Elicitation of Utility Functions and Prior Information in the Design of Complex Sample Surveys

• Activity Number: 464
• Type: Invited
• Date/Time: Wednesday, August 7, 2013 : 8:30 AM to 10:20 PM
• Sponsor: Government Statistics Section
• Author(s): John Eltinge*+
• Companies: Bureau of Labor Statistics
• Keywords: Data quality ; Disclosure limitation ; Measurement error ; Nonresponse ; Responsive design ; Small domain estimation

Abstract

This paper considers three classes of elicitation problems in the design and management of complex sample surveys. First, it is important for designers to understand as much as possible about the utility functions of data users. In many practical cases, these utility functions do not necessarily align fully with standard squared error loss, and may depend heavily on contextual information. Second, design decisions often involve a wide range of options for which the survey organization may have only limited empirical information. Elicitation of prior information from field collection personnel or other experts may provide useful guidance in the selection of specific features to evaluate in small initial laboratory studies or larger field tests for, e.g., effects related to collection mode or respondent incentives. Third, some design decisions involve empirical features that are not readily assessed through small- or moderate-scale field tests, but may be compatible with prior elicitation. The main ideas of the paper are motivated and illustrated by examples from disclosure limitation, small domain estimation, survey redesign and architecture for statistical production systems.