WHAT WORKS IN EVALUATION RESEARCH

British criminology has evolved its own distinctive division of labour . Theorists theorize, ethnographers empathize, statisticians soothsay, and so on. Unlike other areas of scholarship, this arrangement allows little time or gain for anyone bent on fundamental reflection on the nature of criminological research. Whilst (thankfully) this has had th e effect of avoiding the unproductive epistemological brawls which cha racterize neighbouring disciplines, it can lead to a directionless res earch strategy founded on a taken-for-granted pragmatism. This paper e xamines the case of 'evaluation research' in the criminal justice area and attempts to inject some strategic thinking into this, the dourest , most matter-of-fact corner of the literature. On offer is a brief (a nd at this length) polemical case, arguing that the quasi-experimental paradigm has resulted in moribund evaluation, being itself a contribu ting factor to the 'nothing works' lament. Rescue is at hand in the fo rm of a 'scientific realist' approach to evaluation.