Dataset on AI- and VR-Supported Communication and Problem-Solving Performance in Undergraduate Courses: A Clustered Quasi-Experiment in Mexico

Behavioral and educational researchers increasingly rely on rich datasets that capture how students respond to technology-enhanced instruction, yet few open resources document the full pipeline from experimental design to data curation in authentic classroom settings. This data descriptor presents a clustered quasi-experimental dataset on the impact of an instructional architecture that combines virtual reality (VR) simulations with artificial intelligence (AI)-driven formative feedback to enhance undergraduate students’ communication and problem-solving performance. The study was conducted at a large private university in Mexico during the 2024–2025 academic year and involved six intact classes (three intervention, three comparison; n = 180). Exposure to AI and VR was operationalized as a session-level “dose” (minutes of use, number of feedback events, number of scenarios, perceived presence), while performance was assessed with analytic rubrics (six criteria for communication and seven for problem solving) scored independently by two raters, with interrater reliability estimated via ICC (2, k). Additional Likert-type scales measured presence, perceived usefulness of feedback and self-efficacy. The curated dataset includes raw and cleaned tabular files, a detailed codebook, scoring guides and replication scripts for multilevel models and ancillary analyses. By releasing this dataset, we seek to enable reanalysis, methodological replication and cross-study comparisons in technology-enhanced education, and to provide an authentic resource for teaching statistics, econometrics and research methods in the behavioral sciences.