Model-Based Estimation: When Is It Good Enough for Production?

• Activity Number: 361
• Type: Roundtables
• Date/Time: Tuesday, July 31, 2012 : 12:30 PM to 1:50 PM
• Sponsor: Section on Government Statistics
• Author(s): Wendy Barboza*+
• Companies: USDA/National Agricultural Statistics Service
• Keywords: model-based estimation ; research ; production

Abstract

Within many Federal statistical agencies, the primary goal of statisticians working in the research environment is to have their work incorporated into production. In particular, work related to model-based estimation can be a very challenging endeavor. Many questions arise when attempting to convert this type of research into an operational setting. Is there a different methodological approach to consider and investigate? What potential covariates were examined and are there others? What diagnostics were used to evaluate the performance of the model? Was an independent source(s) used to validate the model-based estimates? Are the error levels for the model-based estimates acceptable? How do the model-based estimates and error levels compare to the current methodology? Why is the model better than what is currently in place? In this session, we will focus on model-based estimation and deliberate the criteria for determining when the research methodology is ready for production.