Abstract
Recent research has shown that even minor changes in the visual layout of survey questions can affect the way in which respondents answer. Still, little is known about how these changes affect the actual answering process. The recent development of client-side paradata allows us to better understand the way respondents construct their answers by providing new types of data, including time taken to answer a question and whether respondents change their answers. Although a number of studies have used this technology to examine response latency, few have taken advantage of its potential to collect data on the types of changes respondents make while answering survey questions. This study uses client-side paradata to analyze changes in multiple versions of three types of Web survey questions. The results suggest that the visual layout of survey questions not only affects the number but also the types of changes respondents make.