Wednesday, February 3, 2010

A Ph.D without scholarship?

An interesting and controversial question for Ph.D's in science and engineering is too what extent students should be expected to have become "scholars." This issue also came in an earlier post about whether being able to write should be a necessary requirement for getting a Ph.D.

To me, some key characteristics of scholarship are the ability to frame and answer questions, ability to put a problem in the context of a discipline, a knowledge of and critical appreciation for earlier work in the field, and the ability to communicate about the topic.
On the other hand, a really important Ph.D will probably involve some highly technical (and specialised) achievement whether it is making a new molecule, developing a new computer algorithm, deriving or solving an equation, fabricating a new electronic device, or developing a new experimental technique, ....
Scholarship and technical achievement may be almost orthogonal to each other.

At UQ after one year of enrollment in the Ph.D program students must write a ten page report which defines their project, gives a literature survey, and sets goals and a timeline for finishing their Ph.D. I am on the committee which reviews each student. I increasingly struggle with the lack of scholarship and the highly technical focus of the student reports. I have tried to come up with a list of what I would like a student to be able to do at this stage. Here is a rough attempt which will hopefully generate some discussion. (The numbers are extremely rough).

State three big scientific questions in this field.

State a specific scientific question that your project will answer.

What are five key concepts in this field?

What are five landmark papers in the field from the last century? For each paper write several sentences saying why it is so important.

Which three papers from the last 5 years are most relevant to your project?


  1. Thanks for your thoughts Seth and for your challenge.
    My post was NOT meant to be read as a complaint about students. I think significant (and perhaps greater) responsibility actually lies with supervisors. I agree there is a tension between the big and small pictures and I was trying to define what might be some realistic goals, that balance this tension, for a student to have after one year of the Ph.D. I agree the length of Ph.Ds in Australia is too short. Unfortunately, this is largely determined by national government policies. In contrast, in the US it is determined largely by policies of individual departments. But, I also think 7 years (the average time for a physics Ph.D in the US) is too long. At Princeton, where I did my Ph.D, the average was closer to 5 years, at which time students funding gets cut off.

  2. After some further thought, I decided that I was being too aggressive, and was risking offending people that I didn't want to offend. I understand that the problems are endemic and not just a local issue. I also appreciate that certain points I raised are also probably frustrating to you, but that there are many fingers in the same pie, so to speak.

    I couldn't find scope for editing the comment, so I just removed it.

    I agree broadly with your point, actually. I also think that seven years is too long, although I am extremely grateful to have been given the time to sort my goals out.

    For the interested reader: the removed comment made the case that PhD's at UQ (in Australia, generally) are too short. This hinders scholarship because technical detail can be easier and faster to achieve than deep understanding. It also selects for the individuals who can get from A to B in the shortest time (which often means learning only what you need).

  3. Ross - I think having an explicit requirement of answering questions such as you suggest would be helpful, certainly to supervisors. I often find it hard to get new students started - I try to give students what I think is a relatively small problem that will teach them things relevant to the direction in which I want them to head. Often this works well, it leads to a lot of new ideas and helps to define the project. However, others run into difficulties and end up not thinking about the big picture until confirmation rears its ugly head. With an explict requirement such as this, it can be shown to students from day one and will help keep what they are doing in context.

    Interestingly - at the end of my ~4 year PhD I don't think I wouldn have had very good answers to your questions.

  4. One problem is that, unless a student has had previous research experience, or done a more generous and comprehensive physics degree than is typically offered in the Australia/UK system, it is rare that they will be able to frame and articulate such big picture questions at the start of a PhD.

    Hopefully, one will learn to be a scholar in the course of a PhD, and the process of confirmation should assist with this(though as Seth says, it's not much time!) .

    For a such an established and technical field as physics, a lot depends on the supervisor's ability to frame the big questions and to communicate these to the student. After only one year, perhaps it is more a test of the supervisor, and their ability to formulate sensible research projects , than the student.

    Another issue is that what counts as important and worwhile is rather subjective. Different subfields in physics have different methods and approaches to doing physics. What counts as a 'big question' in quantum science may not be very meaningful to an Astrophysicist! We don't students to get caught in the crossfire in discussions about what counts as 'real' physics!

    Further, when it comes to framing 'big questions', we must beware that we are not encouraging merely the formulation of glib sound-bites, that might sound good in a short, confidently presented talk, but actually do not provide any rigourous justificaiton of the specific research project.

    Overall, I guess I'm saying that we should be realistic with our expections, and cautious in locating the problem when our expectations are not met.


