On the Bias and Variance of Samples of Individuals: A Comparison of the Electoral Registers and Postcode Address File as Sampling Frames

Abstract

General population surveys in Britain tend to use either the electoral registers (ERs) or the postcode address file (PAF) as a sampling frame of addresses, with subsequent selection of one unit (household or individual) per address. The PAF has better coverage, and is clearly the superior frame for sampling households, but the ER has an important advantage when sampling individuals: selection probabilities can be made to vary less than with a PAF sample. This paper provides an evaluation of the relative merits of the two frames for sampling individuals. This is achieved by analysis of an attitude survey which employed a split frame sample design. The conclusion is that, for most variables investigated, the PAF provides estimates which are no less accurate than those from the ER, and that other advantages of the PAF suggest that it is likely to be the preferable frame for many purposes.