A project evaluation study on multiset Likert scale data

This work is part of the evaluation proposal for the experimental phase of the ClassMate Robot project, promoted by the Protom Group. The experimentation consists in testing how a newly developed AI device for social education is received in a classroom environment. To assess usability, likability, and social impact pre- and post-trial surveys were administered to the participating students of 4 schools. The data is arranged in multi-block architectures and then summarized with IRT tools. A classic non-parametric approach is employed for testing before and after differences. Post-experimentation results are explored via PARAFAC2 to model school differences while accounting for a multiset structure.