Wednesday, February 17, 2010

Candid Advice on Taking the PhD Qualifying Exam

One of the rites of passage for earning a PhD in engineering is taking an exam that determines, at an early stage, whether a student will be allowed to continue in graduate school.  The goal is to determine whether the student has the necessary background knowledge, research capabilities, and communication skills to be successful as a PhD candidate before the student spends years of his/her life working on a research project, by testing whether the student displays the characteristics that typically “qualify” successful PhD candidates.  Usually, the qualifying exams are called “quals” for short.  Students who do not pass the quals generally leave the university with a Masters degree, and are often very successful as they move ahead with their careers.  You might say, in the cases where a student fails, the exam saves the student years of frustration and poverty, while at the same time saving their advisor from financially supporting someone who would have a difficult struggle with thesis research and publication.

Before giving advice on taking the qual in the Illinois ECE department, I should relate a little bit of history.

Every university and department has its own process for administering a PhD qualifying exam.  Some departments require the students to take a long written exam with topics selected from a set of core undergraduate and graduate courses.  Other departments require the students to give a short presentation on a research topic.  Yet other qual exams involve a panel of 4-5 faculty asking the student to answer verbal questions by solving equations on a chalkboard.  In many cases, an exam may have components of all three methods!  In ~2007, the Electrical and Computer Engineering Department at the University of Illinois at Urbana-Champaign made a transition from a 6-hour written qualifying exam to a ~1-hour oral exam.  In the previous written exam, the first 3-hour section was taken by every student, with questions all derived from core undergraduate classes.  In the second 3-hour section, students could select from a menu of technology areas, to receive questions close to their area of research.  The area-specific questions were typically difficult ones drawn from graduate level classes. 

The score from the exam was just one component of all the things that the department faculty would consider when determining whether a student would pass or fail.  After all the exams were taken and scored, the faculty would hold an evening meeting (lasting several hours) to discuss every student who took the test, going in alphabetical order.  A student’s class grades, exam score, and research progress (based on verbal comments from the thesis advisor) were all considered, and each faculty member would vote to either pass or fail the student.  While the exam score was not all-determining, a student in the bottom third of the scores would be at risk of failing, since the low score would lead to a discussion about whether the student truly had strong fundamental knowledge.  Although no set number or percentage of students were designated to pass or fail the exam, competition was always very fierce, because no students wanted to end up in the bottom third of the scores.  For this reason, students would spend several weeks (in some cases the majority of a semester) doing nothing but studying for the quals.  Some advisors would allow their students to take the time to study by giving them little or no research responsibilities during the qual semester, but other advisors would not.  Regardless, the time that students would spend rehashing material from their previous classes was seen by the faculty as a drag on productivity, since students who were doing nothing but studying were still being paid their research assistantships from grant sponsors.  One safety net that was (and still is) in place was that a student could have up to, but no more than, two chances at taking the qual.  If a student did not perform well on the first try, the faculty did not feel too bad about failing them and making them prepare better the next time.  If a student did not do well on the second try, there would be much more discussion to determine whether failing them would serve the best interests of the student and the department.  On the downside, students who did not pass on the first try would end up studying twice, resulting in almost an entire year of research being lost.

After several years of discussion and debate, the ECE faculty came to a decision to change the qual exam format from a written exam to an oral exam.  The thoughts behind this decision were that we already measure students’ fundamental knowledge through the grades that they earned in their classes, and that the written qualifying exam was redundant.  Also, the time that students spent studying was seen as a drag on their research productivity, but that students who were given time to focus on study had an unfair advantage.  Finally, because the purpose of the qual was to determine whether the student had the necessary abilities to succeed as an independent researcher, it was felt that a research presentation (in which the student prepared a written paper, gave a short presentation, and answered questions from a panel of faculty) would more accurately measure the students’ research capability.  I am not trying to make a statement about whether the new method is better than the old way – I am just trying to explain how the faculty arrived at this method and to give background that will help students to focus their preparation efforts.

