Longitudinal research in educational settings is notoriously difficult, and when the study is on a large scale becomes even more problematic. The StatSMART project aims to measure changes in both teachers and their students over a period of three years. Using previously validated instruments for students, three overlapping test forms will be used with a rotating design so that each student will only complete each test once. Teacher change will be measured using a profiling instrument at the start, middle and end of the study. Rasch measurement allows for this type of design and should provide good information even with the complexity of students changing classes or schools. The issues associated with complex designs of this nature will be discussed.