Abstract:
|
Web paradata is a relatively inexpensive way to get information about respondents behavior in a survey instrument. It can be a powerful tool for monitoring survey responses and improving survey design. However, because web paradata is complex, many existing measures fail to take advantage of the rich interdependent information available in paradata. Measures that are currently used are often limited to things like time on page, time in instrument or information about an individual page. This study builds on a past qualitative study that sought to understand the interdependent patterns of behavior across the survey pathway. This study uses that information about patterns of behavior to build qualitative measures that take advantage of the complexity and interdependency of the data by considering patterns of backing and answer changes across multiple survey pages. These types of quality measures can help us identify patterns of behavior that we can target for survey improvement.
|