ABSTRACT The biomedical research workforce faces steep declines in recruitment and retention of under-represented researchers, from undergraduate study to the senior professoriate, and an escalating attrition of those at the PhD-level as they are nearing potential careers as independent investigators (Fuhrmann, Halme et al. 2011). To address these challenges, evidence-based mechanisms that enhance mentoring, networking, and career transitions are needed, but the specific mechanisms that make them effective are still unclear. Elucidating these mechanisms is critical to future efforts to diversify and retain the biomedical research workforce. To that end, our objective is to test a predictive model of the influence of skills for communicating across difference (CAD) on long-term outcomes for both students and their junior mentors (graduate students and postdocs). Our central hypothesis is that acquiring CAD skills will influence distal outcomes for diverse trainees' career persistence and network growth, as well as junior mentors' subsequent engagement in mentoring of diverse trainees, and can be modeled using an adapted version of the Tripartite Integration Model of Social Influence (TIMSI). Our specific aims are to measure effects of a CAD workshop series in a longitudinal, randomized controlled trial of dyads of summer students and their matched junior mentors, with additional group-effects analysis for under-represented groups; 2) to identify causal relationships of the variables and create 2 related TIMSI-based models: one for career intention outcomes and one for mentoring intention outcomes; and 3) to assess long-term impact of CAD skills and predictive utility of the student and mentor models by tracking outcomes of known Hallmarks of Success such as completion of STEM degrees, publication, and others. By investigating the role of CAD skills, we expect to model the psychological processes by which research career persistence is strengthened for all participants, and commitment to mentoring diverse trainees is strengthened for junior mentors. Innovations of the proposed research include its unique focus on linguistic skills, the dyadic and interactional analytic approach, the analysis of psychological development processes as senior trainees become junior mentors, and the use of flexible, generative skills rather than acquired knowledge as a mechanism of change. Our long-term goal is help diversify the biomedical workforce through an approach which addresses the interactional roles of both under-represented and well-represented groups simultaneously. This research, conducted by our uniquely qualified team, directly addresses priorities outlined by the NIH National Research Mentor Network and will contribute to our understanding of mechanisms that enhance mentoring and career transition and thereby broaden participation in the biomedical workforce.