Journal of Official Statistics, Vol.15, No.2, 1999. pp. 305–327
Calibration as a Standard Method for Treatment of Nonresponse
Sixten Lundström and Carl-Erik Särndal
Abstract:This article suggests a simple and unified approach to the use of auxiliary information
for reducing both the sampling error and the nonresponse bias in a survey. We propose a
point estimator based on calibration and a corresponding variance estimator. They are
general in regard to both the sampling design and the form of the auxiliary information.
The calibration procedure generates final weights which are as close as possible to
specified initial weights (the design weights), while respecting known auxiliary
population totals or unbiased estimates of these totals. When population totals are used,
the resulting point estimators are consistent in the sense that the final weights give
perfect estimates when applied to each auxiliary variable. A clear tendency in our
empirical findings is that an increased auxiliary information content will reduce both
variance and nonresponse bias of the point estimator. Despite some residual bias, the
coverage rate of our confidence intervals comes close to a nominal 95%.
Keywords:Nonresponse adjustment; nonresponse bias; nonresponse variance; auxiliary information; variance estimation.
Copyright © Statistics Sweden 1996-2018. Open AccessISSN 0282-423XCreated and Maintained by OKS Group