DIAGNOSIS KESALAHAN SISWA BERBASIS PENSKORAN POLITOMUS MODEL PARTIAL CREDIT PADA MATEMATIKA

Abstract

Penelitian ini bertujuan untuk menemukan informasi diagnostik dari kesalahan jawaban peserta pada Ujian Nasional (UN) Matematika. Informasi diagnostik yang ditemukan meli-puti atribut yang mendasari butir soal, ketidaktuntasan atribut, dan jenis kesalahan yang dilakukan oleh peserta. Penelitian ini merupakan diagnosis post-hoc, yang digambarkan sebagai pende-katan retrofitting. Analisis butir soal dan respons butir pada UN mata pelajaran matematika untuk menemukan informasi diag-nostik pada kategori isi, proses, dan keterampilan siswa SMP di Bantul Yogyakarta 2007/2008. Hasil penelitian (1) atribut yang mendasari butir soal pada matematika ada 47 atribut, meliputi 4 atribut isi, 36 atribut proses, dan 7 atribut keterampilan, (2) ketidaktuntasan atribut isi, proses, dan keterampilan yang ter-tinggi pada geometri dan pengukuran, (3) jenis kesalahan ter-tinggi pada bilangan, aljabar, dan geometri dan pengukuran adalah kesalahan konsep, dan jenis kesalahan tertinggi pada sta-tistika dan peluang adalah kesalahan interpretasi bahasa, dan (4) penemuan informasi diagnostik data UN Matematika dapat dilakukan melalui mekanisme identifikasi atribut, pengembang-an rubrik penskoran politomus, perhitungan ketidaktuntasan atribut, dan diagnosis kesalahan peserta tes. Kata kunci: diagnostik, matematika, partial credit model ______________________________________________________________ A DIAGNOSIS OF STUDENTS’ MISTAKES BASED ON POLYTOMOUS SCORNG PARTIAL CREDIT MODELS IN MATHEMATICS Abstract This study aims to reveal diagnostic information from the participants’ incorrect answers in the National Examination (NE) of Mathematics. The diagnostic information includes the attributes underlying test items, the attribute exhaustiveness, and the types of mistakes that the participants made. This study was a post-hoc diagnostic study, described as the retrofitting approach. The analysis of test items and item responses in the NE of Mathematics aimed to reveal the diagnostic information in the Junior High School students’ content, process, and skill categories. The results are that (1) there are 47 attributes underlying the mathematics test items, consisting of 4 content attributes, 36 process attributes, and 7 skill attributes, (2) the highest inexhaustiveness of the content, process, and skill attributes is in the topics of geometry and measurement, (3) most mistakes in the topics of numbers, algebra, geometry, and measurement are those of concepts, and most mistakes in statistics and probability are those of language interpretation, and (4) the findings of diagnostic information in the data on the NE of Mathematics can be revealed through the mechanism of identifying the attributes, developing a polytomous scoring rubric, finding out the attribute inexhaustiveness, and diagnosing mistakes the test participants. Keywords: diagnosis, mathematics, partial credit model

Keywords

• diagnostik, matematika, partial credit model