With survey costs, but not responses, increasing, new book provides best practices for adaptive designs
RESEARCH TRIANGLE PARK, NC—For survey researchers, collecting data has gotten harder and costlier. Fewer people respond to surveys than ever before, which means more people need to be queried to get a representative sample, and this adds costs and time.
With these concerns in mind, experts have already begun to express concerns with the 2020 U.S. Census costs and accuracy. Adaptive Survey Design, a new book that shares a title with the same survey method adopted by the U.S. Census Bureau, outlines a framework for survey design that will help address these issues for researchers and data collectors. The book provides a framework for customizing data collection procedures to different sample members to help minimize costs and reduce bias.
“Tailored and efficient data collection, specifically through adaptive survey design, requires investment in more customizable systems that are better tailored to the individual,” said Andy Peytchev, an RTI International Senior Survey Methodologist and a co-author on the book. “Our book will serve as a best-practice guide in the drive for accurate data and lower costs.”
The objective in adaptive survey design is to tailor survey processes and sampling protocols to improve survey outcomes. Real-life examples in the book highlight the importance and key features of adaptive survey design. In one example, census data was matched to sample telephone numbers to tailor the time of day in which to call each telephone number. The design then incorporated information such as outcomes from prior calls in determining when to make the next attempts. In another telephone survey application conducted by co-author Andy Peytchev, the census data were used to identify sample telephone numbers with lower likelihood of responding (such as in areas associated with low participation), assigning them to more experienced interviewers.
- Addresses elements of adaptive survey design
- Uses realistic examples
- Is suitable for survey managers and survey statisticians alike
- Introduces a view on different survey errors, including nonresponse, measurement, and sampling error
- Provides a research agenda for the near future
The book’s other two authors are Barry Schouten, senior methodologist at Statistics Netherlands and professor at Utrecht University and James Wagner, a research associate professor in the University of Michigan’s Program in Survey Methodology and the Joint Program in Survey Methodology at the University of Maryland.
In addition to his work at RTI International, Andy Peytchev was a research assistant professor in the University of Michigan’s Program in Survey Methodology and the Joint Program in Survey Methodology at the University of Maryland.