The dissertation: II

Today I’m going to write about the choice of the thesis project. This is nominally a collaborative choice between graduate student and mentor, although in the sciences, it’s often really the province of the dissertation advisor. The fundamental reason for this is that there are no givens in scientific research–it’s not cookbook. Essentially, the success of the thesis project is somewhat stochastic, although with expertise this can be minimized.

Thus, ultimately one of the most important criteria for choosing a major professor, is their ability to pick good dissertation projects. And typically professors become better at picking projects as they gain experience with supervising dissertations. Hence, one of the principle disadvantages of choosing a newly-minted assistant professor relates to the possibility that they might pick a very bad thesis project.

So what makes a good thesis project? First, I believe that it should have a very high probability of yielding publishable results no matter which way the experiments turn out. At the same time, it should be very much hypothesis-driven. At first, this might sound like a paradox. But at least in the biosciences, this is often true (most successful NIH RO1 applications are designed this way).

The second very important characteristic, is that the entire experimental scope of the thesis experiments should be able to be conducted within a reasonably compact period of time. It’s a bad thesis project that wont get its first pilot data for five or so years (think longitudinal studies such as the famous Framingham Project on blood pressure).

It’s also very important to have the experiments manifest practicality. If getting your data requires space on a mission to Saturn–something your proposed advisor has no experience with, then probably something is very wrong with the choice of thesis project (not to mention that the trip to Saturn will take a bit of time).

Finally, I always believe that there should be some aspects of a thesis project that can be “cut out” from the professor’s research program and published potentially independently by the graduate student. As an example, one of my students has such a chapter in his thesis where he developed a new computational methodology for solving differential equations in neuroscience simulations. This is a topic that might be easily “cut out” into a separate paper from his thesis as a single author unit. The advantage with this strategy is that it gives the trainee early evidence of independence–something that will be crucial for career advancement later.


The dissertation: I

Over the next several posts, I’ll put forward some of my ideas about what a successful science thesis should encompass. First some general comments:

The dissertation or thesis is of course the culmination of a graduate students work towards the Ph.D. degree. For the sciences, in general we are looking for original and potentially publishable work. That the work could be published (and usually will be) implies that it advances scientific knowledge in the disciplinary area of the dissertation.

Typically, the research leading up to the thesis takes on the order of 2-3 years and represents a coherent body of experiments, which taken together, form the basis of the thesis chapters. A thesis also typically includes a comprehensive literature review and a somewhat speculative discussion chapter (often the last time a scientist gets to be reasonably speculative in print until they reach the rank of full professor).

There is some debate as to whether the middle chapters of a thesis should be essentially separable as distinct publishable papers (the staple 4 papers together concept) or whether it’s better to have an integrated tome. I have my opinion on this which I’ll reserve for one of the next posts.

Of course a thesis must be defended–usually including a public seminar given by the degree candidate, and an examination session with the thesis committee where, in some European universities, any question at all is fair game–even if not related to the thesis. In the US the defense is often somewhat pro forma–we basically don’t allow someone to defend if it’s likely that they will fail.