Understanding the nature of models and engaging in modeling practice have been emphasized in science education. However, few studies discuss the relationships between students’ views of scientific models and their ability to develop those models. Hence, this study explores the relationship between students’ views of scientific models and their self-generated models, and also whether views of models and modeling practice may be influenced by other factors, such as science learning performance and interest. The participants were 402 ninth-grade students in Taiwan. Data were collected using the Students’ Understanding of Models in Science (SUMS) survey and students’ self-evaluations of their own science learning interests and performance on a Likert-scale. The students’ self-developed models explaining why three different magnetic phenomena occur were also evaluated on a schema of five levels, from lower (observational and fragmented models) to higher (microscopic and coherent models).The results reveal that most students’ models remained only at the level of description of observable magnetic phenomena. A small number of the students were able to visualize unseen mechanisms, but these models were fragmented. However, several students with better science learning performance were able to develop coherent microscopic models to explain the three magnetic phenomena. The analyses indicated that most sub-factors of the SUMS survey were positively correlated with students’ self-developed models, science learning performance and science learning interest. This study provides implications for teaching the nature of models and modeling practice.
All Science Journal Classification (ASJC) codes