The Validity of the Comparative Interrupted Time Series Design for Evaluating the Effect of School-Level Interventions