Embedding Logic Checks in an Automated Hot-deck Imputation of Survey Data

• Activity Number: 58
• Type: Contributed
• Date/Time: Sunday, August 8, 2004 : 4:00 PM to 5:50 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Rong Huang*+ and Wei Yen and Jenny Chia
• Companies: University of California, Los Angeles and University of California, Los Angeles and University of California, Los Angeles
• Keywords: skip check ; hot-deck ; imputation

Abstract

This paper introduces a method we developed that incorporates skip pattern checks of survey in an imputation process. Frequently, surveys contain missing data that require imputation. In complex surveys in which a variable may be related to many others due to question skipping, imputation of a variable may not possibly include all related variables as controls. The result is that the intervariable relationship is either evaluated post imputation, most often manually, for the imputed values or is simply ignored. When an imputed value is found to be inconsistent with other variables, it is sometimes manually edited to make it consistent with other variables or is set to missing to be imputed again. The system we developed automates the whole process of imputation and logic checks. The resulting imputed values are then consistent with all known skip patterns. A hierarchical sequential hot-decking method is used for the imputation and the program is coded in SAS.