|
Activity Number:
|
289
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistical Computing
|
| Abstract - #306822 |
|
Title:
|
On Weighted Least Squares for Missing Data
|
|
Author(s):
|
Sergey Tarima*+
|
|
Companies:
|
Medical College of Wisconsin
|
|
Address:
|
817 Watertown Plank Road, Milwaukee, WI, 53226,
|
|
Keywords:
|
parameter estimation ; weighted least squares ; missing data ; hierarchical structures
|
|
Abstract:
|
A robast form of weighted least squares for missing data is considered. This form uses a hierarchical structure for estimating a multivariate vector. At i^th step of this procedure estimates of parameters for non-missing components of the vector is based on combining information in the subset of observations with the non-missing components with updated estimates of the location parameters from all subsets with even more missing components in an iterative fashion. This approach was applied to several MCAR and MAR simulation examples. One of them is median difference estimation for a bivariate exponential case, which shows robustness of the approach to some deviations from normality in the presence of MAR missing data.
|
