Claim Missing Document
Check
Articles

Found 1 Documents
Search
Journal : MaPan : Jurnal Matematika dan Pembelajaran

A COMPARISON OF MANTEL-HAENSZEL AND STANDARDIZATION METHODS: DETECTING DIFFERENTIAL ITEM FUNCTIONING Ahmad Rustam; Dali Santun Naga; Yetti Supriyati
MaPan : Jurnal Matematika dan Pembelajaran Vol 7 No 1 (2019): JUNE
Publisher : Department of Mathematics Education Faculty of Tarbiyah and Teacher Training Universitas Islam Negeri Alauddin Makassar

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (297.226 KB) | DOI: 10.24252/mapan.2019v7n1a2

Abstract

Abstrak:Tujuan penelitian ini untuk meninjau sensitivitas dua metode yaitu metode Mantel-Haenszel (MH) dan metode Standarisasi dalam deteksi perbedaan fungsi butir atau Differential item functioning (DIF). Sensitivitas ditinjau dari banyaknya butir DIF. Data yang digunakan dalam penelitian ini adalah data generasi dengan menggunakan program Wingen3 yang berbentuk respons dikotomi sebanyak 3054. Ukuran sampel (200 and 1000) respons untuk kelompok referensi dan (200 and 1000) respons untuk kelompok fokus. Sampel diambil secara acak sebanyak 35 replikasi. Distribusi kemampuan kedua kelompok adalah distribusi normal dengan rata-rata dan varians yaitu 0 dan 1. Hasil penelitian menunjukkan bahwa metode MH lebih sensitif dari pada metode standarisasi dalam deteksi DIF untuk sampel 400 maupun sampel 2000. Dari hasil penelitian ini ditemukan bahwa ada kemungkinan metode standarisasi lebih unggul ketika menggunakan sampel yang kecil atau jumlah anggota populasi kelompok fokus dan referensi tidak seimbang, dimana kelompok fokus lebih sedikit dibandingkan kelompok referensi.Abstract:The purpose of this study was to review the sensitivity of the two methods, the Mantel-Haenszel (MH and the Standardization methods to detect differences in function items (DIF). Sensitivity was based on the number of DIF grains. The data used in this study were generation data using the Wingen3 program in the form of a response dichotomy of 3054. Sample size was (200 and 1000) responses for the reference group and (200 and 1000) responses for the focus group. Samples were taken randomly as many as 35 replications. The distribution of the ability of the two groups was normal with average and variance, 0 and 1 respectively. The results of the study indicated that MH method were more sensitive than standardization method in DIF detection for samples of 400 and 2000. The finding also assumed there were possibility that standardization method was supreme when using a small sample or the number of population members of the focus group and reference was not balanced, while the focus group was less than the reference group.