PERGAMON Studies in Educational Evaluation 26 (2000) 127-142 Studies in Educational Evaluation www.elsevier.nl/stueduc THE MEASUREMENT OF SCHOOL EFFECTIVENESS 1 Sorel Cahan and J.Gabriel Elbaz The Hebrew University of Jerusalem Introduction If schools were equal in terms of all of the factors affecting achievement other than the quality of schooling, the definition of "school effectiveness" and its measurement would be simple and straightforward: Between-school differences in the quality of teaching would be the only possible source of between-school differences in achievement. Therefore, school effectiveness would be isomorphic with school achievement, and between-school differences in mean achievement would be valid measures of their differential effective- ness. Unfortunately, however, reality is uncooperative in this respect. Allocation of students to schools is not random (Goldstein, 1997) and between-school variability obtains with regard to many of the factors that are presumed to affect achievement, such as parental level of education, family income, size, and achievement orientation, as well as the student's personal ability and motivation. In addition, between-school differences in access to other cultural and educational inputs (e.g., enrichment programs) and/or external financing sources (e.g., municipal) are also likely to affect achievement. Between-school inequality in student characteristics and non-school educational inputs preclude the simple identification of school effectiveness with school achievement. Between-school differences in achievement may be due, to an unknown extent, to these "initial" differences between them, rather than to their differential effectiveness. Measurement of the "true" level of school effectiveness thus requires "partialling-out" the effects of these initial outside factors. In other words, the answer to the following counterfactual question has to be found: "What would the level of achievement of the schools be, had the schools been equal from the outset?" Clearly, this question can have no simple or entirely satisfactory answer. Schools are different on many dimensions in 0191-491X/00/$ - see front matter © 2000 Published by Elsevier Science Ltd. All rights reserved. PII: S0191-491X(00)00012-2