Recent Advancements in Statistical Modeling to Identify Address Updating Areas for the 2020 Census

• Activity Number: 648
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Government Statistics Section
• Author(s): Christine Gibson Tomaszewski*+ and John Luthor Boies
• Companies: U.S. Census Bureau and U.S. Census Bureau
• Keywords: address listing ; modeling ; logistic regression ; cost-benefit ; cost reduction ; statistical prediction

Abstract

The 2010 Address Canvassing (AC) Operation was the second most expensive field operation in the 2010 Census at more than 450 million dollars in direct costs. Since the last Census research effort concluded in 2012, substantial gains have been made in both the predictive accuracy and resulting potential cost reduction of the statistical procedures being researched. A range of new independent variables from diverse sources including property tax records, the United States Post Office, and Federal agencies have resulted in substantial improvements in model fit. Better cost estimates of the 2010 AC operations have greatly improved the accuracy and utility of the cost/benefit analyses stemming from the statistical modeling.