The verbal qualifying exam is not without its problems.  Students with poor English language capabilities (or speaking/presentation skills in general) faced a new challenge.  Because the panel of faculty examining a student would be different for each student, there could not be absolute uniformity of verbal questioning.  Some faculty would be much more rigorous than others, so some students would get asked a greater number of more challenging questions than others.  While some students had some experience with a research project by the time they took the qual, others did not.  Students with more thorough research experience could draw upon material that they had developed through working with their advisor, that perhaps they had already had time to polish from a scientific journal article or a conference presentation.  To adjust for these differences between students, the ECE department has been instructing faculty more thoroughly about how to consider students in these different situations, and to provide some guidance (to students and faculty) about the type of knowledge students should come prepared to answer questions about.

What happens after a student gives their qual presentation?  The faculty on the exam committee meets immediately afterwards to discuss the student’s performance, and answers a series of questions about their background knowledge, handling of questions, the clarity of the written report, presentation skills, and specific knowledge relevant to the research.  Each component is scored on a scale of “Excellent,” “Good,” and “Needs Improvement.”  The faculty member who chairs the exam committee writes a consensus evaluation that summarizes the thoughts of all the faculty on the committee, which is reviewed and approved.  All of the written evaluations and scores are entered into a database, and all the information is available when the entire department faculty meets to discuss every student who took the exam.  In addition, the evaluation comments and scores are shared with the student.  As in the old system, the entire faculty meets to discuss each student’s case individually.  At the faculty meeting, the evaluation comments are available along with the grades (and class ranking) from every class taken by the student, and the student’s thesis advisor is required to make a verbal statement about the student’s research progress.

So how should a student navigate this process to maximize their chances of passing on the first try?

The process of preparation actually should begin as soon as a student enters graduate school with the intention of earning a PhD.  A student cannot earn a PhD (or even register to take the PhD qualifying exam) without a research advisor.  This means that a student must find a faculty member who is willing to advise them and (ideally) to provide financial support in the form of a Research Assistantship (RA).  This process is not trivial, and is so important that it should be a new graduate student’s highest priority.  Many graduate students enter the program with a research advisor already lined up, but this is not the case for everyone.  If a student has financial support from a Teaching Assistantship (TA), they will not always have an advisor.  In a later blog, I will write more about the process of finding an advisor, but why is it important to have an advisor for the qual?  The answer is that if a student begins working on a research project and learning the background knowledge to perform the research effectively, they are developing the skills and presentation material that will get them through the qual with flying colors.  If a student is working with a research advisor from the beginning of graduate school, they are effectively preparing for the qual from their first semester of graduate school.  If a student joins a research group in the same semester that they take the qual, they may only have had ~2 months to build this base of knowledge.  If you are a student in this situation, it is still possible to compensate, as I will discuss later.  However, in my opinion, if a student is actively working on a research project before they take the qualifying exam, it provides a tremendous advantage because they have something to write and talk about.

Secondly, I cannot overemphasize the importance of maintaining good grades.  Remember that at the meeting where your case will be discussed, every faculty member in the department will see the grade and class ranking for every class you have taken at Illinois.  Therefore, even if your instructor for graduate-level courses was an easy grader, all the faculty will still know how you performed compared to everyone else.  In graduate level courses, you may find that grading is on a different scale than what you may have been used to.  By graduate school, all of the students are already excellent, so most grades are A’s and B’s.  Earning a C in a class is almost equivalent to a failing grade, reserved for students who clearly were not doing the work or preparing for exams.  It is rare for a student with excellent grades to fail the qualifying exam, but students whose grade point average is 3.5 or below (earning half A’s and half B’s) will be scrutinized more thoroughly.  When a student is up for discussion, faculty generally know how the student performed in their class.  Was the student someone who participated actively and asked a lot of good questions in class, someone who showed good intellectual curiosity at office hours, or was he/she someone who rarely attended?  These types of things always come up, and help paint a picture of the student.  While it is not necessary for a student to have a near-perfect GPA to pass the qual, it is not wise to do poorly in class at the expense of spending every waking moment in the lab.  It is truly important to achieve a balance.

