Exploring Patterns of and Predicting Nonresponse in the Occupational Employment Statistics (OES) Survey

• Activity Number: 245
• Type: Contributed
• Date/Time: Monday, August 10, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Government Statistics Section
• Author(s): Tian Luo* and Amar Mann and Richard J. Holden
• Companies: Bureau of Labor Statistics and Bureau of Labor Statistics and Bureau of Labor Statistics
• Keywords: nonresponse ; collection analytics ; survey data

Abstract

The Occupational Employment Statistics (OES) from the U.S. Bureau of Labor Statistics (BLS) produces employment and wage estimates for over 800 occupations. When a firm is selected to participate in the survey, they are contacted (mail, phone call, personal visit, etc) and asked to provide the wages they pay to their employees. This study explores and identifies any patterns of nonresponse and builds various models for predicting nonresponse at the establishment level. 480,000 individual collection schedules spanning over the recent 18 years are merged with an administrative dataset that contains a census (98 percent) of business establishments in the western U.S. as well as other BLS market information data and Census data. The information and analyses will be provided to survey managers and staff to assist in decision making and to improve the efficiency of data collection, ultimately reducing the time and cost required to collect.