|
Activity Number:
|
209
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, July 30, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Health Policy Statistics
|
| Abstract - #309067 |
|
Title:
|
A Comparison of Weighting Adjustment and Multiple Imputation Methods To Correct for Nonresponse Bias in a Longitudinal Group-Randomized Clinical Trial for Depression
|
|
Author(s):
|
Lingqi Tang*+ and Naihua Duan and Ruth Klap and Thomas Belin
|
|
Companies:
|
University of California, Los Angeles and University of California, Los Angeles and University of California, Los Angeles and University of California, Los Angeles
|
|
Address:
|
10920 Wilshire Blvd Suite 300, Los Angeles, CA, 90024-6505,
|
|
Keywords:
|
Missing Data ; Hot Deck ; Nornesponse weights ; Multiple Imputation ; Model-Based Imputation ; Logistic regression
|
|
Abstract:
|
Weighting adjustment is a standard method used to correct for unit nonreponse bias. However, in long term follow-up, the association between auxiliary variables and the outcome variables may not be strong. In this paper, we compare two methods of handling unit nonresponse between a baseline assessment and 9-year follow-up in a group-level randomized controlled trial of quality improvement for depression treatment. One approach uses nonresponse weights constructed by fitting logistic regression models to predict follow-up status from baseline clinical and sociodemographic characteristics. The reciprocal of the predicted follow-up probability is used as the nonresponse weight for each participant. A second method uses multiple imputation based on the approximate Bayesian bootstrap. Results from the two approaches are compared.
|
