Design effects in the transition to web-based surveys

Abstract

Innovation within survey modes should always be mitigated by concerns about survey quality and in particular sampling, coverage, nonresponse, and measurement error. This is as true today with the development of web surveying as it was in the 1970s when telephone surveying was being developed. This paper focuses on measurement error in web surveys. Although Internet technology provides significant opportunities for innovation in survey design, systematic research has yet to be conducted on how most of the possible innovations might affect measurement error, leaving many survey designers "out in the cold." This paper summarizes recent research to provide an overview of how choosing the web mode affects the asking and answering of questions. It starts with examples of how question formats used in other survey modes perform differently in the web mode. It then provides examples of how the visual design of web surveys can influence answers in unexpected ways and how researchers can strategically use visual design to get respondents to provide their answers in a desired format. Finally, the paper concludes with suggested guidelines for web survey design.