<![CDATA[Understanding chemistry effectively involves constructing coherent mental models that connect the macroscopic, submicroscopic, and symbolic levels and using these representations in a coordinated manner to explain scientific phenomena. This study constructed and validated a triplet-representation-based instrument for assessing students’ mental models of chemical bonding. Using an instrument development design, the study involved construct formulation, item development, expert content validation, and a limited trial. The instrument comprised eight open-ended essay items, each with three sub-questions targeting the macroscopic, submicroscopic, and symbolic levels. Data were analyzed using the item content validity index (I-CVI), intraclass correlation coefficient (ICC), Pearson correlation, Cronbach’s alpha, and item difficulty and discrimination analyses. The instrument achieved expert validation with a I-CVI values ranged from 0.80 to 1.00 and very high inter-rater reliability among three independent raters (ICC = 0.997). All eight items demonstrated acceptable empirical validity, with item-total correlation coefficients ranging from 0.466 to 0.805, and the instrument showed high reliability (α = 0.844). The items also exhibited appropriate difficulty and discrimination characteristics, confirming their suitability for diagnostic assessment. Students’ responses were predominantly categorized as synthetic mental models, suggesting partially developed but not yet fully integrated understanding across representational levels. Overall, the instrument demonstrated robust validity, reliability, and diagnostic potential for identifying variations in students’ mental models and representational weaknesses in chemical bonding. This study contributes a conceptually grounded and empirically supported instrument for assessing mental models in chemistry education]]>
