Rasch Model Analysis of Item Quality for Measuring Conceptual Understanding of the Human Body System

Abstract

This study aimed to eval_uate the quality of test items in a science learning assessment instrument designed to measure students' conceptual understanding of the human body system using Rasch model analysis. A quantitative descriptive research design was employed, involving 100 junior high school students as participants. These students completed a conceptual understanding test consisting of 35 multiple-choice items. The Rasch model was applied using Winsteps software to assess item fit, analyze the person-item distribution through the Wright Map, and eval_uate the overall validity and reliability of the instrument. The analysis results showed that all test items met the Rasch model criteria, indicating their suitability for assessing students' conceptual understanding. The instrument provided an item reliability score of 0.85, reflecting strong internal consistency, while the item separation index of 2.38 confirmed its capability to distinguish items based on difficulty levels. The Wright Map further illustrated a proper alignment between item difficulty and student ability, ensuring fair measurement across varying levels of proficiency. These findings confirmed that the developed instrument was valid and reliable, making it a valuable tool for both eval_uative and diagnostic purposes in science education