ANALYSIS OF THE VALIDITY AND RELIABILITY OF A NEWTON'S LAWS CONCEPT-UNDERSTANDING TEST INSTRUMENT: A CLASSICAL TEST THEORY AND RASCH MODELING APPROACH

Abstract

This study aims to evaluate the psychometric quality of a Newton's Laws concept-understanding test by integrating Classical Test Theory (CTT) with the Rasch Measurement Model. A quantitative descriptive-evaluative design was applied to N = 100 tenth-grade students from two senior high schools (SMA) in Aceh, Indonesia. The instrument comprised 20 single-tier multiple-choice items covering Newton's First, Second, and Third Laws. CTT analysis (Pearson item–total correlation and Cronbach's Alpha) was performed in SPSS, and Rasch analysis (item fit, reliability, and Wright Map) was performed in WINSTEPS. Validity and reliability: 16 of the 20 items were valid under CTT (rxy > 0.195) and Cronbach's Alpha was 0.78 (acceptable). The Rasch analysis supported unidimensionality (explained variance > 40%), with a Person Reliability of 0.75 and an Item Reliability of 0.92. Four items (Q3, Q6, Q14, and Q17) showed misfit (MNSQ > 1.50) and were also invalid under CTT. Interpretation of misconceptions: the response patterns on these four items suggested possible impetus-based reasoning about inertia, neglect of friction and angled-force components, and the assumption that the normal force equals weight in an accelerating elevator; these interpretations are tentative because the test is single-tier. Together, the two approaches provide complementary evidence on instrument quality and on students' conceptual difficulties in mechanics.