Résumé en anglais

The difference between individual social capital and organizational (or corporate) social capital has been an important topic of research in sociology during the past decade. The existence of this difference between two forms of social capital evokes an old question in a new manner: what matters most in explaining individual actors' performance? Is it personal social or collective resources provided by the organization to which the individuals belong and in which they work? In this paper we provide a preliminary answer to this question based on a multi-level network study of the top 'elites' in French cancer research during 1996-1998. By multi-level we mean that we reconstituted both the inter-organizational networks of exchange between most French laboratories carrying out cancer research in 1999; simultaneously, we reconstituted key social networks of the top individual elites in cancer research in France during that same year. Given our 'linked design' (i.e., knowing to which laboratory each researcher belongs), we were able to disentangle the effects of structural properties of the laboratory from the effects of characteristics of the individual researcher (including structural ones) on the latter's performance. Performance was measured by a score based on the impact factor of the journal in which each researcher published. Our results show that organizational social capital matters more, and more consistently, than individual relational capital in explaining variations in performance by French top cancer researchers.