Measuring social embeddedness : how to identify social networks in science-industry partnerships ?

Social embeddedness appears to be a promising way to analyze knowledge collaborations, notably to better understand their build up and their spatial patterns. Nevertheless, measurement problems and an over-territorialized conception of the notion exist. When studying the formation of these partnerships, authors have underlined the embeddedness of innovators in social ties as a major factor (Walker, Kogut, 1994 ; Zucker et al., 1998); others have shed light on institutional devices (Ponomariov, Boardman, 2010 ; Eom, Lee, 2010), but few have integrated both relational and institutional forms of embeddedness. Moreover, "embeddedness is mostly conceived of as a spatial concept related to the local and regional levels of analysis" (Hess, 2004): scholars argued (Moka et al., 2007) and showed (Fischer, 1982 ; Wellman, 1996; Grossetti, 2002) that social ties easily build-up in the neighborhood. They thus conclude social embeddedness favors local partnerships without demonstrating it really. Finally at the empirical level, precise data are missing to identify social embeddedness (Giuri, Mariani, 2007). Therefore, regarding the existing studies, "the analytical scales and the spatiality of embeddedness needs to be scrutinized" (Hess, 2004) theoretically and empirically to determine "who is embedded, in what and what is so spatial about it ?" (Pike et al., 2000). We propose here to address this deficit thanks to the formulation of a method robust enough. In this perspective, an analytical framework which does not postulate the social network hegemony is needed. We realized further theoretical refinements by introducing the concept of « coordination resources » to indicate modalities that permit connections between actors without using interpersonal ties. To identify embeddedness effects, we then present an original method essentially based on interviews. We use it to a group of 264 cases of science-industry collaborations realized in France. Several results are revealed thanks to statistical and econometric treatments. We reaffirm the major weigh of social embeddedness in the build-up of partnerships and the complementary role of coordination resources. Social embeddedness appears to be independent from the partners features. It nevertheless impacts the geography of the partnership although it is not possible to associate systematically social embeddedness and local collaborations.