Measuring scientific understanding: A pedagogical problem and its potential solution?