Design and estimation issues for rotating business surveys
Design and estimation issues for rotating business surveys
Considerable theoretical developments have occurred in the past forty years for the design and estimation of rotating surveys. A number of issues, relating particularly to rotating business surveys, have received little attention. It is conjectured that this has hindered the application of such techniques to these surveys in spite of their potential for greatly increased sampling efficiency. We address three of these issues - the use of auxiliary information in rotating surveys; design for rotating surveys with changing frames; and the additive consistency of optimal composite estimates of level and change. We derive composite regression estimators which use auxiliary information as well as the information from past responses of survey units. We show that these estimators combine the desirable properties of both composite and regression estimators. Composite ratio estimators are also developed. We consider also the use of auxiliary information in stratification in rotating surveys. We consider design for rotating surveys with changing frames, deriving optimal sampling rates for births and rotation rates for other units. We also consider designs in which the sampling rate of births is optimised given a fixed rotation rate for other units. In both cases these designs may be significantly more efficient than the usual designs. The implication for estimation and for the complex selection procedures which are usually employed for rotating business surveys are clarified. Unless past estimates are revised optimal composite estimates of level and change will not be additively consistent yet we show that revision may be undesirable. We derive optimal unrevised composite estimates of change which may be considerably more efficient than the standard unrevised composite estimates. We also consider alternative designs for improving the efficiency of these estimators. (DX84117)
University of Southampton
1988
Hughes, Phillip John
(1988)
Design and estimation issues for rotating business surveys.
University of Southampton, Doctoral Thesis.
Record type:
Thesis
(Doctoral)
Abstract
Considerable theoretical developments have occurred in the past forty years for the design and estimation of rotating surveys. A number of issues, relating particularly to rotating business surveys, have received little attention. It is conjectured that this has hindered the application of such techniques to these surveys in spite of their potential for greatly increased sampling efficiency. We address three of these issues - the use of auxiliary information in rotating surveys; design for rotating surveys with changing frames; and the additive consistency of optimal composite estimates of level and change. We derive composite regression estimators which use auxiliary information as well as the information from past responses of survey units. We show that these estimators combine the desirable properties of both composite and regression estimators. Composite ratio estimators are also developed. We consider also the use of auxiliary information in stratification in rotating surveys. We consider design for rotating surveys with changing frames, deriving optimal sampling rates for births and rotation rates for other units. We also consider designs in which the sampling rate of births is optimised given a fixed rotation rate for other units. In both cases these designs may be significantly more efficient than the usual designs. The implication for estimation and for the complex selection procedures which are usually employed for rotating business surveys are clarified. Unless past estimates are revised optimal composite estimates of level and change will not be additively consistent yet we show that revision may be undesirable. We derive optimal unrevised composite estimates of change which may be considerably more efficient than the standard unrevised composite estimates. We also consider alternative designs for improving the efficiency of these estimators. (DX84117)
This record has no associated files available for download.
More information
Published date: 1988
Identifiers
Local EPrints ID: 460763
URI: http://eprints.soton.ac.uk/id/eprint/460763
PURE UUID: a2fe33d8-9b48-4089-8c3f-484769093246
Catalogue record
Date deposited: 04 Jul 2022 18:29
Last modified: 04 Jul 2022 18:29
Export record
Contributors
Author:
Phillip John Hughes
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics