Monday, March 17
Tom Vogl: Differential Fertility, Human Capital, & Development
Conrad, Frederick G., M.F. Schober, and T. Coiner. 2007. "Bringing features of human dialogue to web surveys." Applied Cognitive Psychology, 21:165-187.
When web survey respondents self-administer a questionnaire, what they are doing is in many ways similar to what goes on in human-human interviews. The studies presented here demonstrate that enabling web survey respondents to engage in the equivalent of clarification dialogue can improve respondents' comprehension of questions and thus the accuracy of their answers, much as it can in human-human interviews. In two laboratory experiments, web survey respondents (1) answered more accurately when they could obtain clarification, that is, ground their understanding of survey questions, than when no clarification was available, and (2) answered particularly accurately with mixed-initiative clarification, where respondents could initiate clarification or the system could provide unsolicited clarification when respondents took too long to answer. Diagnosing the need for clarification based on respondent characteristics-in particular, age-proved more effective than relying on a generic model of all respondents' need for clarification. Although clarification dialogue increased response times, respondents preferred being able to request clarification than not. The current results suggest that bringing features of human dialogue to web surveys can exploit the advantages of both interviewer- and self-administration of questionnaires.