Evaluation has become a routine exercise in many central government programmes and increasingly is a requirement for voluntary and community groups receiving project funding. The pragmatic monitoring that this requires has often encouraged a concentration on the techniques of audit at the expense of reflection on the significance of evaluation, its relationship to research and the theoretical concepts on which it is often shakily based. This volume has tried to redress some of this imbalance through a wide-ranging discussion of the politics of evaluation grounded in real examples.

The conference from which this volume grew expressed considerable scepticism and pessimism about the politics of evaluation. The audience was concerned at the obvious inadequacy of the rationalist model that assumed ‘knowledge’ acquired from evaluation could be fed neatly into strategic planning and policy implementation. Concluding the day's proceedings, Michael Hill reflected, however, that suspicion of the aims and intentions of agencies involved in neighbourhood renewal, youth justice, social services assessments and health partnerships – to name a few of the major issues discussed in this volume – should not make us apologetic about being evaluators.

Drawing on his own research (Hill and Hupe, 2003), Hill argued that recognising that evaluation and assessment procedures are themselves social constructs operating in a politically charged environment did not prevent us either from attempting a rational analysis of policy nor attempting to learn from experience. To want to ask honest questions about ‘what works’ and to try to ‘speak truth to power’ can still be defended as honourable and important activities. To be fair, as Rowe and Taylor (Chapter Thirteen of this volume) point out, New Labour's use of ‘what works’-type language does suggest a clear intention to learn from evaluation in order to replicate good practice and thus improve public services. Yet to understand ‘what works’, we need to know so much more: why things work, for whom they work and what features of any programme have the most impact. With this in mind we are able to identify some ways forward which will take the politics of evaluation into account while supporting evaluation as a valuable and viable form of research.