Comparing Estimates and Variances for a Dataset with Missing Values and Multiple Hot-deck Imputations

• Activity Number: 28
• Type: Contributed
• Date/Time: Sunday, August 8, 2004 : 2:00 PM to 3:50 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Nuria Diaz-Tena*+ and Frank Potter
• Companies: Mathematica Policy Research, Inc. and Mathematica Policy Research, Inc.
• Keywords: item nonresponse ; imputations ; missing values ; multiple imputation ; hot-deck imputation

Abstract

The National Survey of SSI Children and Families (NSCF) collected data about families with children with disabilities who have applied for Supplemental Security Income (SSI) since 1992. The NSCF supports descriptive analysis of the characteristics of current beneficiaries and their families, and studies the effects of welfare reform. The national sample consisted of 8,535 eligible completed interviews, and some interviews had item nonresponse. We imputed data for missing values for some variables. We present an example of how to impute the missing values (imputing the values of the households' cars and associated debts). The household may have one or two cars for the parents, besides a car for the SSI applicant. These data have been imputed by hot-deck imputation procedures taking into account the income, the kind of employment, and all the household reported data. We imputed multiple times using the hot-deck procedure to check if we can detect a between variance among the different imputations. We compared the difference in the estimates between the multiple imputations, as well as the variances of the estimates with missing values, and the multiple complete datasets.