Kualitas tes literasi numerasi siswa SMP ditinjau dari validitas, reliabilitas, tingkat kesukaran, dan daya beda

Abstract

This study evaluated the psychometric quality of a five-item essay test of numeracy literacy in three-dimensional geometry for Grade VII students, focusing on content and empirical validity, reliability, item difficulty, and discrimination. A quantitative descriptive design was used. Participants were 35 students in class VII-2 at SMP Swasta Gajah Mada, selected through cluster random sampling from a population of 102 students across three classes. The test was developed with reference to Minimum Competency Assessment (AKM) indicators and scored using a partial-credit rubric from 0 to 4. Content validity was reviewed by a mathematics-education lecturer and a mathematics teacher. Empirical item validity was examined using Pearson product-moment correlations between item and total scores. Reliability was estimated with Cronbach's alpha; difficulty indices were calculated from mean item scores relative to the maximum score, and discrimination indices compared the 17 highest- and 17 lowest-scoring students. The average expert rating was 3.55, categorized as very valid. All five items exceeded the critical correlation of 0.334, and the reliability coefficient was 0.82, indicating very high internal consistency. Four items had moderate difficulty and one was easy; four had good discrimination and one had adequate discrimination. These results suggest that the test can support classroom assessment in settings similar to the pilot school. However, the evidence is preliminary because the trial involved one class in one school and only five items. Broader use should follow further review and testing with larger, more diverse samples.