|
Activity Number:
|
23
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, August 8, 2004 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #302195 |
|
Title:
|
Trimming Extreme Weights in Household Surveys
|
|
Author(s):
|
Benmei Liu*+ and David L. Ferraro and J. Michael Brick and Erin Wilson
|
|
Companies:
|
Westat and Westat and Westat and Westat
|
|
Address:
|
, Rockville, MD, 20850,
|
|
Keywords:
|
outliers ; sampling weights ; mean square error
|
|
Abstract:
|
Outliers in household surveys usually arise when there are extreme values for the observed characteristics or when the sampling weights for a few of the cases are very large relative to the other cases. Outliers can seriously inflate the variance of the survey estimates, and may be highly influential for the estimates for small subgroups. The literature describes a few weight-trimming procedures that have been developed to handle outliers due to variable weights. However, most of these procedures were developed for use in surveys in which the variability in the weights is caused by weighting adjustments, or else the methods are very survey specific. In many household surveys, these methods may not apply because extreme weights are often a result of differential probabilities of selection rather than weight adjustments. This paper will explore various methods to trim extreme weights where such weights naturally occur due to differential probabilities of selection. We will evaluate the advantages and disadvantages of various trimming techniques when applied to household surveys of this type.
|
- 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 2004 program |
