Visual and Interactive Issues in the Design of Web Surveys

Regents Of The University Of Michigan - Ann Arbor, Ann Arbor MI

Investigators

Abstract

The rapid growth of the Internet as a vehicle for surveys raises important questions about this new method for data collection. Web surveys are the latest in a series of methods in which computers administer questions directly to respondents. Like many of the earlier methods, Web surveys present questions visually and offer some capability for interacting with respondents. This project encompasses a series of experiments that examine the implications of the visual and interactive character of Web surveys. One set of experiments examines how respondents interpret visual cues in Web questionnaires. These studies test the general proposition that incidental visual features of the questions (e.g., the spacing of response options) can give rise to unintended inferences about their meaning. The studies examine specific hypotheses about the heuristics respondents use in interpreting visual features of questions. These hypotheses state that respondents expect response categories to proceed in a logical progression from left to right; that they expect spatially isolated options also to differ conceptually from the other options; and that they assume that items that look alike must somehow be conceptually related. Two additional experiments examine the effects of including images to supplement the text of a question. Images are necessarily concrete, and one experiment tests whether this leads respondents to interpret the question more narrowly than they otherwise would have. Another experiment tests whether items depicted in an image may serve as standards of comparison for respondents' judgments. The final series of studies examines when respondents take advantage of interactive features of a questionnaire. These experiments test three general hypotheses: Respondents are more likely to utilize information available to them interactively when 1) the information is easy to obtain, 2) it is clearly helpful, and 3) they are highly motivated to seek help. Collectively, the results are likely to lead to practical guidelines for the design of Web surveys. For example, the initial experiments should help settle some important practical questions-Does it matter whether the response categories are arrayed vertically or horizontally? Does the spacing of the response options affect the answers? Does the use of color in response scales influence the distribution of answers? The final experiments could yield better methods for getting respondents to take advantage of features that might improve their answers. This research is supported by the Methodology, Measurement, and Statistics Program and a consortium of federal statistical agencies under the Research on Survey and Statistical Methodology Funding Opportunity.