Seite - (000350) - in Knowledge and Networks

348 Hypothesis 3a The greater the general inventive or innovative experience of both partners is, the higher the likelihood of their continued collaboration. Hypothesis 3b The greater the general collaboration experience of both partners is, the more likely it is that their collaboration will continue. The embeddedness of an actor as defined by the number of collaborative ties that the actor has established also determines the number of opportunities for additional collaborations. The mechanism by which the rich eventually get richer explains a certain path dependency in the evolution of networks: Central actors tend to become more central over time (Barabási & Albert, 1999). This phenomenon is known as preferential attachment, or cumulative advantage (Barabási & Albert, 1999; Dahlander & McFarland, 2013). This process might be explained by the broad access that central actors have to information about potential partners and by the high visibility that central actors have for other potential partners (Ahuja 2000). However, the reciprocity criterion applies as well. When seeking to maximize the benefits of the collaboration, central actors are more likely to find that their invested efforts are reciprocated by actors who exhibit the same degree of popularity. Moreover, the bargaining power of central firms is greater than that of the less con- nected actor (Gilsing et al., 2008). If collaboration is to continue, then that power needs to be equally distributed among the partners so as to avoid unilateral depen- dence (Hamel, 1991). Partners are therefore more likely to connect with each other and to maintain this connection if they possess a similar number of collaborative ties (Dahlander & McFarland, 2013). Hypothesis 3c The more similar the degree of popularity of two actors is, the more likely it is that their collaboration will continue. Methodology In our theoretical considerations we identified three main factors that might explain the repetition of innovative linkages in our longitudinal study: (a) cognitive proxim- ity between the cooperation partners, (b) social proximity between the cooperation partners, and (c) similarity in competencies that the partners bring to the collabora- tion. This section presents the database we used, the variables we created, and the methodology we applied. Data To construct potential and realized linkages, we used relational information found in patent applications. Successful collaboration leaves a trail in public patent data because patented inventions can be considered the output of a preceding intensive U. Cantner et al.