The aim of this paper is to compare different methods for calculating sampling errors in the Spanish Labour Force Survey (EPA). Half sample replication (HSR) is the method currently employed to this end. We compare its results with those obtained with two other more recent techniques, standard delete-one jackknife and Rao-Wu-Yue bootstrap. The paper begins with a brief description of the EPA methodology, and goes on with a theoretical presentation of the above mentioned methods, followed by the coefficient of variation (CV) calculated for the estimates of the most important EPA variables in 2009.
Finally, we present a more detailed study for the autonomous community of Galicia. In this NUTS2 the sample has been enlarged in the third quarter of 2009, and this fact allows us to study the changes in the estimates of the variance, in relation to the change in sample size.
Palabras clave / Key words: Sampling errors, half sample replication, jackknife, bootstrapStudy of variance estimation methods in the Spanish Labour Force Survey (EPA) (Pdf 112 KB)