Research seminar: Causal mediation analysis with double machine learning
- Martin Huber, University of Freibourg
- 13 October 2020
- 11:45 - 12:45
- Follow the seminar at TEAMS: https://teams.microsoft.com/dl/launcher/launcher.html?url=%2F_%23%2Fl%2Fmeetup-join%2F19%3Ameeting_M2Q2OWVlMjEtMjlkZC00MDdjLWE5YmUtZDA3OTVjMzQzZWIy%40thread.v2%2F0%3Fcontext%3D%257b%2522Tid%2522%253a%2522c7217092-b240-4e1d-bd61-fa97ba975cbc%2522%252c%2522Oid%2522%253a%2522448837b1-9373-4be1-adb1-27de4e3addab%2522%257d%26anon%3Dtrue&type=meetup-join&deeplinkId=4e93f57e-918f-498a-9855-4211a4b791b1&directDl=true&msLaunch=true&enableMobilePage=true&suppressPrompt=true
Martin Huber, University of Freibourg: https://www3.unifr.ch/appecon/en/chair/team/prof/
Causal mediation analysis with double machine learning
Abstract: This paper combines causal mediation analysis with double machine learning for a datadriven control of observed confounders in a high-dimensional setting. The average indirect eect of a binary treatment and the unmediated direct eect are estimated based on ecient score functions, which are robust w.r.t. misspecications of the outcome, mediator, and treatment models. This property is key for selecting these models by double machine learning, which is combined with data splitting to prevent overtting. We demonstrate that the eect estimators are asymptotically normal and root-n consistent under specic regularity conditions and provide a simulation study as well as an application to the National Longitudinal Survey of Youth.