China Dissertations in France. A preliminary survey (3)
This blog post examines dissertation committees. It is a modest attempt to explore network analysis methods on a small sample of data.
The standard practice in French academia to receive the Ph.D. degree is the formal presentation of the dissertation by the candidate before a dissertation committee (jury de these). It is not, as in the U.S. system, a dissertation committee that oversees the progress of the candidate throughout his/her years until the completion of the dissertation. It is a one-time formal and public event when the candidate can finally discuss his/her work with an array of 4-6 scholars, including the advisor(s). It is a crucial moment for the candidate both intellectually and in terms of his/her future career. After the defense, the committee produces a report (rapport de soutenance) which is key to recognition by the peers and a necessary passport for ulterior applications for jobs.
I was hoping to do a systematic study of the dissertation committees before which candidates presented their work for China-related topics. Unfortunately, theses.fr does a very poor job of recording the members of such committees. The metadata contains a “persons” variable that lists names of committee members. Yet in most cases, the information is limited to the name of the candidate and his/her advisor(s). From the data on the 1,265 dissertations, I was only able to retrieve the composition of the committee for 291 dissertations, hardly a quarter of the total.
My take on committees was to examine them through network analysis to see what kind of connections may exist between their members, especially for a given field or discipline. The restrictive sample I have used did not allow for such a systematic approach. I will thus present a preliminary picture that still show the potential for such a study.
To implement this approach, I used my favorite network analysis tool, Cytoscape. To introduce a bit of network analysis lingo, the sample data contained altogether 1,142 nodes, distributed in three categories: graduates (291), advisors (216), and committee members (635). The data, however, was incomplete even for this set of dissertation committees. The average size of the committees was 3.5 scholars, which is below the composition of most committees (4). The projection of the data in Cytoscape produced a very high number of connected components — 99 — which reflects two parallel phenomena: the high number of “one-time” China dissertation by a majority of advisors and the incompleteness of the data. With complete date, there would still be a high number of isolated components, but one one would expect a larger main component. The figure below illustrates the disconnection between these isolated components.
Yet one can also see in Figure 1 that many links exist between dissertation committees. The “connectors” in this case are the scholars who took part to theses committees and, among them, advisors play an important role since they usually contribute mainly to the constitution of these committees. Those who organized a number of such committees appear as “brokers” within this network. Figure 3 highlights some brokers in the global network. Here we can see scholars who seem to play a larger role, but even if they did contribute more to the constitution of dissertation committees, this also reflects the degree of more complete information on them in the data.
Of course, this exercise tends to create an artificial connection between its members. First, this is a network that represent a long period. Scholars were rarely of various committees with the same colleagues at the same time. Second, while this network may reflect interpersonal relationships — this could be measured by the number of cross-participation to committees, it should be considered more like a network of citations. Somehow, the participation of a scholar to a committee is the sign of his/her citation as an intellectual contributor to the assessment of a dissertation. Each one brought his/her specific mark.
With more compete data, exploring the sub-networks by discipline could reveal much better the dynamics of interpersonal exchange between scholars. In the sample data I used, I was only able to produce three subsets with enough data to provide an interesting perspective on what could be done. Figure 4 , 5 and 6 present networks based on history, economy and law respectively. They reveal potential patterns that deserve further study with more complete data.
The use of network analysis offer a genuine possibility to examine how scholars alternatively play the role of advisor and committee member in dissertation committee. It also allows a window on how a scholar chose to configure the dissertation committee of his/her students over time. Since this is the only case for which I have complete data, I will put myself to the test here by presenting the network of committee members of my own students since 2000. With more complete data on all dissertations, we could probably improve our knowledge of the dynamics of inclusion and exclusion at work in the constitution of dissertation committees.