The new written part of the qual, in which the student is required to prepare a short paper describing their research, is your chance to show that you have been studying the background literature surrounding your research topic, and gaining a solid understanding of the important issue that impact your work.  The exam committee members will have read your paper before the oral exam, so the paper serves to introduce yourself to the committee before they even meet you.  The oral presentation will generally follow the outline of the paper, presenting the same information, so the paper serves to help students with poor presentation skills, since there is no time limit on preparing a good paper.  However, there are a number of pitfalls to avoid.  Make sure the paper is logically organized, written with proper English, and that all equations and figures are clearly easy to understand.  Student’s thesis advisors are not really supposed to edit and rewrite the paper, but it occurs occasionally.  Usually the committee can tell when something has been written by a professor rather than by a second year graduate student.  Instead of asking your advisor to edit your paper, ask a senior graduate student in your group, or a student who is a strong writer to proofread your paper.  Remember that some of the faculty in your area will be totally unfamiliar with your research area.  It is extremely important to explain things clearly, as if you are trying to educate someone who is learning about your area for the first time.  Errors such as failure to define variables in equations, using highly specialized jargon, failure to label axis on graphs, and failure to put size scale bars on images will drive your committee crazy, and you will always get negative comments for doing these things.  There is a page limit on length of the paper that should never be exceeded for any reason.  More is definitely NOT better.  You may think that you are showing enthusiasm by having paper that is three times longer than the page limit, but what this really shows is a lack of ability to describe complex concepts in a concise manner, and an inability to follow directions.  Your advisor may be asked how much of the paper is the student’s own work.  I have seen instanced in which the text of a student’s qual paper was lifted directly from journal papers from the thesis advisor’s published papers.  You should avoid doing this – the paper should represent your own words, and any technical descriptions taken from other work should be referenced.

Next, let’s discuss the presentation itself.  Even though this part is what students worry about the most, it is really only ~1/5 of everything that goes into the exam.  Just like the paper, the presentation should be logically organized (with an outline, and sections like “Introduction,” “Background,” “Methods,” “Results,” and “Conclusion.”)  Just like the paper, all the vugraphs should be clear to read (hint – use large fonts – your professors are old and can’t see that well!), all the variables should be defined, all the axis should be labeled, and all the images should have scale bars.  The presentation is meant to take 20 minutes, so you should definitely not try to deliver 60 slides in that amount of time (I have seen some students try).  You should PRACTICE.  A lot.  In front of other people.  If a student has given a conference presentation on their research, they often use the same material in their qualifying exam presentation, but for some students the qual may be the first presentation that they have ever delivered.  The members of your research group are extremely valuable resources because they can serve as your practice audience, and point out areas where your presentation was not clear.  It is also a great idea to have your friends throw questions at you, to give you the opportunity to think and answer on your feet.  Even though your talk should last for 20 minutes when it is not interrupted, the faculty on your committee are definitely going to ask you many questions, resulting in you presentation lasting for approximately 60 minutes.  The committee expects you to know about everything that is in your presentation.  If you present an equation that you do not completely understand (for example how to derive it, or its physical meaning) or if you describe background information that you have not thoroughly studied, you are asking for trouble.  Favorite questions by the faculty might be “tell me more about that technique you that the competing research group is using” or “where does that equation come from” or “what would happen if …”  These types of questions are meant to determine whether you truly understand the material that you are presenting, or whether you just copied it down from somewhere.

