The use of group work in student centered collaborative learning in STEM courses has widely increased over the past decades. There is significant evidence available in literature about the benefits of group work in terms of student learning, peer acceptance, and retention of knowledge. However, group work is not beneficial if the groups are not structured strategically. There is conflicting evidence in the literature about what type of group structures are most beneficial for enhanced learning. There is a significant gap in the literature about how to compose groups in order to maximize cognitive engagement, performance, and satisfaction and minimize interpersonal conflicts in undergraduate inquiry‐based introductory labs. One of the goals of our study is to understand the impact of the group selection methods on student network formation in association with group dynamics and conflicts that arise.To explore this idea, we used a socio‐cognitive perspective in combination with the framework of Cooperative Learning Effects on Learning to conduct a longitudinal study in thirty‐five sections of an undergraduate inquiry‐based biology laboratory at the University of Georgia (n=796). Our data was collected over a 3‐week period in the form of 7 surveys over the summer semester and over 7 weeks in the Spring semester. After each lab, students were given a survey that asked them to report who they worked with, things they noted as good and bad traits in their group mates, how frequently they communicated, how they rated effort from their group mates, and if they planned to work with these students again the following week. The data from the first five surveys in each of the semesters was analyzed quantitatively by social network analysis.The student network analysis showed that there is clustering with respect to gender and ethnicity regardless of whether students are in the structured or unstructured lab. Empirical observations suggest that pre‐class friends choose to work together. In the unstructured sections, students chose to work with the same group members throughout the lab course, even when given the option to change groups. Students would rather stay with a problem group member that they know rather than confront the issue and change groups. In the preliminary data, students were asked to identify “good” and “bad” group member traits. The top three good group member traits were “prepared” “interested in class”, and “participated.” Students in the unstructured labs (9%; n= 540) are less likely to report negative comments about their group members in comparison to students in structured labs (16%; n=532). In the future, we are going to run qualitative analysis in case studies to understand at what point students will report intragroup conflicts.Support or Funding InformationThis study was funded by the National Science Foundation (NSF) under grant no. 1659423 as part of the UBER‐REU program, however; the findings of this study do not explicitly represent the views of the NSF. We would like to thank the GLAs and students who contributed data to this study. Additionally, we would like to thank Destiny Williams, Franklin College of Arts and Sciences, the Department of Cellular Biology, and the SEER Centre for contributions to this project.