Perceived Human-Likeness of Social Robots: Testing the Rasch Model as a Method for Measuring Anthropomorphism

Anthropomorphism is generally defined as the attribution of human-like characteristics to social robots and other non-human objects. We argue that different researchers have different interpretations of this concept, leading to measuring instruments that focus on different subsets of human-like characteristics. In the current paper, we discuss these different interpretations and explore a new method for measuring anthropomorphism, based on the Rasch model. The aim of the current work is to map anthropomorphism as a range of human-like characteristics on a one-dimensional scale. The scale's validity and sensitivity were tested by comparing it with two available measuring instruments and by comparing people's responses to different types of agents. In three studies, we explored whether the Rasch model is suitable for measuring anthropomorphism. Despite some limitations, results showed that the Rasch model can successfully be applied to the measurement of anthropomorphism. Implications for future work on anthropomorphism of social robots are discussed.