Virtual learning environments have the potential to support students' development of design skills in engineering education. However, few approaches exist for modeling and measuring design learning as it emerges in authentic practices, which often includes collaboration. This study merges learning sciences research with engineering design education to develop an approach for modeling and measuring design thinking. I propose a "connected design rationale" model which identifies relationships among design moves and rationale. Results from a qualitative examination of how professional engineers make connections among moves and rationales were used as the foundation to examine students in virtual internships. Using digital collaborative chat data and Epistemic Network Analysis (ENA), the discourse networks of students who had high and low scores in the virtual internship were compared to the discourse patterns of professional engineers to determine if measuring connected design rationale reveals meaningful differences between expert and novice design thinking. The results show a significant difference between high and low-performing students in terms of their patterns of connections and that high-performing students in the virtual internship made connections that were more like experts than low-performing students. Results suggest that a connected design rationale model distinguishes between experts and novices in meaningful ways and can be a robust approach for research in learning sciences and engineering education.