Improving Income Imputation by Using Partial Income Information and Ecological Variables

• Activity Number: 96
• Type: Topic Contributed
• Date/Time: Monday, August 12, 2002 : 10:30 AM to 12:20 PM
• Sponsor: Section on Survey Research Methods*
• Author(s): Michael Battaglia*+ and David Hoaglin and Ali Mokdad and Meena Khare and David Izrael
• Affiliation(s): Abt Associates, Inc. and Abt Associates, Inc. and National Immunization Program and National Center for Health Statistics and Abt Associates, Inc.
• Keywords: Random-Digit-Dialing ; Income Cascading ; National Immunization Survey

Abstract

In the National Immunization Survey (NIS), the household telephone interview includes questions about family income. These questions ask first for the amount of total family income in the past calendar year. If the respondent does not know or refuses to answer, the interview continues with a sequence of cascading income questions that can place the family income into an income category. In the 2000 NIS, approximately 14% of household interviews resulted in a missing income, but around 39% of these completed part of the cascade questions. This partial information can be used in improving the imputation of family income (which is needed, for example, in the calculation of the ratio of family income to the corresponding poverty level). The imputation methods studied include regression models whose sets of predictor variables include the partial income information from the cascading questions, telephone-exchange-level ecological variables, and whether the answer to the first income question was "Don't know" or a refusal.