Monitoring and evaluating school effectiveness: the case for longitudinal datasets