New Variance Estimation Method Proposed for Calibrating to Random Control Totals
Westat’s Jean Opsomer, Ph.D., and Andreea Erciulescu, Ph.D., propose a new variance estimation method that directly uses the replicate weights from 2 surveys, one survey for control totals for calibration of the other survey weights, to estimate the overall variance. Their article, Replication variance estimation after sample-based calibration, appears in Survey Methodology.
Their approach modifies the replicates of the survey to be calibrated by using the replicates from the control survey directly. They show how this method can be used even when the replication methods and/or the number of replicates differ between the 2 surveys.
“To provide correct variance estimates, a method needs to account for the variance contribution of the random controls,” notes Dr. Opsomer. “The approach in this article does this and is also very easy to implement.”