This dissertation is comprised of an introductory chapter that explores the proliferation of mixed-mode surveys, where data is collected using multiple modes, and their influence on social research. Following the introductory chapter, three chapters of journal article length form the main body of the dissertation and address measurement issues facing mixed-mode surveys. Finally, a concluding chapter summarizes the main findings and considers future directions for mixed-mode surveys. The data presented are from methodological experiments embedded in two mixed-mode web and telephone surveys of random samples of Washington State University undergraduate students.
Chapters Two and Three examine how the mode of data collection and the format of response scales influence responses in web and telephone surveys. Respondents surveyed by telephone consistently provide significantly more positive responses than those surveyed by web across a variety of different questions and response scales. In addition, more respondents select the most positive category when all of the categories are labeled and when the scale is presented in two-steps (respondents first choose the direction and then the intensity of their response) rather than in one-step. These effects are compounded where telephone respondents to the twostep format provide the most extreme responses.
Chapter Four describes the forms of communication available in each mode and addresses how differences in visual and aural communication influence the response process. This chapter presents a multi- stage model that explains how respondents perceive and process visual information and synthesizes research on the visual design of surveys from the past decade to provide a practical framework for surveyors to apply when designing questionnaires. Examples are used to illustrate how visual design can help guide respondents in completing self-administered surveys.
Since social research is often based on data from various types of surveys, threats to survey data quality need to be examined and understood by all social researchers. Features of different survey modes influence responses sometimes threatening the quality of our measurements and therefore the conclusions we draw based on them. Further research is needed to understand the biases that may arise as new technologies and mixed-mode surveys continue to profoundly shape survey research practices.