ESTIMASI MAKSIMUM LIKELIHOOD PADA ITEM RESPONSE THEORY (IRT) DAN APLIKASINYA PADA COMPUTERIZED ADAPTIVE TESTS (CAT); MAXIMUM LIKELIHOOD ESTIMATION FOR ITEM RESPONSE THEORY (IRT) WITH APPLICATIONS TO COMPUTERIZED ADAPTIVE TESTS (CAT)

Yazid Shofwan, Dedi Rosadi

2011 | Disertasi | PROGRAM STUDI S2 MATEMATIKA

In this thesis, it wil be discussed how to estimate paramaters on Item Response Theory (IRT). There are three models of logistics on IRT, One Parameter Logistics Model (1-PL), Two Parameter Logistics Model (2-PL), and Three Parameter Logistics Model (3-PL), the differences lay on how many parameter we used. Basic assumption on IRT are Unidimensional, Local Independence, and Invariance Parameter. Estimation medthod which isused on IRT in the thesis is Maximum Likelihood Estimation (MLE). Estimator of item characteristics, difficulty level (b), disciminant (a), pseudo guessing (c), and examinee ability (?) gain by using Numeric Methode, Newton Raphson Iteration. The item characteristics are used for Adaptive Tests and Computerized Adaptive Tests (CAT).

Kata Kunci : Item Response Theory (IRT),estimasi parameter