An empirical study of how much a social robot increases the rate of valid responses in a questionnaire survey

In this paper, we report that how much the presence of a robot increases the rate of valid responses in a questionnaire survey. We conducted a field trial in a university campus based on a single-case design method that is used in applied behavior analysis. We made two conditions: One was a condition that a robot interacting with people was installed (robot condition), and another was a condition that the robot was not installed (no-robot condition). We measured once a day for eight days, each day lasting 125 minutes from 11:25 to 13:30. The robot condition and the no-robot condition were assigned alternately to each day. The robot was controlled by an operator. The results showed that the valid response rates of the robot condition were 8.9%, 3.8%, 6.1%, and 1.9%, and those of the no-robot condition were 4.9%, 3.7%, 3.1%, and 1.8%. Considering both these results and complete answer rates, we found that although a robot can attract people’s attention and increase the response rates a little and short-term, the valid response rates do not increase so much as we expect because people who are attracted by the robot are likely to quit answering the questionnaire halfway. In order to increase the valid response rates, we will need to consider a new interaction design for preventing people from quitting their answer.