Existing studies on crowdsourcing have focused on analyzing isolated contributions
by individual participants and thus collaboration dynamics among them are under-investigated. The value of implementing crowdsourcing in problem solving lies in the aggregation of wisdom from a crowd. This study examines how marginality affects collaboration in crowdsourcing. With population level data collected from a global
crowdsourcing community (openideo.com), this study applied social network analysis and in particular bipartite exponential random graph modeling (ERGM) to examine how individual level marginality variables (measured as the degree of being located at the margin) affect the team formation in collaboration crowdsourcing. Significant effects of marginality are attributed to collaboration skills, number of projects won, community tenure, and geolocation. Marginality effects remain significant after controlling for individual level and team level attributes. However, marginality alone cannot explain collaboration dynamics. Participants with leadership experience or more winning ideas are also more likely to be selected as team members. The core contribution this research makes is the conceptualization and definition of marginality as a mechanism in influencing collaborative crowdsourcing. This study conceptualizes marginality as a multidimensional concept and empirically examines its effect on team collaboration, connecting the literature on crowdsourcing to online collaboration.