Analisis Butir Soal Penilaian Akhir Semester (PAS) Ganjil Mata Pelajaran Bahasa Indonesia Kelas XI Akuntansi SMK Negeri 2 Padang

Abstract

This study was motivated by the absence of item analysis for the Indonesian Language End-of-Semester Assessment (PAS) for the 2024/2025 academic year, so the quality of the test items in terms of validity, reliability, difficulty level, discrimination power, and distractor effectiveness was unknown. The study aimed to analyze the quality of the PAS items and to identify whether the items were too easy or too difficult and whether they could differentiate between students who had mastered the content and those who had not. A descriptive quantitative method was employed. The data consisted of the test papers and students’ answer sheets from the odd-semester PAS 2024/2025. Students’ responses were tabulated by option choice and analyzed using measurement theory, covering item validity, test reliability, item difficulty, item discrimination, and distractor effectiveness. The results showed that in Package A, 7 items were valid (70%) and 3 were invalid (30%); the reliability was moderate (0.620); difficulty levels were dominated by the medium category (7 items; 70%) and easy category (3 items; 30%); discrimination indices were good to very good for 7 items (70%), but 3 items (30%) were poor; and all distractors were good to very good (100%). In Package B, 8 items were valid (80%) and 2 were invalid (20%); reliability was high (0.843); difficulty levels were dominated by easy items (6 items; 60%) and medium items (4 items; 40%); discrimination indices were good to very good for 7 items (70%), with 3 items (30%) in the fair to poor range; and distractors were good to very good for 9 items (90%), while 1 item (10%) had poor distractors. The study recommends revising or replacing invalid items and items with low discrimination, balancing item difficulty—especially in Package B—and conducting routine item analysis before reusing the test.