Activity Number:
|
212
|
Type:
|
Contributed
|
Date/Time:
|
Tuesday, August 13, 2002 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Biometrics Section*
|
Abstract - #301598 |
Title:
|
Limitations in Instrumental Variables Estimation of Dose-response
|
Author(s):
|
Marshall Joffe*+
|
Affiliation(s):
|
University of Pennsylvania
|
Address:
|
602 Blockley Hall, 423 Guardian Drive, Philadelphia, Pennsylvania, 19004-6021, USA
|
Keywords:
|
causal inference ; instrumental variables ; randomized trials
|
Abstract:
|
Instrumental variables methods have been proposed for use in estimating treatment effects in randomized trials with noncompliance, using the randomization indicator as the instrument. Unfortunately, the coefficients in structural models fit using instrumental variables or G-estimation methods do not necessarily represent the correct population dose-response curve, i.e., the dose-response curve were it possible to make subjects take the stated dose. This may be seen by elaborating the models fit. These points are illustrated with data from the Lipid Research Clinics-Coronary Primary Prevention Trial, a randomized trial of a cholesterol-lowering drug.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2002 program |