We explore the role of organizational factors in research collaboration networks among European universities. The study of organizational drivers in shaping collaboration patterns is crucial for policy design aimed at reducing research fragmentation and fostering knowledge creation and diffusion. By using Exponential Random Graph Models (ERGMs) and controlling for spatial factors, we investigate the role of two main mechanisms guiding the partners' selection process: organizational attributes and homophily. We investigate two distinct scientific collaboration networks (i.e., projects and publications) and two research domains (Physical Sciences and Engineering, and Life Sciences) over the 2011-2016 time period. Our empirical evidence reveals that, among the main dimensions indicated by the literature, research capability (measured by the dimension of doctoral programs) has the clearest and most stable impact either on the tendency to establish collaboration ties or as homophily effect. In terms of policy implications, it emerges that organizational similarity in research capability matters and policy makers should consider doctoral programs as a strategic variable to promote successful collaborations in scientific research.