Activity Number:
|
244
|
Type:
|
Topic Contributed
|
Date/Time:
|
Tuesday, August 13, 2002 : 2:00 PM to 3:50 PM
|
Sponsor:
|
Section on Survey Research Methods*
|
Abstract - #301423 |
Title:
|
Numerical Comparison of Small Domain Estimators Computed from Current Employment Statistics Data
|
Author(s):
|
R. Harter*+ and Julie Gershunskaya and John Eltinge and Larry Huff
|
Affiliation(s):
|
National Opinion Research Center and U.S. Bureau of Labor Statistics and U.S. Bureau of Labor Statistics and U.S. Bureau of Labor Statistics
|
Address:
|
55 E. Monroe, Suite 4800, Chicago, Illinois, 60603, USA
|
Keywords:
|
synthetic estimator ; weighted sum estimator ; direct estimator
|
Abstract:
|
In recent years, there has been strong interest in the development of estimators of total employment and one-month change in employment for small domains defined by the intersection of geographical areas (generally metropolitan areas) and industrial classification (generally SIC-based major industrial division or NAICS-based supersectors). Researchers have proposed several possible estimators based on data from the Current Employment Statistics program. The properties (e.g., bias, variance, and mean squared error) of these estimators depend on several factors, including the magnitude of model error, relative to sampling error. This paper presents an empirical comparison of the properties of several small domain estimators that are of potential value in Current Employment Statistics applications. These estimators include: (i) a direct estimator based primarily on CES data from the domain of interest; (ii) a synthetic estimator based on state-level estimates of relative change at the two-digit SIC level; and (iii) a weighted sum of several competing estimators.
|