Equating Test Instruments Using Anchor to Map Student Abilities Through the R Program Analysis

Abstract

This study aims at investigating the equivalence of mathematics try out scores of junior high school through anchor. In the study, the 2 Parameters Logistic (2PL) model of Item Response Theory (IRT) analysis was used through the Haebara method. The estimation step of the equating parameter utilized the equateIRT package in the R program. The results show that the relationship between try out questions of package 1 and package 2 cannot be said to be equivalent. After analyzing and calculating the regression equation, it can be concluded that package 2 of Mathematics National Examination (MNE) try out questions has a higher level of difficulty than package 1. Equating package 1 to package 2* through the Haebara method for the 2 PL model will provide equal or fair scores for the test takers who work on package 1 and package 2. The samples of questions were taken from MNE questions distributed in Makassar in 2014 which consisted of 2 question packages. Question package 1 was responded to by 2099 students, while question package 2 was responded to by 2068 students. The equating design used was a common item design. With the equating made, in the future, the test takers' ability can be detected properly if they work on similar tests, even though test packages are relatively different.

Which one is more accurate, BILOG or R program? (a comparison for score test equating)

“…This will prevent test takers from being adversely affected by the easier or harder test packages they are given. Going through equating, even though test packages are relatively varied, the test takers' aptitude can be detected correctly in the future if they work on similar exams [19].…”

“…A newly developed application that can be used for equating tests using the IRT is the R program. There are several studies on equating tests using this program [19], [48]- [50].…”