In large multipurpose surveys, it is common to select the sample systematically proportional to some measure of size (PPS) which is correlated with an important variable of interest. The problem with systematic samples is that variance estimators are biased. This paper presents a bootstrap variance estimator, which can have less bias than standard methodologies, such as half-sample replication. The results will be demonstrated with a simulation study based on an important National Center for Education Statistics' survey-The Schools and Staffing Survey.
Online Availability
