Randomized controlled trials (RCTs) have become a standard tool of evidence-based policymaking across U.S. government, and in response the corresponding field of policy research ethics has grown substantially in recent years. Local governments run experiments in education, policing, service delivery, and welfare among many other domains. Federalism introduces a challenge for local government policy experiments operating under overlapping jurisdictions. For example, one problem for a recent wave of RCTs involving cash payments was interference between the experimental local cash payment and existing federal cash assistance payments – call this 'policy interference'.

While experimental cash payments are generally expected to benefit recipients, complex and varying benefits rules across federal, state, and local means-tested programs generate risks that the recipients may lose some or all of their existing public benefits due to policy interference. This is a well-known problem among the investigators of cash RCTs, who have implemented a variety of mitigation strategies to minimize associated harms. The investigators in the first contemporary local government-involved cash RCT used the professional ethics of the social work field as the framework for harm mitigation due to policy interference, but the mitigation strategy for policy interference in local government cash RCTs can alternatively be grounded in policy research ethics.

The principles of policy research ethics have been developed with explicit consideration that governments-as-researchers have competing obligations that may not necessarily apply to policy researchers more generally. In this paper I specify the problem of policy interference, catalog approaches to subject selection in the case of recent US cash transfer experiments, consider the variety of solutions that were deployed to minimize exclusions due to policy interference, and extend existing principles of fair subject selection in research ethics to the specific case of local government actors facing trade-offs as they seek to fairly distribute the benefits and risks of research inclusion.

Preliminary reports suggest that among people with low incomes, those who rely on federal disability benefits faced limited opportunity for participation in the initial wave of cash RCTs. While fair inclusion of vulnerable groups in research does not generally turn on the subject selection of any individual experiment, government researchers have especially strong obligations to cooperate across levels of government and over time to ensure that policy research does not perpetuate the systematic exclusion of such marginalized groups, compounding historical exclusions from biomedical research that are still not fully rectified.