  5. I'm not sure I entirely agree that 3.5 years is too short for the PhD. I finished my PhD (in the UK) in <3 years and felt like it was a good apprenticeship. I do think I had a reasonable grasp of some of the questions Ross raises by the time I finished. On the other hand, I felt really insulated from the pressures of academia (ie publish or perish) as a PhD student. This might not have been so helpful when it came to getting a postdoc - I got stuck in Brisbane ;).

    I think actually some of the issue is how well prepared to start doing research. Seth, how much of your 7 years was spent taking courses? I guess ~2. So the closer equivalent in Australia would be a 2 year masters + a 5 year PhD. Ross, how much of the five years at Princeton is spent taking courses?

    The real advantage of the British/Australian system is it is quick. 4 year undergrad + 3 year PhD means your 25 when you start your first postdoc. Plus most British undergrads spend (almost) all of those 4 years just studying physics, which means they get to cover more ground.

    I think given the constraints of funding (and I'd be happy if those changed or if someone had a good idea how to change them), Australia should look towards how things are done in successful places in the UK, rather than the US. I think this because the time constraints are much closer to the UK model. Of the 7 students who passed through my office during my PhD (including me), one dropped out without writing-up, one had a job to go back to in Iran, one has a nice senior fellowship (an independent position), and the rest of us have continuing positions at good universities. However, I think this is much better than average for Bristol.

    One question I have is: has anyone seriously tried to measure this. Which countries/universities have the best long term outcome for students? How can we control this number for quality of student intake?

    It would seem to me that %age of PhDs getting continuing postions in some timeframe (10-15 years, say) after there PhD would be an (relatively) easy and appropriate number to measure. At least for the first part of the question.

  6. These are really interesting issues! I have some thoughts on the structure of the Australian PhD, as raised in the comments, and some thoughts on Ross's original post which I will break into two.

    I am a bit of a bore on the subject of the Australian Ph.D but I'll put pen to paper for anyone who has yet to hear my rant. Consider yourself warned!

    When I completed a B.Sc (Hons) in Physics at an Australasian university 15 years ago it was a focussed four year program with a syllabus modelled very closely on Oxford's, and advanced stream courses in both Maths and Physics starting from first year.

    I then completed a Ph.D in three years at another Australasian university, did a postdoc at a respected international research institution, and felt by the end of all that that I had a fairly good education in Physics.

    I agree with Ben that there is nothing intrinsically wrong with the 4+3 year timescale for a Ph.D but would emphasize that historically it has depended on a high degree of specialisation at an early stage and the availability, at all levels of the undergraduate degree, of more challenging courses designed for those who want a research career.

    I am not sure if any Australian Physics depatment now offers a comparable Honours degree to the one I did. Maybe Melbourne, and maybe Sydney? And I don't believe the issue is confined to Physics and Maths, from what I can see it is common to all Australasian science programs.

    We now have four year undergraduate programs that are more comparable to those at US institutions where students will do a wide range of courses. While there is a research project in the final year, coursework is not at a level that should be regarded as preparation for a research career.

    This sort of degree structure should have a lot of appeal for many students and I don't have a problem with it per se. However, in the US those who are serious about research in government, industry or academia do a 5 or so year Ph.D with graduate coursework.

    The issue is that the Australian science Ph.D program now has effectively NO graduate level coursework. This really is a Ph.D without scholarship, and I think it is why students do not always live up to the sort of expectations that Ross was talking about in his original post.

    The outcome of this is that our graduates typically have very narrow expertise in their area of specialization, and are very much less well prepared to perform independent scientific research than their international competitors.

    If this situation persists it will be VERY destructive of the scientific community in the region and have a serious negative impact on all the quality of research performed here.

    I believe this problem is to some degree appreciated by both academics and Ph.D students, in our department at least. I also believe that our colleagues internationally are often aware of the issue.

    It is not at all appreciated, so far as I can tell, by university management or government, both of whom are essential components of any solution.

    I believe we have a responsibility to explain what has been lost to both groups and to implement one of the two obvious solutions to the problem: a more American system with graduate course-work as part of all research higher degrees or a return to the British model.

  7. Oh, and could I second Ben's call for data on this. My concern on that front is that relevant data would take a disturbingly long time to accumulate.

    Perhaps it would also be important to have a wider appreciation of the enormous changes in science education over the last 20 years.

  8. On the specific proposal in your post Ross, my initial reaction was that perhaps the list is overly ambitious.

    Specifically I am not sure that I could placed my work in the context of the highlights of 100 years of research with any real authority. Of course this may just reflect my inadequacies!

    The questions relating to the shorter timescale context definitely seem pretty reasonable though.

