The perception of agency in human-robot interaction has become increasingly important as robots become more capable and social. However, there is no accepted method to measure perceived agency. This study defines perceived agency as the extent to which an entity's actions are assumed to be driven by internal thoughts and feelings rather than the external environment. Using this definition, the authors developed and validated a scale to measure perceived agency. They compared their scale to two existing scales and found that their PA and PA-R scales provided a better fit to empirical data. They also validated their scale by showing that it reflects the hypothesized relationship between perceived agency and morality. The study used a Rasch analysis to construct a unidimensional scale of perceived agency. This method allows for the measurement of perceived agency on a logit scale, which is an interval scale suitable for parametric statistical analysis. The authors tested their scale on a variety of entities, including robots, AI characters, and humans, and found that the scale was unidimensional, reliable, and had high separation. They also found that the scale was able to predict the perceived agency of novel entities. In Experiment 2, the authors evaluated their scale against two other existing scales. They found that their scale was significantly better than the other two methods of measuring perceived agency. The PA-R scale, which uses the weights from the Rasch scale, was particularly effective in predicting the perceived agency of entities. In Experiment 3, the authors tested the relationship between perceived agency and morality. They found that participants were more willing to save entities with higher perceived agency. This suggests that perceived agency is linked to morality and that entities with higher perceived agency are more likely to be protected from harm. Overall, the study provides a reliable and valid method for measuring perceived agency in human-robot interaction. The PA-R scale is recommended for use in future research.The perception of agency in human-robot interaction has become increasingly important as robots become more capable and social. However, there is no accepted method to measure perceived agency. This study defines perceived agency as the extent to which an entity's actions are assumed to be driven by internal thoughts and feelings rather than the external environment. Using this definition, the authors developed and validated a scale to measure perceived agency. They compared their scale to two existing scales and found that their PA and PA-R scales provided a better fit to empirical data. They also validated their scale by showing that it reflects the hypothesized relationship between perceived agency and morality. The study used a Rasch analysis to construct a unidimensional scale of perceived agency. This method allows for the measurement of perceived agency on a logit scale, which is an interval scale suitable for parametric statistical analysis. The authors tested their scale on a variety of entities, including robots, AI characters, and humans, and found that the scale was unidimensional, reliable, and had high separation. They also found that the scale was able to predict the perceived agency of novel entities. In Experiment 2, the authors evaluated their scale against two other existing scales. They found that their scale was significantly better than the other two methods of measuring perceived agency. The PA-R scale, which uses the weights from the Rasch scale, was particularly effective in predicting the perceived agency of entities. In Experiment 3, the authors tested the relationship between perceived agency and morality. They found that participants were more willing to save entities with higher perceived agency. This suggests that perceived agency is linked to morality and that entities with higher perceived agency are more likely to be protected from harm. Overall, the study provides a reliable and valid method for measuring perceived agency in human-robot interaction. The PA-R scale is recommended for use in future research.