Influence of co-authorship networks in the research impact: Ego network analyses from Microsoft Academic Search José Luis Ortega 1 Cybermetrics Lab, CCHS-CSIC, Madrid, Spain, jortega@orgc.csic.es Abstract The main objective of this study is to analyse the relationship between research impact and the structural properties of co-author networks. A new bibliographic source, Microsoft Academic Search, is introduced to test its suitability for bibliometric analyses. Citation counts and 500 one-step ego networks were extracted from this engine. Results show that tiny and sparse networks –characterized by a high Betweenness centrality and a high Average path length– achieved more citations per document than dense and compact networks –described by a high Clustering coefficient and a high Average degree–. According to disciplinary differences, Mathematics, Social Sciences and Economics & Business are the disciplines with more sparse and tiny networks; while Physics, Engineering and Geosciences are characterized by dense and crowded networks. This suggests that in sparse ego networks, the central author have more control on their collaborators being more selective in their recruitment and concluding that this behaviour has positive implications in the research impact. Keywords: Bibliometrics, Academic search engines, Ego networks, Research impact, Co-authorship 1. Introduction Scientific collaboration is an intrinsic part of the research activity because it makes possible to share technical resources and to exchange new ideas that help to face new scientific challenges (Katz and Martin, 1997). However, and as argued by Latour and Woolgar (1986), the scientists’ behaviour could be influenced by the type of research that they perform that makes possible producing different research products. In that sense, it could also be suggested that each research discipline generate a particular collaboration pattern. On the other hand, the current professionalization of Science has increased the number of scientific partners such as in large and dense institutional research groups, due to mobility of doctoral students or visiting professors, and thanks to the ubiquitous telecom facilities. The role and involvement of these partners differ considerably when it comes to carry out a research project. In this way, the disciplinary differences and the multiple roles that each actor acquires in a research process are key elements to understand co-authorship activity in relation with the research impact and productivity (Narin and Withlow, 1990). Even more, the existence of multiple and diverse interactions between all the contributors could emerge complex partnership structures which properties could also affect the research performance. This new point of view based on networking structures and helped by social network analysis (SNA) techniques can puzzle out how the collaboration structures are able to configure the 1 Cybermetrics Lab, CCHS-CSIC, Albasanz, 26-28 28037 Madrid, Spain, Tel. +34 916022603 1