Definition and Diagnosis of Problematic Attrition in Randomized Controlled Experiments

53 Pages Posted: 30 Jul 2013

See all articles by Fernando Martel García

Fernando Martel García

Cambridge Social Science Decision Lab Inc.

Multiple version iconThere are 3 versions of this paper

Date Written: May 26, 2013

Abstract

Attrition is the Achilles' Heel of the randomized experiment: It is fairly common, and it can completely unravel the benefits of randomization. Using the structural language of causal diagrams I demonstrate that attrition is problematic for identification of the average treatment effect (ATE) if -- and only if -- it is a common effect of the treatment and the outcome (or a cause of the outcome other than the treatment). I also demonstrate that whether the ATE is identified and estimable for all units in the experiment, or only for those units with observed outcomes, depends on two d-separation conditions. One of these is testable ex-post under standard experimental assumptions. The other is testable ex-ante so long as adequate measurement protocols are adopted. Missing at Random (MAR) assumptions are neither necessary nor sufficient for identification of the ATE.

Keywords: attrition, randomized controlled experiments, field experiments, causal diagrams, directed acyclic graphs, average treatment effect, nonparametric

Suggested Citation

Martel García, Fernando, Definition and Diagnosis of Problematic Attrition in Randomized Controlled Experiments (May 26, 2013). Available at SSRN: https://ssrn.com/abstract=2302735 or http://dx.doi.org/10.2139/ssrn.2302735

Fernando Martel García (Contact Author)

Cambridge Social Science Decision Lab Inc. ( email )

Washington, DC
United States

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
328
Abstract Views
2,241
Rank
72,910
PlumX Metrics