Government Policy Experiments and the Ethics of Randomization

Philosophy and Public Affairs 48 (4):319-352 (2020)
  Copy   BIBTEX

Abstract

Governments are increasingly using randomized controlled trials (RCTs) to evaluate policy interventions. RCTs are often understood to provide the highest quality evidence regarding the causal efficacy of an intervention. While randomization plays an essential epistemic role in the context of policy RCTs however, it also plays an important distributive role. By randomly assigning participants to either the intervention or control arm of an RCT, people are subject to different policies and so, often, to different types and levels of benefits. In this paper, I identify one necessary condition as well as a set of sufficient conditions for the permissible use of random assignment by government agencies. I argue first that random assignment is permissible only if it is consistent with governments’ duty to realize morally important outcomes. I argue second that random assignment is permissible in cases where investigators are in a state of genuine equipoise regarding all arms of the experiment and the policy to which people have a claim of justice. Finally, I defend a set of conditions under which random assignment is permissible in cases where one or more arms of a policy RCT are reasonably expected to be either superior or inferior to this policy.

Author's Profile

Douglas MacKay
University of North Carolina, Chapel Hill

Analytics

Added to PP
2020-08-31

Downloads
370 (#60,400)

6 months
135 (#32,438)

Historical graph of downloads since first upload
This graph includes both downloads from PhilArchive and clicks on external links on PhilPapers.
How can I increase my downloads?