Abstract / Synopsis
In the health and social sciences, when it comes to causal inference, randomized controlled trials (RCTs) are considered the gold standard. The justification for granting RCTs this status is the claim that random allocation to different treatment groups addresses the problem of confounding variables by ensuring that any differences in observed outcomes across treatment groups can be attributed only to differences in treatment. I contend that this is a misunderstanding of what random allocation does and that once we see where this argument goes wrong, it becomes clear that the difference, concerning causal inference, between a single observational study versus a single RCT isn't nearly as large as many health and social scientists seem to think.
DOI
10.5642/jhummath.ZBMH7125
Recommended Citation
Michael A. Lewis, "What Does Randomization Do?," Journal of Humanistic Mathematics, Volume 15 Issue 1 (January 2025), pages 264-276. DOI: 10.5642/jhummath.ZBMH7125. Available at: https://scholarship.claremont.edu/jhm/vol15/iss1/15
