Examining the Predictive Power of Response Propensity Models in Varied Survey Designs

• Activity Number: 361
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Survey Research Methods Section
• Author(s): Chandra Erdman*
• Companies: U.S. Census Bureau
• Keywords: paradata ; adaptive design ; responsive design

Abstract

Over the past decade, there has been increasing interest in responsive and adaptive survey designs. These designs use paradata and, in some cases, survey response data gathered throughout the data-collection period, to determine whether and when to modify survey design features. The reasons for changing survey protocol vary -- some survey managers want to maximize response rates for a fixed level of effort, some want to reduce respondent burden or harassment, and others want to reduce costs while maintaining sample balance. While these goals are varied, many adaptive survey designs require estimation of response propensity models. Furthermore, most of the large, national surveys conducted by the U.S. Census Bureau collect a common form of paradata with the Contact History Instrument (CHI). This paper explores the use of these paradata in the construction of response propensity models for several surveys that vary widely in their content and design features.