87 – Developments in Survey Design and Implementation
Evaluation of Alternative Measures of Size for Sampling of Establishments in the National Compensation Survey
Joan Coleman
Bureau of Labor Statistics
Chester H. Ponikowski
Bureau of Labor Statistics
Bradley Rhein
Bureau of Labor Statistics
The National Compensation Survey, conducted by the Bureau of Labor Statistics, is an establishment
survey sampled yearly from a national frame using probability proportionate to establishment
employment size. The national frame is developed from administrative files maintained quarterly by the
States for the Unemployment Insurance program. Each establishment on the frame is assigned a measure
of employment size equal to the employment in the third month of the frame quarter. In 2011,
approximately 15% of the establishments in the frame, which includes seasonal businesses, reported an
employment of zero, an employment that must be adjusted to ensure that the establishment has a chance
of selection. In the past, establishments with zero employment have been assigned an employment equal
to one employee, with some cases resulting in large weights for the occupations selected from these
companies. The large weights lead to an over-representation of these occupations in the sample. This
paper presents alternative measures of size values and discusses the options for best determining the
measure of size for all establishments in the frame.
