Abstract
Internet surveys are currently used in many academic and marketing research fields. However, the results for these surveys occasionally show traces of response bias. In our study, we analyzed how response bias appears in lengthy preference judgments. 1042 respondents participated in lengthy sequential preference judgments. Three stimuli series were used: scene pictures, Attneave nonsense shapes, and point-symmetric figures. One hundred stimuli were selected for each series and individually displayed on a computer screen, with presentation order randomized for each respondent. Respondents were then asked to rate their degree of preference for each stimulus. Mean preference scores increased over the first 10–20 trials, then, gradually decreased from the middle to the last trial. Furthermore, participants tended to produce the maximum and minimum score during early trials. These results demonstrated that response bias can be a function of presentation order.
Highlights
Questionnaires have been used in psychological, social, and marketing research, national population censuses, and many other fields for a long time
We analyzed how response bias appears in lengthy preference judgments. 1042 respondents participated in lengthy sequential preference judgments
The purpose of this study was to examine response bias in lengthy sequential judgments and to examine how this form of bias differs between different stimuli series
Summary
Questionnaires have been used in psychological, social, and marketing research, national population censuses, and many other fields for a long time. Internet surveys have become as widely used as questionnaires Such survey methods are common tools for examining respondents’ ‘‘attitudes’’ or ‘‘thinking’’. Respondents are asked to expend a great deal of cognitive effort such as interpreting question meanings, recalling memories, integrating information, and reporting their judgments Because of these intensive processes, respondents tend to choose midpoint responses to minimize cognitive effort. In a survey, respondents provide too many midpoint responses as a means of representing nonresponses the survey results may be concentrated near the midpoint of the response scale to an excessive degree. This could cause misunderstanding of ‘‘actual’’ respondents’ answers
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
