Multiple Imputation of Missing Income Data for the Redesigned National Health Interview Survey

• Activity Number: 58 - Leading the Dance with Dirty Data
• Type: Topic Contributed
• Date/Time: Sunday, July 29, 2018 : 4:00 PM to 5:50 PM
• Sponsor: Government Statistics Section
• Author(s): Guangyu Zhang* and Yulei He and Pavlina Rumcheva and Aaron Maitland and Suresh Srinivasan and Alain Moluh and Matthew Bramlett and Chris Moriarity and Tina Norris
• Companies: National Center for Health Statistics and CDC/NCHS and National Center for Health Statistics and National Center for Health Statistics and National Center for Health Statistics and NCHS and NCHS and National Center for Health Statistics and NCHS
• Keywords: Multiple imputation; missing data; National Health Interview Survey

Abstract

The National Health Interview Survey (NHIS) is a cross-sectional survey conducted by the National Center for Health Statistics since 1957. It provides rich information for studying relationships between income and health and health care of the civilian, noninstitutionalized household population. However, the nonresponse rates are high for two key items, total family income and personal earnings. To address the missing data issue, a multiple imputation model was developed (Schenker, et al., 2006) and has been used to impute missing family income and personal earnings. Starting in 2019, the NHIS questionnaire will undergo a major revision. One of the key income variables, person-level earnings, and some health-related items related to family income will no longer be asked during the interview. In this research, we study the impact of the NHIS questionnaire redesign on imputation of family income information. We rebuild a multiple imputation model using the 2016 NHIS data, with common questions contained in both 2016 and 2019 questionnaires. We compare the results of the new imputation model with those of Schenker's model and evaluate the impact of the NHIS redesign.