Validating policy indicators

Although validity of an indicator, that is, whether it means what it purports to, is critical to its legitimate use in policy, there are few systematic approaches to validation. The reason lies largely in the traditions of positivist social science which require the separation of fact and theory and allow no place for meaning and valuation. The paper proposes approaches to test the validity of both indicators and the conceptual frameworks on which they are based. The most powerful forms of validation require the application of a priori concepts and models. Validation is partly judgmental and never simply mechanical. The validity of a measure is dependent on the context of its use. Social and political processes are an important part of validation, particularly in determining the appropriateness of the assumptions. If an indicator is to be valid, it should be chosen or designed iteratively with the policies it will help to shape.The author wishes to thank Sir Geoffrey Vickers, Professor Martin Krieger and Dr. Norman Shulman for their thoughtful comments on an earlier version of this paper.