Membandingkan Irt Model 3pl Dengan Irt Model 4pl Untuk Penilaian Menggunakan Computerized Adaptive Test

Abstract

One of the computer-based testing is the Computerized Adaptive Test (CAT), which is a computer-based testing system where the items were given to the participants adapted to test the ability of the participants. Assessment methods are usually applied in CAT is Item Response Theory (IRT). IRT models are most commonly used today is the model 3 Parameter Logistic (3PL), which is about the discrimination, difficulty and guessing. However 3PL IRT models have not provided information more objectively test the ability of participants. The opinion of the test participants were tested items were also to be considered. In this study using CAT in combination with IRT model of 4PL.In this research, the development of CAT which uses about 4 parameters, namely the discrimination, difficulty, guessing and questionnaires. The questions used were about UAS 1 English subjects. Samples were taken from 40 students answer with the best value of the total 172 students spread across 6 classes to measure the parameter estimation problem. Further testing using CAT application 4PL IRT models compared to CAT 3PL IRT models.From research done shows that the CAT application combined with IRT models 4PL can measure the ability of the test taker shorter or faster and also opportunities participants correctly answered the test items was done tend to be better than the 3PL IRT models. Keywords : Ability, CAT, IRT, 3PL, 4PL, Probability, TestÂ

Perbandingan Penerapan Algoritma Teori Respon Butir Dan Fuzzy Tsukamoto Pada Computerized Adaptive Test

“…Sebuah penelitian telah mencoba untuk menambahkan parameter baru untuk metode IRT sehingga menjadi model empat parameter logistic (4PL) yaitu carelessness, lalu membandingkan IRT 3PLdengan IRT model 4P untuk penilaian menggunakan computerized adaptive test [2].Hasil dari penelitian tersebut adalah algoritma model 4PL dapat mengukur kemampuan peserta tes lebih singkat dan cepat, dan peneliti menegaskan peluang peserta tes menjawab benar cenderung naik. Namun pada penelitian ini, tidak dijelaskan apakah hasil pengukuran atau estimasi kemampuan peserta menjadi lebih baik atau tidak.…”