Physics learning requires deep scientific thinking skills, especially in complex topics like rotational dynamics. Unfortunately, current evaluation instruments are still dominated by single-tier formats prone to random guessing and mathematical calculation bias. This study aims to develop and validate a Two-Tier Multiple Choice diagnostic instrument free from mathematical interventions and analyzed using a polytomous Rasch model (Partial Credit Model / PCM). This exploratory research employs a Research and Development (R&D) approach based on Djemari Mardapi's model. The developed instrument consists of 11 paired questions (22 items). Content validity involved 7 expert panels, while empirical testing was conducted on 75 high school students (N=75). Data were analyzed using a 4-category polytomous model via Winsteps to evaluate empirical validity, dimensionality, and item fit. The results show that the instrument meets the essential unidimensionality requirement (1st contrast eigenvalue = 1.68). The calibration yielded ideal average infit and outfit MNSQ values (1.00 and 1.04) and a strong item reliability of 0.80. The monotonic progression of the average logit across the four scoring categories (-0.18, -0.06, 0.12, 0.43 logits) empirically validated the efficacy of the scoring structure. In conclusion, the application of polytomous Rasch analysis to two-tier tests accurately diagnoses students' formal reasoning barriers without being distorted by basic mathematical flaws
Copyrights © 2026