Low-Effort, Unqualified, and Malicious User Behaviors in Online Human-Robot Interaction Studies

This work was motivated by a previous study that saw unintended behaviors from some participants in online survey interactions. We conducted a post-hoc analysis of the study's participant data, including an analytical retrospective of effectiveness of reactive process changes, qualitative analyses of user behaviors, and a quantitative analysis of user responses. We defined novel user types of Low-Effort, Malicious, and Unqualified for this survey data and discussed general study design recommendations to proactively identify and discourage these user types, such as participant identity verification and communication of expected effort. We also suggested directions for future work to further investigate these user behaviors, especially in contextual domain frameworks. Analyses and discussions were conducted in the context of human-robotics interaction and adolescent users, generalizable to user research studies in human-computer interaction, in the hopes of standardizing study design discourse and improving user research data quality in these underexplored areas.