Evidence, Argument and Prediction

Abstract

In this paper I propose a theory of evidence – which I call the Argument Theory – for domains where it is appropriate to demand high standards of rigor, explicitness and transparency, as in evidence for scientific conclusions and especially for evidence-based policy, which is where the need for such a theory first became apparent to me. I then apply the Argument Theory to answer a question that is too seldom asked, and never properly answered, in evidence-based policy where randomized controlled trials (RCTs) are taken as the ‘gold standard’ for evidence for predicting policy effectiveness: What does it take to makes positive RCT results evidence for policy predictions? The answer it turns out is quite a lot: information is required both about the causal role of the policy in the local circumstances and the helping factors required for it to work there.