Measuring Fundamental Aspects of the Social Perception of Robots: Development and Validation of a Shortened Version of the RoSAS (RoSAS-SF)

Abstract

The robotic social attributes scale (RoSAS) is an 18-item measure used to assess the social perception of robots (Carpinella et al. in ACM/IEEE International Conference on Human-Robot Interaction, 2017:254–262). The full scale has sound psychometric properties, but its length makes it suboptimal for use in many studies of human-robotic interaction (HRI). Here, we report the development of a shortened form of the RoSAS (the RoSAS-SF), a 6-item scale useful for assessing judgments of warmth (compassionate, social), competence (competent, reliable), and discomfort (scary, awkward) toward robots. Across two studies, the psychometric properties of the RoSAS-SF are compared with the full RoSAS using participants’ judgments of both synthetic and real images of robots. Results indicate that the RoSAS-SF yields patterns of trait inferences and effect sizes comparable to the original scale. Moreover, confirmatory factor analyses reveal that the RoSAS-SF offers acceptable model fit with a 3-factor solution. One notable finding is that the discomfort subscale showed variable reliability in capturing responses across different robot types, suggesting caution in relying upon the RoSAS-SF as an exclusive measure of this judgment dimension. Applications, advantages, and recommendations for using the RoSAS-SF in HRI research are discussed.