Exploring Chinese high school students' performance and perceptions of scientific argumentation by understanding it as a three‐component progression of competencies

Argumentation plays a significant role in science as a scientific practice in which knowledge is constructed, evaluated, and modified. Scientific argumentation (SA) is thus a promising activity for students to pursue in order to think and act like scientists and to enhance their understanding of science. Considering the impact that assessments usually have on teaching and their role in understanding students' ability, this study aims to explore the assessment of scientific argumentation competence (SAC) and Chinese high school students' performance and perceptions of SA. This study, therefore, proposes a three‐component framework for understanding SAC, including the competencies to identify, evaluate, and generate the SA elements outlined in Toulmin's argument model. Based on this framework, we hypothesize three levels of SAC and develop a test including 22 items to see whether these proposed SAC elements are valid and to what extent they represent students' levels of SAC. An iterative procedure is adopted to develop and validate the assessment instrument, validity evidence is collected from the analysis of teachers' interviews, students' think‐aloud and follow‐up interviews, and from confirmatory factor analysis and Rasch analysis of students' scores on the test. The results suggest that the measure, which includes three SAC components has acceptable reliability and validity, and the three‐level progression of SAC is aligned with and further expands previous learning progressions of SA. We also find that most students in our sample are at level 1 of SAC and they in general hold positive attitudes toward learning through argumentation, but their willingness in engaging with SA tends to be affected by their perception of the intrinsic and practical value of SA and their personality. Providing them with explanations of SA elements does not improve their test performance.