    But as I alluded to above, one big question for me is what is the right standard to ask of students at confirmation given the actual level of preparation that they have in the current system. And given that they are just starting out rather than finishing their Ph.D work. A lot of our colleagues pretty clearly prefer that students start developing their technical capabilities and performing their laboratory or calculational work in the early stages of their Ph.D and I can see many good reasons for doing this.

  9. Thanks for all the posts. This is a record for this blog!

    I agree my proposal is ambitious and maybe unrealistic. On the other hand, maybe this is what the student and supervisor might be able to do at the end of the Ph.D. At least after one year they could just say where they up to at trying to address these questions.

    I don't think that the goal or measure of success of a Ph.D program should be how many graduates end up with permanent jobs at research universities. Rather, I would like it to be how many have learnt the art and discipline of scholarship and are now able to use those skills in any area of employment.

  10. Ross, I agree that it would be great to know how many students "have learnt the art and discipline of scholarship and are now able to use those skills in any area of employment" - but I don't think it's possible to measure that.

    I was just suggesting that getting permanent jobs at research universities is a reasonable proxy and much easier to measure - although Andrew's point about timescales is well made. I am assuming that the %age of students wanting academic jobs is roughly universal and that hiring committees are looking for the kinds of skills you'd like to measure. Clearly neither of those is strictly true.

    I certainly didn't mean to imply that anyone who doesn't end up in an ivory tower has somehow failed.

    However, if anyone has a better suggestion for how to measure PhD outcomes it would be better to hear about. Is there any preexisting literature on this? All the discussion I've heard about it has be very anecdotal.

  11. I think there is a broader context that frames the discussion of the nature of phd's done in Australian (and NZian?) universities.

    Firstly, in regard to Ben's commendable challenge to measure what we're talking about, newly minted british and australian phd graduates should not be directly compared to their US and European counterparts, who often graduate several years older. Rather, a graduate from the british model might be considered still within their apprenticeship throughout the first few years of their postdoc(s). In contrast, graduates from the US or European model are often fully baked researchers, and are capable of parachuting straight into senior postdoc roles upon graduations.

    As a consequence, I've heard it said that one should simply compare outcomes of the various research education models at a given age, rather than at a given career milestone. Perhaps one could measure the productive output of physicists at various ages: say 25, 30 and 40 (in one of the bean-counter metrics we have at our disposal), and correlate this with the educational histories of the individuals.

    Secondly, the role of universities in education has changed dramatically, and continues to do so. Given that many jobs require a university education, students are at university principally to qualify for entry into their future careers. The vast majority of these careers are outside of a formal research setting, and so universities must tailor their offerings to match this demand. Perhaps physics, along with a few other classical disciplines, is a bit special, in that a relatively larger fraction of graduates do go on to do phd's. Nevertheless, the main market for university graduates relies on universities to provide employers with a way of discriminating (via GPA's etc) amongst the different levels of talent in job-applicants.

    The reason I bring this point up is that it gives context to the kinds of courses a university can offer to undergraduates: it will generally favour a broad over deep educational model. This contrasts with Andrew's experience in a notionally elite, research-directed undergraduate. I would certainly support the idea that Australia should have at one elite undergrad institution (I don't think it can afford two: Britain only has 2 or 3, for a population treble Australia's. I know melbourne, sydney and ANU each compete for that niche, though I don't know that they succeed.). However, given other social pressures unique to Australia, such as the relative lack of interstate mobility amongst undergrads, it means the onus of giving a deep education falls on "graduate schools", such as they exist here.

    The problem with this is obviously expense: running courses to transform an australian undergraduate qualification into something useful for a phd student is costly, and with relatively small departments, it seems that this can only be done by pooling resources across universities.

    I've chewed people's ear about this before, and to summarise, I'd advocate the creation of a pilot trial in which a few physics departments collectively offer a course into a pool available for grad students in participating institutions can enrol, using a local course code. The courses would be streamed live over tinternet, so would be a close approximation to having a local lecture. Doing this would enable a relatively complete offering of graduate courses, whilst minimising duplication of costs around the country.

    Excuse the long ramble.

  12. Specifically regarding UQ, as someone who went through the confirmation process, I can see that if the supervisor and student saw before confirmation the sort of list Ross suggests, then you might start seeing a greater percentage of reports that address these issues. I felt after a year of drilling technical skills it was "obvious" a technical report would be requested. What cause do the students have to think otherwise? They would consider: If one goal of my PhD was to the ability to put my project in context, why wouldn't my supervisor have me spend a significant percentage of my time on it in my first year?

  13. ps. this seems like a useful topic to bring up in the HES supplement in the The Australian.... I'd be interested in getting it set up.

  14. I think that's a great idea Tom, I'd be happy to be involved.