Mansoor A. F. Kazi is Reader and Director of Centre for Evaluation Studies at the

This paper outlines the new, emerging realist paradigm in evaluation research, and applies it to social work practice in two countries. This paradigm has the potential for a ‘white box ’ evaluation that not only systematically tracks outcomes, but also the mechanisms that produce the outcomes, the contexts in which these mechanisms are triggered, and the content of the programme. First, realist evaluation investigates what outcomes were achieved, and the extent to which each desired outcome was achieved. Next, the main mechanisms and contexts from the sample of service users are aggregated. The main components of the content of programmes are also tracked across the service users. Finally, inferential statistics are used to identify the key mechanisms and content that were responsible for the outcomes and in what contexts. The statistical methods used include Chi-squares, significance tests, and multiple regression analysis. In this way, it is possible to determine what works, for whom and in what contexts. This new paradigm has been used in the evaluation of a number of social care programmes in England and in Finland, and the paper will outline experiences in the replication of realist evaluation methodologies in the two countries