Evaluating and Redesigning Imputation Methodologies for the 2015 American Housing Survey

• Activity Number: 363
• Type: Contributed
• Date/Time: Tuesday, August 6, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Survey Research Methods Section
• Author(s): George Carter and Brian Shaffer*+
• Companies: US Dept. of Housing & Urban Development and U.S. Census Bureau
• Keywords: imputation ; variance ; mean imputation ; hot-deck imputation ; multiple imputation ; longitudinal

Abstract

The American Housing Survey (AHS) is the largest nationally representative survey of housing in the United States. It is conducted every two years by the U.S. Census Bureau for the Department of Housing and Urban Development (HUD) and has followed the same sample of housing units since 1985. The AHS imputes some items for non-response using hot-deck, cold deck, and regression-based methods. The AHS is currently undergoing a survey redesign and will introduce a new sample in 2015. As part of the redesign, the Census Bureau is evaluating current imputation methodologies and designing new approaches to impute items in 2015 and beyond. Utility costs are a financial housing characteristic that is imputed in the AHS. Utility cost data are currently adjusted and imputed using a regression-based method that utilizes utility bill and housing characteristics from the AHS and housing characteristics, consumption, and cost data from the 2001 Residential Energy Consumption Survey (RECS) adjusted for inflation. This research re-estimates utility models with data from the 2009 RECS and explores the implications of estimating utility costs without AHS utility billing data.