Multidimensional Assessment of Emotional Attribution to Social Robots Among Ghanaian University Students with Depression: A Bifactor Item Response Modeling Approach in Classroom Contexts

The study explored how undergraduate university students in Ghana, exhibiting symptoms of depression, attribute emotional qualities to social robots within classroom settings. A quantitative descriptive survey design was employed to collect data from 752 students at the University of Education, Winneba (UEW) and the University of Cape Coast (UCC) during the 2023–2024 academic year. The survey design was selected to gather standardized data from a large sample, allowing for the identification of prevailing attitudes, beliefs, and experiences related to emotional attribution towards social robots. This methodology was further enhanced by the use of bifactor Item Response Theory (IRT) modeling, which is well-suited for exploring multidimensional constructs such as emotional attribution. Utilizing the Emotional Attribution to Social Robots Scale (EASRS), the research examines how variations in depression severity, measured using the PHQ-9, influence students’ perceptions of social robots’ emotional capacities. Descriptive statistics revealed that students had higher mean scores for the affective (M = 3.84) and social-emotional (M = 3.67) dimensions, with cognitive attribution showing the lowest mean (M = 3.21). A bifactor model analysis confirmed excellent model fit indices (CFI = 0.948, RMSEA = 0.042, SRMR = 0.035), suggesting that the proposed structure of emotional attribution, encompassing both a general factor and specific emotional dimensions, is psychometrically sound. Regression analysis indicated significant negative relationships between depressive symptoms and emotional attributions across all dimensions, with cognitive attribution being the most strongly affected (β = -0.34, p < 0.001). Gender differences were negligible, and academic level differences revealed that postgraduate students attributed higher cognitive and social-emotional capacities to robots compared to undergraduates. These findings underscore the role of depression in shaping emotional perceptions of social robots, providing insights into the psychological factors influencing human-robot interactions in educational and mental health contexts.