Harm Jan Boonstra | N.J. Nieuwenbroek An empirical comparison of BRR and linearization variance estimators
Gedrukt boek
The paper carries out a simulation study of two qualitatively different variance estimation methods. The first method is based on an analytical derivation of the design variance, after Taylor linearization, and is referred to as the Taylor method. The second method is balanced repeated replication (BRR) which is an example of a resampling method