Satisficing in Web Surveys: Implications for Data Quality and Strategies for Reduction.

With the increasing use of the Web in mixed mode surveys, especially those conducted by the Census and other federal statistical agencies, it has become more urgent than ever to develop methods to enhance online measurement quality. This dissertation research (which includes three studies) focuses on respondent satisficing as a source of online measurement errors, and an intervention approach to reduce satisficing behaviors. The first study evaluates speeding (or very fast responding) as an indicator by investigating how it is associated with another wellknown satisficing behavior – non-differentiation in grid questions. The second and third studies aim to extend the scope of previous research on Web survey interventions by investigating how the design of the intervention might affect its success in curtailing respondent satisficing. Specifically, the second study examines whether intervention for different satisficing behaviors could produce different impact on overall response quality. The third study explores whether intervention that feels like from a human (with the manipulation of image and text displayed in the intervention prompt) performs differently compared to the intervention that is obviously automatic computer feedback. The findings lead to better understanding of satisficing behaviors and the mechanism for interventions to reduce these behaviors.