(Un-)persuasive robots: Exploring the effect of anthropomorphic cues on the foot-in-the-door effect across three experimental studies

The foot-in-the-door is a sequential request strategy (Freedman and Fraser, 1966): The probability of compliance with a larger request increases if a smaller request is submitted in advance. We conducted three studies in different experimental settings investigating the foot-in-the-door (FITD) effect in human-robot interaction (HRI): An online questionnaire, laboratory experiment and field study. We investigated the FITD effect when applied by a robot and varied anthropomorphic cues along two factors: (1) more or less facial features displayed in the robot face and (2) the robot either self-disclosed ‘personal’ information or not. The experimental setting significantly influenced compliance with the robot's requests, with more items answered in the laboratory experiment compared to the online questionnaire and field study conditions. There was no difference between the latter two. There were no significant effects of facial features, self-disclosure, or the use of the FITD technique, neither on compliance with the large request nor on robot perception. The participants' trusting disposition significantly affected how they perceived the robot. These results highlight the complexity of HRI and the need to carefully consider the experimental setting when conducting and comparing studies on the FITD effect in HRI.