A self-evaluation procedure for schools using multilevel modelling

Argues that self-evaluation should primarily be based on the outcomes of educational practice. When designing a self- evaluation system one has to cope with 2 problems: how to define a fair school effect measure and how to locate those practices that may lead to malfunctioning. It is shown how pupil monitoring systems can be used to construct a school monitoring system. The statistical aspects of such a system can be handled well by using multilevel statistical models. The proposed approach is illustrated using data on the development of pupils in mathematics achievement. The indicator proposed for self-evaluation purposes is compared with 7 other indicators. Striking differences between indicators are discussed. Moreover, it is discussed how such a monitoring system could be modified to detect educational practices that lead to malfunctioning of pupils. Finally, the practical aspects of implementing the proposed system are depicted.