Variance of the generalized regression estimator under measurement error

In this paper a variance approximation is proposed for the generalized regression estimator that accounts for the additional uncertainty if the weighting scheme contains population tables observed with measurement error.

The source of measurement error in population tables are not necessarily the result of sampling errors. In this sense, the proposed method is an extension of the variance of the regression estimator for two-stage sampling. The method is applied to quarterly data from the Labor Force Survey where weighting is done to population tables estimated with a time series model.

Brakel, J. van den, J. Michiels (2025). Variance of the generalized regression estimator under measurement error. Discussion paper, Statistisc Netherlands, The Hague/Heerlen.

Downloads