Thursday, January 14, 2016

Can we make life better for graduate students? The two advisor trick

The latest reported sexual harassment case occurred at Caltech with a professor I had known and thought very highly of. Caltech issued the typical response saying that they have "zero tolerance for sexual harassment".  Scientists across the world agree that departments turn a blind eye to various forms of abuse (see We Suck (but we can be better) by Sean Carroll or Something deeply wrong with Chemistry from some time ago).  I believe that it is timely to brainstorm, propose and implement changes that improve the life of graduate students (and postdocs).

A hiking trip
my best friend from grad school and I
My PhD  advisors were Ira Wasserman and Saul Teukolsky (Saul is also on wikipedia). They were both full professors at the peak of their careers. When I was their student, Saul was the head of the physics department and Ira was the head of the astronomy department. My thesis was on the nonlinear dynamics of R-modes, a type of oscillations driven unstable by gravitational radiation emission, in neutron stars.  Every time I stumbled and one of my advisors would say something discouraging, the other would give me the hope I needed to move on.  Since they complemented each other quite well, I cannot imagine finishing my PhD with only one of them.

Multiple committee members is a common requirement for most universities, but a dual advisor system is not. The committee members are people a student sees in exams (universities typically have a qualifier exam, a thesis defense, and sometimes an exam in-between that presents the thesis idea) and to get signatures. These entail a few meetings over a period of five or six years. The advisors play a much more important role. They regularly meet with the student, and shape her or him into the researcher she/he becomes. Most often professors do not mean to hurt their students. They hope to teach them everything they know, and want to make sure that what comes out of their group is daring, correct and as perfect as it can be. It is hard for a student to make the difference between legitimate concerns about performance or scientific goals and a personality conflict. Yet to progress one must act on valid criticism and have reasonable expectations on both the scientific goals and how much work is necessary to achieve them. A second opinion from someone with equal power and experience who understands and is part of the project can be crucial at preventing many types of problems.

On top of Space Sciences: Dave, me and Andy
the office next door: Ali with the others
I was a PhD student at Cornell between 2003 and 2008. In the physics department, we were four women in a class of over 40 students and at universities who actively try to recruit women, the situation is similar today. I enjoyed my time there among many talented overachievers. When I was a first and second year graduate student my friends and I watched Kip Thorne's gravitational wave lectures that had just been recorded by my brother, went hiking, and frequently talked about science. We were thrilled to interact with the brilliant Cornell professors on the same floor (the rest I did not see unless I took a class with the person), and with incoming speakers over lunches and dinners. My science discussions with Andy, Mihai and Dave eventually resulted in a paper on the importance of finite mirror effects for advanced LIGO detectors that was published in Physical Review D. I had worked on other projects before and after this, but this paper was the only project with no senior people on it, i.e., where none of the authors had a PhD to begin with. We all hold PhD diplomas now.

Why two advisors?  
Two people are less likely to make the same mistake. Emails and meetings on the details of a project include all collaborators, which typically prevents people from sharing personal or harassing information, and when there is a problem you get it explained in two different ways by two very smart people.

Typical problems
1) Very long PhDs or no doctorate degree after many years of working on a project.  I have had an office mate who worked on his PhD for ten years and yet did not have time to publish any of his findings in a journal.  There were also plenty of people who did not graduate at all. If after a number of years (sometimes up to six) the research is not deemed successful, students receive a masters. A masters can also be achieved in one year of study; I got my first masters in one semester, but this was atypical. Projects that last too long are a sign of mismanagement.

2) Taking too long to read a paper written by a student & allow the publication of the results. This is quite common. It is also extremely stressful and frustrating. While the paper sits on the advisor's desk, it can happen that somebody else produces the same results and the student has no thesis left at all or that the methods/equipment used change to make the problem obsolete.

3) Obvious mental problems. Both the student and/or the professor can have mental problems ranging from depression to bipolar disorder to other forms of problems. I have limited experience with such situations. If you have two advisors, it is unlikely that both will suffer from the same thing. If the student is ill, it can help to have two people to rely on. Either way, emails sent to/conversations that include more than one person prevent harassment.

We had a professor at Cornell, who was very talented, but on alcohol & drugs. His problems were known to everyone in the department, and yet he had four students. Two of his students quit science without a PhD. The other two had the strength to continue after he left and were encouraged to choose new advisors. Unfortunately, this was four years later, and so the duration of their PhD extended, too. If they had more than one advisor to begin with, the sane person could perhaps have taken over or at least provided some advice and support.

4) Unhealthy competition. If you work with two smart people, they may help you gauge what is important better.

5) Sexual Harassment. It has recently been estimated that close to 90% of women suffered from some form of sexual harassment in the tech community (estimate based on a sample of 200). The numbers must be similar in technical fields in academia.  A two-advisor model provides another person to turn to if there is a problem in the department. It is also less likely that a colleague/professor would harass a student in group meetings or through group emails. More people in the group/room and more diverse groups help diffuse awkward situations. If a person behaves inappropriately, they have to be removed from the project for the group to function fairly.

Attraction is common between people, and it has to be acknowledged that we do not choose how feel, but choose how we act. Different people react differently when they are attracted to someone. Some are kind to people they are attracted to, and some make unkind comments and put them down. The latter category is surprisingly common and causes more disturbances in the work place than the former. It is hard to control this behavior that comes naturally and may in part arise due to repressed feelings by telling people to repress their feelings more. Training may help, but I worry it only has short term effects.  I do not believe that someone can change their personality because it is part of who they are; a "training to be nice" program that starts at the age of 45 or 50 is not likely to work efficiently. On the other hand, choosing people who are compatible in the first place to work together should make a difference.

So far (in the past 20+ years) we have not made (much) progress towards equal treatment & equal numbers of women and men in the technical world or towards making the work environment more healthy for everyone. On the bright side, most of the hassles that I have to face today are likely to be less than what my mother/grandmothers/great-grandmothers faced years ago. Most women can work, vote and drive, and with no major upcoming war, things should get better if we actively try to change dysfunctional programs.

Drawbacks of the two-advisor model.
Few professors are as kind and as talented as Ira and Saul. My friend and colleague, Ali Vanderveld tells me Ira and Eanna Flanagan made a wonderful two-advisor team, too, and I am sure there are other people out there who were happy with such graduate school experience. But there will be cases when this does not work well. If the professors do not get along, the student may be pulled in different directions and forced to act as a mediator. Sometimes a less senior person works (e.g., a postdoc) with the student, and the professor is on the paper because he provides the funding and work-environment, but little else. This is less likely to happen at top US institutions, but more common in Europe. In this case, it may not make sense to add another uninvolved person to the work. In smaller universities, there may not be two professors who work in the same field. Then the second person would have to be external.