 0.7.