Pengembangan Model Waktu Respon Untuk Memahami Proses Kognitif Peserta Tes

Abstract

The popular measurement theory is Item Responds Theory (IRT). IRT measures the parameter of item test and examineer based on the examinee responds, true or false answer. In computer based test, beside responds of the examinee, there is another variable that should be notice, that is the time for item responds. Responds time shows how long examinee can administer item test. The information about responds time in test can be the valuable information source for test administrator , for instance when we analyze the speeded test, calibrate test items, detect about the cheating, and design a test. The responds time can also be used to measure the item parameters of a test (item difficulties and item slowness) and examinee parameters of a test (ability and speedness). Relating with those purpose, the responds time be modelled based on the basic distribution. Researcher choose the lognormal distribution as time responds distribution, because of the easy intrepretation. The responds time is dependent variable, and independent variable are item difficulty, item slowness, ability dan speedness. The method used to estimate parameters is Bayesian, Markov Chain Monte Carlo.