La struttura e l'evoluzione del network dei ricercatori e l'impatto sulle performance

Collaboration occurs when a group of autonomous stakeholders of a problem domain engages in an interactive process, using shared rules, norms and structures, to act or decide on issues related to that domain (Wood et al. 1991). In particular in scientific research collaboration is at the same time working method and study subject, since the scientific interest comes from realization that, so far, collaboration is the common way of doing research (Moody 2004). This work is in response to the need of deeply understand the evolutionary dynamic of individuals' collaboration in research group and the connection with performance. This can be done supposing that we can measure a researcher performance from his publications and that this can be related to the network evolution which is at the same time effect of selection and influence regarding the behavior. We will study the scientific collaboration network, which is an evolving self-organizing network based on similarity of researchers' scientific interests. The peculiarity of scientific collaboration compared with collaboration in general is that it is referred to a non hierarchic model. This leaves the researcher free to decide to whom and how to collaborate. Some hypotheses are tested in order to understand this pattern. First we will test if the pattern of attachment is homophily versus functionality or status expectations. Second we will test if the rise of performance of a researcher is positively correlated with network evolution. Finally we will test if the researchers' performance evolution is explained by influence. To explore these questions, applying three methodologies, the Discrete-Time Network Visualization (Powell et al. 2005), the Actor Oriented Modeling (Snijders 1996) and the semi-structured interviews (Spradley 1979; Wengraf 2001), we will study researchers' co-evolution of network and performance in four departments at Bologna University from 1996 to 2007. Results show that homophily, meaning common research interests is the main mechanism to explain the attachment to a research group. Researcher performance is influenced by the groups he belongs to, the principal investigator and his capabilities. The co-evolution mechanism requires strong ties inside the group and weak ties outside. Results do not support the hypothesis that performance rises is positively correlated with network increase. Implications of these findings include an empirical contribution, as the uniqueness of the sample; a methodological contribution, as the refinement of the Actor Oriented Model for non-directed networks; and a theoretical contribution, as the understanding of research groups' evolutionary mechanisms. In the last section of this work we discuss about policy and management implications, limitations and directions for future research.