Comment les docteurs deviennent-ils directeurs de thèse ? Le rôle des réseaux disponibles How do PhD students become PhD advisors?

In this paper we study how doctors become PhD advisors. The fi rst contribution of this paper is methodological. We show how we can use administrative data, the French PhD data- base (1970-2002), that is relatively poor and lacks of individual information, but that is quasi-exhaustive and can help in order to reconstruct academic careers. It is therefore possible to cal- culate the time separating a researcher's own PhD defense and that of his fi rst doctoral student. The second contribution is theoretical and empirical. We would like to return to the ques- tion of the strength of social ties and to nuance Granovetter's conclusion concerning the fact that this strength is due to infor- mation diffused through the network. Networks count in aca- demic world because they generate also some support (espe- cially support of the adviser for his PhD student). This support is much more exclusive than just the spread of information. In order to test the importance of contacts, we link the recruit- ment process with the mobility network of advisors. When they change universities, advisors increase for a while their number of contacts and make them profi table for their PhD students or those of their colleagues. We estimate the impact of network with Mantel Haenszel odds ratios in the one hand and Cox survival model in the other hand. We use the fi xed effects tech- niques in order to control for unobserved heterogeneity. Those models show that the reproduction of faculty highly depends on the local conditions of competition, especially during the recruitment process underlying our observations. We confi rm by our research the importance in France of an academic in- breeding due more to a preference for institutional proximity than to a preference for geographic proximity. We show also the importance of informal queuing lines, with the eldest shading the chance of the youngest. But contacts do not solely have a role locally but also at distance, through networks. Outside candidates linked to the recruiting department by intermedia- tes are more likely than other outside candidates to be hired and to have an academic career. We show however that in this very competitive universe, contacts do not act systematically as supporters. They do so if they are available, if they do not have their own interest to defend and their own candidates to support.