<![CDATA[Correctly naming buoyancy does not establish the ability to apply displaced volume. This exploratory study integrated within-person response analysis and item auditing in archived basic science tests from 37 prospective Madrasah Ibtidaiyah teachers at IAIN Sorong. The 2024 dataset comprised 80 multiple-choice items. Analyses used archived-score summaries, conditional analytical keys, Wilson confidence intervals, and six exact McNemar comparisons with Holm adjustment. Thirty-two participants (86.49%) recognized the buoyancy label, whereas four (10.81%) correctly calculated buoyant force with 80% immersion. Twenty-eight recognized the label but failed the partial-immersion item, a paired difference of 75.68 percentage points (Holm-adjusted p < .0001). Fully versus partially submerged items differed by 35.14 percentage points (adjusted p = .001221). These differences concern specific items rather than causal format effects. Another item lacked a scientifically adequate neutral-buoyancy option and was excluded from correctness scoring. The study contributes a transparent way to distinguish response discrepancies warranting follow-up from item defects requiring revision. Findings suggest testing coordination among verbal, spatial, and mathematical representations; mental representations and misconceptions were not directly measured. Teacher education should use audited tasks with explanations and confidence ratings before assigning diagnostic labels.]]>
Copyrights © 2026
