Measures
The study included three objective measures of participant compliance: total time spent working on the task, total number
of puzzles solved, and total number of disks moved across all instances of the puzzle. The follow-up questionnaire contained
several seven-point rating scales for assessing the performance of the robot. Five items from this questionnaire were combined
into a single construct of perceived robot performance, including questions about the robot’s skills as an instructor and motivator, as well as questions relating to the usefulness of the
robot’s information and advice. This construct was found to have good internal reliability (Cronbach’s = .75).
The follow-up questionnaire also included open-ended questions about why the participant chose to participate for as long
as they did and why they chose to eventually terminate the interaction. In order to obtain a more task-specific measure of
intrinsic motivation, these open-ended responses were coded for explicit mention of the participant’s inherent desire to
solve the puzzles, without mention of any external motivation coming from the robot.