MRC/CSO Social and Public Health Sciences Unit

Many policies and other large-scale interventions that affect population health are not amenable to experimental evaluation via trials in which exposure is manipulated by the researcher in order to identify the health impact of the intervention. There is growing interest among researchers and decision-makers in alternatives, such as natural experimental approaches, in which the variation in exposure generated by implementation of the policy is used to identify health impacts.

Such approaches can be highly efficient - for example, if they use data routinely collected from large populations - but they are also prone to bias resulting from selective exposure to the intervention.

We aim to improve understanding of the range of natural experimental approaches that may be useful for policy evaluation, through a mixture of substantive evaluation studies and methodological work to compare alternative methods and identify good practice.