• learning progression;
  • chemistry;
  • Rasch modeling;
  • substance


Previously, a small scale, interview-based, 3-year longitudinal study (ages 11–14) in one school had suggested a learning progression related to the concept of a substance. This article presents the results of a large-scale, cross-sectional study which used Rasch modeling to test the hypothesis of the learning progression. Data were collected from 4,450 students, aged 11–14, across 30 secondary schools in England using a computer-based assessment instrument. The construction of the items was informed by the research literature on students' understanding in chemistry. One hundred seventy-six fixed response items, in three formats, involving the use of video and animation were developed. Scored dichotomously, overall, the data show a good fit to the Rasch model. Item difficulties have a high degree of invariance across ability, schools, gender, and year group. Conceptually, when items are placed in order of difficulty, a coherent progression of ideas emerges which matches the expectations from the longitudinal study. The learning progression is presented. Independent, nationally standardized data allow tentative projection of student performance to the wider population in England. Implications for research and curriculum design are discussed. © 2011 Wiley Periodicals, Inc. J Res Sci Teach 48: 849–877, 2011