Public policy experimentation: Lessons from America

Abstract

The years from the early 1960s to the late 1980s in the United States represent a unique era in the history of public policy experimentation. During these years a number of social programmes were implemented and evaluated using the ‘medical’ model of the randomised controlled trial design. The reasons for the widespread acceptance among policy‐makers of this approach to evaluation included recognition of the chaos and inefficiency in existing welfare services, and the consequent need to adopt a more rational means of solving policy issues; a government mandate specifying that a proportion of social programme budgets be devoted to evaluation; a long‐standing tradition of social experimentation among behavioural scientists; and the conceptual and methodological tools for applying experimental models of evaluation to the social domain. The paper examines nine public policy experiments carried out during this period in the fields of income maintenance, employment‐support and penal policy. These experiments demonstrate that rigorous evaluations of public policy programmes can be carried out using a randomised controlled design. Other lessons to be drawn from the ‘golden era’ of evaluation in the USA are the complex relationship between research findings and policy; the changing nature of the social contexts in which such experiments are done; the need to include process and qualitative data, and to consider research participants as active agents rather than merely as passive subjects; and the importance of involving different social science disciplines in evaluating different approaches to public policy.