In current survey practice, the creation of a data collection instrument involves two distinct steps. The first is survey design, in which a researcher defines the questions and flow of a survey. The second is survey implementation, in which a researcher or programmer turns the design into an electronic or paper survey instrument.

This paper presents an alternative approach to survey development. Metadata-driven survey design means that the actions taken to define a survey are the same actions that create the survey instrument. Using a metadata-driven approach to survey design provides several benefits, including:

• Less redundant work

• Better, easier data documentation

• Reusability of key survey components

• Increased data harmonization potential

• Greater research integrity