Advanced Commercial Survey Methodsn(COMSURV) Volume 3:Demonstration of Optimum Sample Design fornNonresponse Bias Control

摘要

This report documents demonstration of a procedure for controlling nonresponse bias in survey estimates. The procedure requires subsampling and collecting data from nonrespondents, using more intensive and costly procedures than the original survey. The report describes statistical methods for optimizing the overall design, including the initial sample and the follow up subsample. The optimization can be constructed to either minimize error at fixed cost, or minimize cost for fixed precision. Necessary equations are provided. The report documents the results of applying these procedures in a 1985 mail survey of the commercial sector conducted by the Potomac Electric Power Company, serving the District of Columbia and suburban communities in Maryland.