The measurement of change is one of the challenges in social science research discussed in the past, present and probably in the future. One foundation of measurement of change is the analysis of the instrument's quality and validity does the instrument tap into the same construct over time? Given it does, probably after modifications or accommodating some restrictions, a sound model of change has to be specified and estimated. Numerous approaches using Rasch models or generalized Rasch Models have been presented to model change and growth. Fischer's approach using LLTM and LLRA models or mixture distribution Item response modeling should be named here. In the present paper a modeling framework for change making use of multidimensional Rasch Models is presented (Andersen 1985, Embretson 1991, Wilson 2007). These Models are implemented as special cases of the Multidimensional Random coefficient Multinomial Logit (MRCML) Models (Adams, Wilson, Wang, 1997). With this approach, many flexible ways of setting up a measurement model are offered in the first place. It gives many options to model change or growth, like polynomial growth curve modeling or change according to other approaches. Using the latent regression part of the model, relations of growth parameters and conditions like parental support, classroom variables or students' motivations can be accommodated. Many of these models may be estimated using the software ConQuest which is used for the analyses presented. Other models may require the use of more general software like the nlmixed package of SAS, GLAMM or Mplus. The approach is illustrated by analyzing interest in math data collected from grade 5 to grade 9 in German secondary education schools in the study "Project for the Analysis of Learning Mathematics: PALMA". Polynomial growth curves are fitted to the development of students' interest in mathematics. In contrast to traditional growth curve modeling the measurement model for item responses and the growth modeling part are estimated simultaneously. Growth is regressed onto conditions that are theoretically assumed to have an impact on the interest growth trajectories. Results show a decline in interest in mathematics in this age, some of the theoretically predicted impacts on the interest growth seem to hold.