From Good Strategy, Bad Strategy by Richard Rumelt:
Whereas a social scientist seeks a diagnosis that best predicts outcomes, good strategy tends to be based on the diagnosis promising leverage over outcomes. For instance, we know from research that K-12 student performance is better explained by social class and culture than by expenditures per student or class size, but that knowledge does not lead to many useful policy prescriptions. A very different strategic diagnosis has been provided by my UCLA colleague Bill Ouchi. His book Making Schools Work diagnoses the challenge of school performance as one of organization […] Decentralized schools, he argues, perform better. Now, whether the organization of a school system explained most of the variations in school performance is not actually critical. What is critical, and what makes his diagnosis useful to policymakers, is that organization explains some part of school performance, and that, unlike culture or social class, organization is something that can be addressed with policy.