Much of the literature on research-oriented teaching relies on the impressions of instructors who have experimented with such practices. Often authors are enthusiastic. This article shifts the focus to assess student satisfaction with such methods. We hypothesize that a student preference for research-based learning in Political Science would grow out of having a positive research experience during their studies, and that on average this retrospective satisfaction will depend on the nature of the collaboration. Using multicountry survey data spanning eleven institutions, we offer a series of linear models that distinguish three different research experiences: research with their professors, research with their peers, and solo research. Our results indicate that research experience does predicts student preference for research-based learning, but the direction of that pattern depends on whether they performed the research with their professors or other students. Whereas research experiences with other students positively predict a preference for research-oriented teaching, past research collaboration with professors generated a negative assessment, especially among students with no methodological training, and who did not use their involvement in such research projects to improve their methods skills.