Undergraduate research experiences confer benefits on students bound for science, technology, engineering, and mathematics (STEM) careers, but the low number of research professionals available to serve as mentors often limits access to research. Within the context of our summer research program (BRAIN), we tested the hypothesis that a team-based collaborative learning model (CLM) produces student outcomes at least as positive as a traditional apprenticeship model (AM). Through stratified, random assignment to conditions, CLM students were designated to work together in a teaching laboratory to conduct research according to a defined curriculum led by several instructors, whereas AM students were paired with mentors in active research groups. We used pre-, mid-, and postprogram surveys to measure internal dispositions reported to predict progress toward STEM careers, such as scientific research self-efficacy, science identity, science anxiety, and commitment to a science career. We are also tracking long-term retention in science-related career paths. For both short- and longer-term outcomes, the two program formats produced similar benefits, supporting our hypothesis that the CLM provides positive outcomes while conserving resources, such as faculty mentors. We discuss this method in comparison with course-based undergraduate research and recommend its expansion to institutional settings in which mentor resources are scarce.