Nonresponse Followup Modeling and Microsimulation: Examining Cost-Benefit Tradeoffs for 2020

• Activity Number: 142
• Type: Contributed
• Date/Time: Monday, August 5, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Government Statistics Section
• Author(s): Kevin M. Shaw*+ and John Boies
• Companies: U.S. Census Bureau and U.S. Census Bureau
• Keywords: 2020 Census ; Administrative Records ; Nonresponse Followup ; AR Enumeration ; Cost Benefit Analysis ; Modeling, Microsimulation

Abstract

The 2010 Nonresponse Followup (NRFU) operation, at $1.6 billion in execution costs, was the single most expensive operation in the 2010 Census and thus a primary target for cost avoidance research. The use of statistical modeling to predict the likelihood of whether a Housing Unit (HU) in the NRFU operation is vacant or nonexistent, and whether data from Administrative Records (AR) for an individual HU are accurate offers substantial opportunities for examining the cost-benefit tradeoffs of alternative operational designs. This paper describes HU status and AR concordance modeling and the resulting outcomes from microsimulations conducted on the 2010 Census. Following the general strategy the Census Bureau used in the Targeted Address Canvassing (TAC) research, we use data from a range of government AR and census sources as predictor variables in binomial logistic regression models to predict 2010 NRFU HU status outcomes (vacant, deleted, occupied) and 2010 NRFU household size agreement with AR data. The universe for this research is the 47,197,405 HUs identified as the NRFU analytic universe for the 2010 NRFU Operational Assessment. Population count, census Mailout/Mailback outcomes, and address quality indicators at the HU level were all found to be important predictors of NRFU HU status. HU-level cost data from numerous sources were then used to estimate the potential cost avoidance. This paper examines cost-benefit tradeoffs - coverage degradation vs. dollar cost avoidance - based on preliminary models to predict HU status and whether and where to use AR data for NRFU.