The amount of data available in the context of educational measurement has vastly increased in recent years. Such data are often incomplete, involve tests administered at different time points and during the course of many years, and can therefore be quite challenging to model. In addition, intermediate results like grades or report cards being available to pupils, teachers, parents, and policy makers might influence future performance, which adds to the modeling difficulties. We propose the use of simple data filters to obtain a reduced set of relevant data, which allows for simple checks on the relative development of persons, items, or both.