|
Activity Number:
|
405
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 15, 2002 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Survey Research Methods*
|
| Abstract - #301370 |
|
Title:
|
Extreme Design Effects: Why They Occur and What to do About Them
|
|
Author(s):
|
Franklin Winters*+ and Margaret Carroll and Sylvia Dohrmann and Lester Curtin
|
|
Affiliation(s):
|
National Center for Health Statistics and National Center for Health Statistics and Westat, Inc. and National Center for Health Statistics
|
|
Address:
|
6525 Belcrest Rd, Room 915, Hyattsville, Maryland, 20782, USA
|
|
Keywords:
|
|
|
Abstract:
|
The design effect (DEFF) of a variable is an indicator of the impact of the sample design upon the sample variance. Many national probability samples have design effects greater than one, due to clustering and differential weighting. The National Health and Nutrition Examination Survey (NHANES) is a rather unique population-based survey that collects interview data, conducts standardized medical examinations, and obtains laboratory measurements on blood sera. For the current NHANES, conducted in 1999-2000, the majority of estimated DEFFs fall between 0.8 and 3.0. However, for some variables, estimated DEFFs are as small as 0.02 and as large as 60. It should also be noted that estimation of sampling errors for NHANES 1999-2000 variables is complicated by the relatively small number of PSU's (26) in the two years of data collection. Extra sources of variation reflected in the standard errors for the complex design may not be reflected in the standard error assuming a simple random sample. This paper will focus on investigating several factors that might contribute to the magnitude of these extreme design effects. Use of the DEFFs in the analysis of NHANES data will also be discussed.
|
- 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 |
