Zero-Inflated Regression Modeling for Coverage Errors of the Master Address File

• Activity Number: 648
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Government Statistics Section
• Author(s): Derek Young*+
• Companies: U.S. Census Bureau
• Keywords: Coverage ; Master Address File ; Negative Binomial ; Poisson ; Zero-Inflated Count Regression

Abstract

To meet the strategic goals and objectives for the 2020 Census, the Census Bureau must make fundamental changes to the design, implementation, and management of the decennial Census. The changes must build upon the successes and address the challenges of the previous Censuses. Of particular interest is to gauge the ongoing quality of the Census frames. We address this topic by discussing a set of statistical models for the Master Address File (MAF) that will produce estimates of coverage error at levels of geography down to the block level. The distributions of added and deleted housing units in a block are used to characterize the undercoverage and overcoverage, respectively. As will be shown, these distributions are highly right-skewed with a very large proportion of 0-counts. As such, we utilize zero-inflated regression modeling to determine the predicted distribution of adds and deletes. The models are built using action codes from the 2010 Address Canvassing operation as responses with variables from various other datasets used for predictors. A brief discussion will also highlight the future maintenance and updating of this model.