UC Berkeley

With the increasing focus on incorporating argumentation in the teaching and learning of science in recent years, how to assess students' scientific argumentation ability becomes an emerging challenge. There exists very little work on the development and validation of new kinds of assessments for scientific argumentation. In response to this challenge, a progression-based assessment was developed to assess scientific argumentation ability in middle school. This dissertation examined validity evidence for the scientific argumentation assessment with three separate yet related studies. The first study applied factor analysis and Rasch modeling to investigate the dimensionality and psychometric properties of the argumentation items. The second study applied latent class modeling to examine whether the hypothesized structure of the scientific argumentation learning progression, which is the backbone of the studied assessment, could be observed in the empirical response patterns. The third study applied a multidimensional item response modeling framework to investigate the differences, or similarities, in the function of the studied argumentation items and traditional content items. Collectively, these three studies contribute to the understanding of the performance of the assessment of interest, and inform future research on assessments of argumentation within the domain of science.