4.24 Assumptions: SUTVA

  • SUTVA: Stable Unit Treatment Values Assumption
  • -1- No interference & -2- No hidden variations of treatment
      1. The potential outcomes for any unit do not vary with the treatments assigned to other units
      • A subject’s potential outcome is not affected by other subjects’ exposure to the treatment
      1. For each unit, there are no different forms or versions of each treatment level, which lead to different potential outcomes (Consistency) (Imbens and Rubin 2015, 10; Keele 2015b, 5)
      • Unambiguous definition of exposure, e.g. “15 min of exercise” (Keele 2015b, 5)
      • Controversy around this assumption (Keele 2015b, 5)

  • Q: Imagine you all have a headache (Pain = Yes) and we try to randomly assign aspirin to one half of you to test its effect: What would the two assumptions (No interference, No hidden variations) mean in that situation? Discuss with your neighbour!

References

Imbens, Guido W, and Donald B Rubin. 2015. Causal Inference for Statistics, Social, and Biomedical Sciences: An Introduction. Cambridge University Press.

Keele, Luke. 2015b. “The Statistics of Causal Inference: A View from Political Methodology.” Polit. Anal. 23 (3): 313–35.