|
Activity Number:
|
160
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 9, 2004 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #301114 |
|
Title:
|
Analysis of NHANES 1999-2002: One Four-year or Two Two-year Surveys?
|
|
Author(s):
|
Lester R. Curtin*+ and Margaret D. Carroll
|
|
Companies:
|
Centers for Disease Control and Prevention and Centers for Disease Control and Prevention
|
|
Address:
|
3311 Toledo Rd., Hyattsville , MD, 20782,
|
|
Keywords:
|
complex survey ; variance estimation ; sample size ; percentiles ; design-based inference
|
|
Abstract:
|
The current National Health and Nutrition Examination survey is designed to be a nationally representative sample for each year and for combinations of years. Public use data files for the combined years 2001-2002 were released in late spring 2004 while data files for 1999-2000 have been previously released. Estimation and analysis of current NHANES data is complicated by several factors including the change in survey design for 2002 due to the 2000 Census of population becoming available, confidentiality restrictions on two-year data releases, and the need to examine relatively rare events. The CDC/NCHS recommends that the two-year datasets be combined and primarily analyzed as a single (four-year) survey. This paper provides support for that recommendation and compares the two-year versus four-year survey in terms of the demographic detail for reliable estimation for a range of commonly used statistics such as means, proportions and percentile distributions. The paper also examines the impact of small sample size and design constraints on annual variation to establish conditions under which the data may be appropriately analyzed as two two-year surveys.
|
