Instability in the school population between school entrance and school leaving is not "just a problem of missing data" but often the visible result of the educational problems in some schools and is, therefore, not merely to be treated as missing data but as indicator for the quality of educational processes. Even the most superior value-added model with corrections for premeasurement, gender, intelligence, age, socioeconomic and ethnic backgrounds of students is only valid for the detection of schools with the highest raw scores and the highest learning gains, but it is a very moderate predictor for the detection of schools with the lowest amount of students that lag behind and a bad predictor for the equity and efficiency of schools. (Contains 3 tables.)