Time series methods for postcensal estimation of household counts / Susana Rubin-Bleuer.: CS11-613/95-4E-PDF

"We are concerned with the estimation of a population characteristic Y which changes over time, using data from a repeated sample survey carried out at regular time intervals. We propose to achieve accuracy by careful modelling of the survey error, acknowleging a bias component in the direct survey estimator, and the use of auxiliary information represented by a 'known' survey error in estimating another characteristic X. We will use state space models and the Kalman Filter for the estimation. We also address the problem of internal consistency between the sum of the published estimates of individual domains and the estimate of the aggregate. We impose calibration conditions which can be expressed in state space form. We present an application to postcensal estimation of the current number of households: the calibration equations induce a 'joint model' which effectively reduce the variability of the estimators. A measure of the model bias is obtained by the relative error in estimation at June 1991 when the corresponding census counts are available"--Abstract.

Permanent link to this Catalogue record:
publications.gc.ca/pub?id=9.837185&sl=0

• MARC XML format
• MARC HTML format

Request alternate formats