Abstract
Mental models are mental representations created on the spot to solve problems, to construct knowledge at the global-structural level and to understand scientific phenomena (eg, Johnson-Laird, 1983; Vosniadou, Skopeliti & Ikospentaki, 2004). Some view mental models as central to the analysis of conceptual change. In other words, for successful science learning to take place, students' mental models are the target for correction and expansion. However, critics of the mental model account suggest that naive mental models are simply methodological artifacts. Despite this disagreement in the value of mental models in terms of science teaching and learning, both sides agree that the most effective means of accessing mental models are open-ended interviews (eg, Vosniadou et al, 2004) and forced-choice questions (eg, Nobes, Martin & Panagiotaki, 2005). The advantage of the former method is that a clearer understanding of students' thinking processes is obtained, whereas the latter method has the advantage of being a more standardized assessment procedure with objective scoring. Other methods like concept-mapping tools (eg, Schaal, Bogner & Girwidz, 2010) and thinking aloud protocols (eg, Gadgil, Nokes & Chi, 2012) are also frequently used in research into mental models. These methods tend to deal with small-scale sampling of participants and involve time-consuming data analysis procedures which become limitations for generalization of outcomes. Moreover, the dynamic nature of mental models makes their diagnosis difficult because examinees can change their thinking or expressed behaviors if their situation changes. This poses great challenges to the diagnosis of mental models. Although there exist technology-based assessment tools (eg, Thelwall, 2000), the focus of these approaches is mainly on the correctness of learners' outcomes instead of the internal consistency and coherence of students' knowledge. Our Web-Based Mental Models Diagnosis (WMMD) system goes beyond the capabilities of the existing technology-based assessment tools to investigate whether learners develop and possess single or multiple mental models while learning science.