Weighted Multiple Regression Estimation for Survey Model Sampling
by James R. Knaub, Jr.
Abstract:
Model-based inference has performed well for electric power establishment survey
data at the Energy Information Administration (EIA), using weighted, simple linear regression, as
pioneered by K.R.W. Brewer, R.M. Royall, and others. Cutoff sampling is used because it is not
practical to collect frequently from among the smallest members of these highly skewed
distributions, where nonsampling error and respondent burden can be particularly serious
problems. Further, certain generation and sales for resale data have proved to be relatively
difficult candidates for sampling. A weighted, multiple linear regression model, using a cutoff
sample, where one regressor is the data element of interest as captured in a previous census, and
another regressor is the nameplate capacity of the generating entity, has proved to be extremely
valuable. This has been applied to monthly sampling, where regressor data have come from
previous annual census information. Estimates of totals, with their corresponding estimates of
variance, have been greatly improved by this methodology. Heteroscedasticity with respect to
each regressor is addressed. Key Words:
variance of totals, heteroscedasticity with respect to
each regressor, establishment surveys, cutoff sampling, stability
Author:
James R. Knaub, Jr.,
jknaub@eia.doe.gov
Editor:
Aurelio Tobias,
atobias@imim.es
READING THE ARTICLE: You can read the article in
portable document (.pdf) format (240845 bytes.)
Errata (in .pdf format).
NOTE: The content of this article is the intellectual property of the authors, who retains all rights to future publication.
This page has been accessed 650 times since April 1, 2007.
Return to the 
Home Page.