Online Program Home
  My Program

Abstract Details

Activity Number: 306 - Causal Inference
Type: Invited
Date/Time: Tuesday, August 1, 2017 : 10:30 AM to 12:20 PM
Sponsor: IMS
Abstract #322241
Title: Inferring Causal Effects with Invalid Instruments in Mendelian Randomization
Author(s): Hyunseung Kang*
Companies: Economics, Stanford Graduate School of Business, Stanford University
Keywords: Instrumental variables ; Mendelian randomization ; Econometrics ; Weak instruments ; Invalid instruments ; Pleiotropy
Abstract:

Recently, in a subfield of genetic epidemiology known as Mendelian randomization (MR), instrumental variables (IV) techniques have been used to estimate the causal effect of an exposure on an outcome using genetic markers as instruments. These IV techniques require (i) instruments that are strongly associated with the exposure and (ii) a complete knowledge about all the instruments' validity; a valid instrument must not have a direct effect on the outcome and not be related to unmeasured confounders. Often, this is impractical in many MR studies where genetic instruments are weak and complete knowledge about genetic instruments' validity is absent.

The talk discusses recent work that provides honest inference of the causal effect in MR studies in the presence of invalid instruments. We discuss methods based on (i) sequential hard thresholding, which we call two-stage hard thresholding (TSHT), (ii) selective inference, and (iii) union intervals. We also discuss application of the methods to summary-level MR data, which is common in many MR studies.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2017 program

 
 
Copyright © American Statistical Association