The views expressed here are those of the individual authors and not necessarily those of the JSM sponsors, their officers, or their staff.
Online Program Home
Abstract Details
Activity Number:
|
627
|
Type:
|
Contributed
|
Date/Time:
|
Thursday, August 2, 2012 : 8:30 AM to 10:20 AM
|
Sponsor:
|
Section on Nonparametric Statistics
|
Abstract - #305145 |
Title:
|
Weighted Kernel Regressions in Longitudinal Studies with Dropout
|
Author(s):
|
Lu Wang*+ and Xihong Lin and Andrea Rotnitzky
|
Companies:
|
University of Michigan and Harvard University and Di Tella University/Harvard School of Public Health
|
Address:
|
, Ann Arbor, ,
|
Keywords:
|
Asymptotics ;
Dropout at random ;
Efficiency ;
Inverse probability weighting ;
Kernel smoothing ;
Seemingly unrelated
|
Abstract:
|
We consider nonparametric regression for longitudinal data when some subjects drop out of the study at random. Drop out may depend on the history of both outcome and covariates, but is independent of future outcome and covariates. We propose inverse probability weighted (IPW) kernel generalized estimating equations (GEEs) and IPW seemingly unrelated (SUR) kernel estimating equations using either complete cases or available cases. None of these approaches require specification of a parametric model for the error distribution. We show that these IPW kernel estimators are consistent when the probability of dropout is known or estimated correctly. The most efficient IPW kernel GEE estimator is obtained by ignoring the within-subject correlation, while in contrast the most efficient IPW SUR kernel estimator is obtained by accounting for within-subject correlation and is more efficient than the most efficient IPW kernel GEE counterpart. The IPW kernel estimators obtained using all available cases are more efficient than those using complete cases when appropriate covariance matrices are used. We perform simulations to evaluate the finite sample performance of the proposed methods.
|
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 2012 program
|
2012 JSM Online Program Home
For information, contact jsm@amstat.org or phone (888) 231-3473.
If you have questions about the Continuing Education program, please contact the Education Department.