The main purpose of this study was to design and verify the quality of an assessment aimed at diagnosing scientific misconceptions among students enrolled in 10th-grade biology subjects. A sample consisted of N=200 students from schools under the administration of the Office of Secondary Educational Service 31, Nakhon Ratchasima province. We employed the design-based approach, which consists of four phases, namely, construct map, item design, outcome space, and Wright map. Multidimensional Random Coefficient Multinomial Logit Model (MRCMLM) was used to evaluate the quality of the assessment tool. The assessment tool consists of two dimensions, namely, knowledge and reasoning. The assessment tool is comprised of 40 items, with 20 items tapping each of the dimensions. Findings form item analysis and modeling revealed sufficient evidence on the internal structure and validity of the instrument. It can be concluded that the unidimensional model is more appropriate than the multidimensional model for diagnosing scientific misconceptions in the genetics topic for dichotomously scored with four options in each item. The polytomous scored type should be further investigated to determine the best fit model in the items relating to evidence-based reasoning for supporting students' answers.