A word of advice for students who may have been working with their advisor for only a short time.  It is perfectly acceptable to mention this fact to your committee as you begin your presentation, by way of introducing your self.  You might start your presentation with a statement like: “Hello, my name is Joe Genius.  Thanks for serving on my exam committee.  I started working with my current thesis advisor, Prof. Andrea Wickedsmart, at the start of this semester.  Prof. Wickedsmart’s group does research in the area of fabric based invisibility cloaks, and my project will be to more fully develop the fabrication processes for building optical nanostructures into the fabric.  In my presentation today, I will briefly summarize the current state of knowledge around these invisibility cloaks, and describe the plan that Prof. Wickedsmart and I have developed for fabricating the nanostructures over large surface areas.”  This kind of statement (with the names changed of course) will serve to set the expectations that the committee will have for your presentation.  You have not been working on the research project for long enough to have new results of your own, but you are going to show them that you understand the current state of the art, some of the fundamental theory of operation, and that you have developed a plan for your research going forward.  To prepare, you should be studying journal literature very intensely, and asking your advisor (and senior graduate students in your group) a lot of questions.  Your classwork hopefully would be serving to build your background knowledge to help you understand the literature around your research topic.

The way that you handle questions about your presentation is probably more important than the prepared part of the talk, and this is where the greatest danger lies for failing the qual.  Even though you are giving the presentation about your research project, the committee is going to try to find out how you translate material that you learned from class into your research.  Some of the members of your exam committee will be faculty who know your research area very well, or they may be faculty who taught one of your previous classes.  It is extremely important to be able to show the committee that you retained and can use the most basic principles that you learned in class.  If, for example, a student is working on a project that relies heavily on the use of electromagnetic theory, it is not uncommon for the committee to try to find out how strong the student is in electormagnetism.  You might be asked to write one of Maxwell’s equations on the white board, or to write the equation for a propagating electromagnetic wave.  Anyone who is working with electromagntics should, in principle, be able to do that!  You might be asked how you would analyze a particular situation (such as what happens when an electromagnetic wave encounters a boundary between two dielectrics, or what happens when light is scattered inside a tissue).  The questions are usually tangentially related to the research topic, but emphasize the basics.  It is best to be able to not just wave your arms, but to be prepared to write down from memory some of the basic equations.  Many students fail to do this, and the faculty always shake their heads and say “I can’t believe that students working in this area can’t remember the (fill in the blank) equation.  It is fundamental to everything they are doing!”  Students working on transistors get asked transistor questions, while students working on lasers get asked laser questions.  There is no way to know in advance what question you will be asked, so this is where it is important to study and to be prepared in advance.  Being able to recall and apply even a simple concept from your former instructor’s class will make them incredibly happy.

Often, the committee tries to find the boundaries of your knowledge, to discover what you DON’T know.  In this case, the committee might lead with a simple question, and then start diving into more and more detail.  You should not be nervous about this, but just do the best that you can.  If you don’t know, it is better to say that you don’t know, or to outline how you would approach the problem, rather than to try to fake it.  Usually, the committee will give hints about how to think about it, to see if you can make connections to the next step.   It is definitely not a good idea to say things like “that question is not fair,” or “you can’t be serious” although I have seen that happen.  In one instance, a very senior faculty member asked a large number of questions, starting out with deep ones, and then making the questions simpler and simpler until the student could answer.  I could tell that the student was totally flustered, and was so nervous that he could not think clearly.   Even though the exam seemed like a disaster, after the exam was over the faculty member said “I thought he did pretty well!”  Remember that it is normal to be asked questions that might be beyond what you know.  If you are strong in the fundamentals and can show it, you will fare well in the scoring.

So as you can see, there is unfortunately no magical easy way to pass the quals.  Performing well requires a great deal of preparation, some long-term dedication to your research topic, and the ability to display that you have retained some fundamental principles.  The exam also requires students to work on learning how to write clearly and how to prepare an effective presentation.  I believe that it really focuses students to work on developing the same skills that they will need to be a successful researcher.