February 2013
Sun Mon Tue Wed Thu Fri Sat
          1 2
3 4 5 6 7 8 9
10 11 12 13 14 15 16
17 18 19 20 21 22 23
24 25 26 27 28    

Authors' Committee


Matt Blackwell (Gov)


Martin Andersen (HealthPol)
Kevin Bartz (Stats)
Deirdre Bloome (Social Policy)
John Graves (HealthPol)
Rich Nielsen (Gov)
Maya Sen (Gov)
Gary King (Gov)

Weekly Research Workshop Sponsors

Alberto Abadie, Lee Fleming, Adam Glynn, Guido Imbens, Gary King, Arthur Spirling, Jamie Robins, Don Rubin, Chris Winship

Weekly Workshop Schedule

Recent Comments

Recent Entries



SMR Blog
Brad DeLong
Cognitive Daily
Complexity & Social Networks
Developing Intelligence
The Education Wonks
Empirical Legal Studies
Free Exchange
Health Care Economist
Junk Charts
Language Log
Law & Econ Prof Blog
Machine Learning (Theory)
Marginal Revolution
Mixing Memory
Mystery Pollster
New Economist
Political Arithmetik
Political Science Methods
Pure Pedantry
Science & Law Blog
Simon Jackman
Social Science++
Statistical modeling, causal inference, and social science



Powered by
Movable Type 4.24-en

« App Stats: Carpenter on "R&D Abandonment in Regulatory Equilibrium: Evidence from Asset Price Shocks Induced by FDA Decisions" | Main | App Stats: Mozaffarian on "Estimating the Global Impact of Poor Dietary Habits on Chronic Diseases" »

18 February 2013

App Stats: Garcia on "When and Why is Attrition a Problem in Randomized Controlled Experiments and How to Diagnose It"

We hope you can join us this Wednesday, February 20, 2013 for the Applied Statistics Workshop. Fernando Martel Garcia, a Research Fellow at the Harvard School of Public Health, will give a presentation entitled "When and Why is Attrition a Problem in Randomized Controlled Experiments and How to Diagnose It". A light lunch will be served at 12 pm and the talk will begin at 12.15.

"When and Why is Attrition a Problem in Randomized Controlled Experiments and How to Diagnose It"
Fernando Martel Garcia
Harvard School of Public Health
CGIS K354 (1737 Cambridge St.)
Wednesday, February 20th, 2013 12.00 pm


Attrition is the Achilles' Heel of the randomized experiment: it is fairly common, and it can unravel the benefits of randomization. This study considers when and why attrition is a problem, and how it can be diagnosed. The extant literature remains ambiguous because it relies on the language of probability, whereas problematic attrition depends on the underlying causal relations. This ambiguity arises because causation implies correlation but not vice versa. Using the structural causal language of directed acyclic graphs I show attrition is a problem when it is an active collider between the treatment and the outcome, or when the latent outcome is a mediator between the treatment and the attrition. Moreover, whether observed outcomes are representative of all outcomes, or only comparable across experimental arms, depends on two d-separation conditions. One of these is directly testable from the data.

Posted by Konstantin Kashin at February 18, 2013 12:30 AM