Current performance measurement methods are a-contextual and unable to measure the effectiveness of policies. They usually result in rankings that measure the extent of the policy problem, but they fail to reveal anything substantive about the quality of policies, and they are equally unhelpful in the search for best practices. The goal of this article is to further develop and validate a mixed method of performance measurement - a combination of statistical analysis, residue analysis and case study approach - which considers the policy context, relates the outcomes to the policies, and provides guidance in the systematic search for best practices. In this article, this approach will be tested against one of the problems mentioned in the Millennium Development Goals, e.g. literacy rates.