摘要：False positive test results that overstate intervention impacts can distort and constrain the capability to learn and adapt in governance, and are therefore best avoided. This article considers the benefits of using the Bayesian techniques used in signal processing and machine learning to identify cases of these false positive test results in public sector evaluations. These approaches are increasingly used in medical diagnosis—a context in which (like public policy) avoiding false positive and false negative test results in the evidence base is very important. The findings from a UK National Audit Office review of evaluation quality are used to illustrate how a Bayesian diagnostic framework for use in public sector evaluations could be developed.
Loading...
