In previous studies of educational systems, Markov chains have often been used in simulations in which history played no role because all events were described as independent of each other. Mono-level dynamics (i.e., the social context in which learning occurs is ignored) have also been applied to simulations. Educational theories, however, are usually multilevel in nature, since they describe pupil learning as a function of the characteristics of the group in which learning takes place. Moreover, teacher behavior is affected by organizational interventions. A three-level simulation model for educational effectiveness is presented. This mathematical model adequately describes the functioning of standard setting by teachers, which can be seen as the lever for raising performance levels. Experiments with the model show that it is almost impossible to achieve different, and possibly conflicting, policy goals simultaneously. The model is applied to two major educational policy programs in the Netherlands, the introduction of a Common Core Curriculum and a program addressing inequality of educational opportunities, the Educational Priority Programme. An appendix contains the source code for the model. (Contains 9 tables, 6 figures, and 28 references.) (Author/SLD)