Cut-off Sampling in Federal Establishment Surveys: An Inter-Agency Review

• Activity Number: 210
• Type: Invited
• Date/Time: Monday, August 1, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Section on Government Statistics
• Author(s): Daniel R. Yorgason*+ and Janice Lent and Benjamin Russell Bridgman and Yan K. Liu and Alan Dorfman and Yang Cheng and Javier Miranda and Scot Rumburg
• Companies: Bureau of Economic Analysis and Energy Information Administration and Bureau of Economic Analysis and Statistics of Income/IRS and Bureau of Labor Statistics and U.S. Census Bureau and U.S. Census Bureau and National Agricultural Statistics Service
• Keywords: Cutoff sampling ; Federal surveys ; Model-based estimation ; Establishment surveys

Abstract

Government policy makers, economic analysts, and the general public rely heavily on data gathered from federal establishment surveys for information on the U.S. economy. Faced with increasing data collection costs and concerns about heavy respondent burden, some statistical agencies -- including the Bureau of Economic Analysis, the Census Bureau, and the Energy Information Administration -- utilize "cutoff sampling," a model-based estimation technique, to maximize the information they extract from their survey data. Cutoff sampling, a highly cost-effective solution for many establishment surveys, involves selecting only the largest units in the population for the sample and using statistical models-often with auxiliary data-to extrapolate the survey information to the smaller units. In this paper, we discuss the use of cutoff sampling in the federal statistical system and present some new results on error measures for survey estimates based on cutoff sample data.