Replication Variance Estimation for Balanced Sampling: An Application to the PIAAC Study

• Activity Number: 405
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Survey Research Methods Section
• Author(s): Jianzhu Li*+ and Sixia Chen and Thomas Krenzke and Leyla Mohadjer
• Companies: Westat and Westat and Westat and Westat
• Keywords: Spectral decomposition ; jackknife variance estimation ; balancing conditions

Abstract

This paper describes the application of the replication variance estimation method to the sample data drawn from a balanced sampling design in the Programme for the International Assessment of Adult Competencies (PIAAC) Cycle 1 Round 1 study. One of the participating countries, France, used the cube's method to select a balanced sample of the first stage sampling units, Interviewer Action Areas (IAAs), within administrative regions. To be consistent with other participating countries, the variance estimation should use a replication method with 80 replicate weights. Most existing methods for constructing replication weights are not valid for the balanced design. Within each stratum we applied a method outlined by Fay (1984) to generate replication weights such that the replication variance estimator produces algebraically equivalent results to the linearization variance estimator proposed by Deville and Tille (2005). Across strata, the replication weights were allocated in a block-diagonal fashion to reduce the correlation.