Scientific Literacy Assessment Based on Local Wisdom in Testlets Models

Abstract

This study aims to develop a scientific literacy assessment using a local wisdom-based testlets model that fulfills the criteria for a good assessment. Research and development (R&D) with ADDIE research design (Analysis, Design, Development, Implementation, and Evaluation). The instrument validation was carried out by content by two experts and a teacher to determine CVR and CVI, and empirically carried out on a small group of 30 students and a large group of 216 students. Empirical testing to determine the validity and reliability of the items. Test the effectiveness of the scientific literacy assessment using the Rasch model. The results of the analysis of local wisdom that are used as the context of science in the assessment have 10 contexts, including batik, keris, angklung, flute, and other local wisdom. The number of questions developed were 50 questions consisting of 40 questions to measure competence and science content and 10 questions to measure scientific attitudes. The form of questions used varied, namely ordinary MC, Complex MC, MC with reasons, sorting, filling, description, and questionnaire. The questions are arranged in one testlet based on the context of local wisdom. The results of expert validation showed an average of 94% (very suitable) and the acquisition of CVR and CVI of 1 and 1. The results of small-scale trials showed that the scientific literacy assessment developed was all valid with a reliability coefficient of 0.977. The results of large-scale trials show that all items are valid with a reliability coefficient of 0.850. The effectiveness test using Rasch modeling shows an Eigenvalue of 48.7%. Based on the results obtained, the scientific literacy assessment based on local wisdom developed is valid and reliable and effective for measuring the scientific literacy of junior high school students.