Simulation Study to Validate Sample Allocation for the National Compensation Survey

• Activity Number: 247
• Type: Contributed
• Date/Time: Monday, August 5, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Survey Research Methods Section
• Author(s): Hyunshik Lee*+ and Tiandong Li and Klaus Teuter and Chester H. Ponikowski and Gwyn R. Ferguson
• Companies: WESTAT and Westat and Westat and Bureau of Labor Statistics and Bureau of Labor Statistics
• Keywords: Replicated population ; random rounding of the weighting factor ; Yates-Grundy-Sen variance formula ; approximate variance formula for PPS designs

Abstract

The National Compensation Survey (NCS) conducted by the Bureau of Labor Statistics (BLS) is an establishment survey which produces as one of its outputs an estimate of the employer costs for employee compensation (ECEC) for establishments in the United States. The survey has been redesigned by reducing sampling stages from three to two, and a new set of sample allocation goals have been derived. This current study is to validate the new allocation goals in comparison with the current sample allocation goals via simulation. The study uses a simulated population dataset, and calculates the population level relative standard errors of the ECEC estimates under the two sets of sample allocation goals using the variance formula for the two-stage PPS design, where the PPS method is used to select establishments at the first stage and to select occupations within the selected establishments at the second stage. This paper discusses these steps and the comparative merits of the two sets of sample allocation goals, demonstrating that the current allocation needs to be revised.