Exploring National Examination Quality Using Rasch Measurement Model and Revising Suggestions: A Case Study of the National Examination of Mathematics Subject in the Three-Year Executive National Exams (2019, 2021 & 2023)

This study examines the quality and effectiveness of the National Mathematics Examinations for 12th-grade students in Timor-Leste, focusing on item difficulty, student performance, and teacher observations over three examination periods: 2019, 2021, and 2023. The population consists of 12th-grade students in the Science and Technology program from six selected Secondary General Education schools. Data were collected from 347 students randomly selected from a population of 2,647 who took the exams in 2019, 2021, and 2023. Additionally, 20 mathematics teachers provided insights into the examination process. Using the Rasch Measurement Model and Bloom's Taxonomy, the analysis reveals significant variations in cognitive performance. "Remembering" items were found to be more difficult than expected, while "Understanding" items required slight adjustments. The "Applying" and "Analyzing" items were well-aligned with expected difficulty levels. Regional disparities in student performance were also identified, with students from Dili and Manatuto demonstrating stronger comprehension and lower guessing rates, while Covalima and Palaban-Oecusse faced greater challenges in answering questions accurately. Teacher observations confirmed consistent vigilance during the examination process, though mixed perceptions emerged regarding the transparency of correction mechanisms. The findings highlight the need for targeted educational support in underperforming regions and revisions to the assessment design, particularly to simplify "Remembering" items and refine "Understanding" tasks. This study provides essential insights for improving the quality, fairness, and transparency of national mathematics examinations in Timor-Leste.