Split-questionnaire Designs for National Health Surveys

• Activity Number: 95
• Type: Invited
• Date/Time: Monday, August 9, 2004 : 10:30 AM to 12:20 PM
• Sponsor: General Methodology
• Author(s): Neal Thomas*+ and Myron Katzoff and Trivellore E. Raghunathan and Nathaniel Schenker
• Companies: Pfizer Inc. and National Center for Health Statistics and University of Michigan and National Center for Health Statistics
• Keywords: split questionnaire design ; matrix sampling ; multiple imputation ; NHANES

Abstract

The National Center for Health Statistics (NCHS) conducts nationally representative probability samples that include expensive and time-consuming clinical exams and labs. The amount of time and measurements these surveys require result in high costs and respondent burden. One potential solution to this problem is to split the questionnaire into several parts and administer a carefully selected subset of the questionnaire to each respondent. We develop a method for creating split-questionnaire forms that can be applied in complex settings which include the use of skip patterns. The method creates forms that include items that are good predictors of the excluded items so the subsequent analyses can recover much of the information about the excluded items that would have been collected in a complete survey. The form-design method is applied using pilot data from NHANES II to create forms for a simulation study based on the NHANES III sample. The simulation study demonstrates the feasibility of the approach applied to a real health status survey, and it produced practical advice about the appropriate items to include in split-questionnaire designs in future health surveys.