St-Louis, B., Roberge, G., Lavoie, R., Campbell, D., Côté, G., and Archambault, E. (2016)
Presentation by Bastien St-Louis (April 3, 2016)
Homophily is a measure used in social network analysis to highlight the extent to which actors exhibit a preferential attachment to other actors sharing a common characteristic for some attribute (e.g., geographic location). Traditionally, homophily is calculated by comparing the average path length between actors sharing a particular characteristic to the average path length between all actors in the network. However, datasets often contain isolated actors (or components consisting of small sets of connected actors) not tied to all others in the complete network. This was the case for the datasets considered in a network analysis Science-Metrix conducted for the European Commission.
In such a context, the traditional definition of homophily is for the most part useless because it becomes impossible to compute the average path length between disconnected actors. To overcome this limitation, Science-Metrix defined a new homophilic metric for the Commission’s analysis. The metric ranges from -1 (highly heterophilic) to 1 (highly homophilic), with 0 representing a neutral network. A neutral network is one in which the observed occurrences of homophilic (a tie between two actors sharing the same characteristic) and heterophilic (a tie between two actors with different characteristics) pairs are equal to their expected occurrences based on a random network of the same size (with same number of actors and edges) and with the same frequency distribution of actors across the observed characteristics of the attribute of interest. This new metric was then applied to the set of real-world data representing the co-participation network of researchers funded under the European Commission’s 7th Framework Programme for Research and Technological Development (FP7).
Under FP7, a number of actions were implemented to boost innovation by promoting increased linkages between the public (academic and governmental) and private sectors. In this context, it is highly relevant to investigate whether the FP7 network—relative to its predecessor under FP6—evolved towards a more heterophilic structure in regard to the clustering of the participating organizations according to their sector (in this case, higher education, private, public body, research organizations and "other").
The global sectoral homophily was found to decrease from FP6 to FP7, indicating that integration of sectors was promoted under FP7 (i.e., the different sectors collaborated more together). As a result, one can hypothesize that innovative ideas under FP7 reached the private sector more easily, potentially boosting innovation and, in the long term, the economic competitiveness of Europe. An extended analysis collating multiple sources of evidence (collaboration networks, surveys and interviews) confirmed that the presence of SMEs in FP7 projects significantly increased knowledge transfer from research to market as well as the propensity of projects to introduce innovation in the form of new products or processes. A finer analysis by specific funding mechanisms under FP7 highlighted the efficiency of the projects that were purposefully designed to achieve this goal.
Image: iStock Photo