| | | Student Opinions |
The Quals are a controversional process. Even those students who have done
well question both the necessity and relevance of the exam. We have
provided below a sample of student quotes taken from the 1998 Quals
survey. Quotes are followed by summary information about the student who
gave them. The format is [Year in Grad. School],[Year took
Quals],[Affiliated Lab], [Pass Status]. For example,
-1st year, 1998,
[ISL STAR], [EngPhys, Ckts, EM, Sig, Ctrl], 1st pass
is a 1st year grad
student who took the Quals in 1998, is affiliated with ISL and STARLAB,
and passed the Quals on the first attempt.
The whole process of Quals was a little disquieting in that it did
not seem to test raw ability, but rather was biased towards quick
responses and snazzy presentation. If that is the measurement basis,
then it should be publically stated. As it stands, it is my opinion
that the Quals process needs to be revamped to properly measure what
it claims.
- 1st year, 1998, ISL, 1st pass
Quals doesn't measure how well one will do exactly. Maybe it's roughly
correlated, but what it really is is a flow valve for letting people into
the PhD program.
- 1st year, 1998, CSL, 1st pass
The Quals are a useless test that was put in place because the department
admits too many PhD students and needs to weed them out. The Quals do a
poor job of finding out who is ready for PhD level work and who isn't, but
they've become such an institution that I can't imagine anyone actually
taking the effort to make Quals relevant.
- 2nd year, 1998, ISL, not yet passed
Although I can't say the process is unfair, but it really depends on
whether one is lucky or not; and I don't see any thing in qual relevent to
the ability of doing research. This is more like a "Jeopordy" rather than
a PhD qualifying exam.
- 1st year, 1998, CSL, 1st pass
I can't say the Quals are totally perfect, but they are better than any other methods I've seen in selecting talented people.
- 4th year, 1995, ISL, 1st pass
Short answers in time pressure cannot possibly be a good measure for
research skills and therefor should not be used as a quallifying measure.
- 1st year, 1998, CSL, not yet passed
Make life simple and have a standard written exam, one for each area of
interest. This way, there's no guessing about whether one will get
professors they want - they're guaranteed the same number of desired
questions as everyone else, and everyone is graded equally, as everyone
with a particular interest DEFINITELY gets the same questions as someone
else.
- 2nd year, 1998, ICL, not yet passed
Profs should not ask the questions that they have asked in the previous
years, that happened to me. The prof asked a question i had seen in the
previous years papers but did not solve and and also did badly in front of
him. Those who had solved it with their study groups must have done well.
- 1st year, 1998, ICL, 2nd pass
Maybe we should have high standard for admission instead of using qual for PhD entrance.
- 1st year, 1998, ICL, not yet passed
Sometimes prof. didn't ask questions in the area they were supposed
to. Actually, mmore than a half asked me about other areas.
- 1st year, 1998, CSL, not yet passed
Bell curving of each professor's score really threw me off. For example,
in Cox's qual, if you got below 8, you will be belled to 5!
- 1st year, 1998, ISL, 1st pass
I was disappointed that there were not more circuits and EM profs. This
meant I had to put down some professors from areas that I had little
experience (and received 2 professors from my last group).
- 1st year, 1998, other, not yet passed
My problem with quals is how the importance of the exam is hidden during the application and admittance process. Many people don't realize what it means until they get here - that fully 2/3 of the admitted class will leave, regardless of their ability or intent. That statment should be somewhere, and I resent like Hell the way its importance is downplayed to incoming prospective and first year students.
If EE wants to keep the exam, dual track admissions is the only way to go.
Otherwise they should implement objective testing. Yes, you can make an
objective test within the limits of what that means.
- 4th year, 1994, 1995, ICL, 2nd pass
A one-day shot cannot really tell about the long term research
capabilities of an individual. It does, however, tell about his/her
ability to tackle problems quickly and to keep himself/herself in good
shape under high stress which is good.
- 1st year, 1998, SSPL, 1st pass
1) The format definitely is advantageous for some students (including
myself) and not for others. 2) If quals claim to test students' POTENTIAL
to do a succesful PhD, how can anybody fail one year and do really well
the next ? Since one's potential doesn't change dramatically from one year
to the next, it seems that preparation (studying) is at least as important
as just the potential of a student for a PhD.
- 2nd year, 1997, SSPL, 1st pass
I think that it is very difficult to assess a student's capability in only
10 minutes. Some students often require more time to think through a
problem thoroughly before answering. Quals gives an undue advantage to
students who can answer quickly and present themselves well verbally. As
such, it does not seem to directly measure the research capabilities of
any particular student.
- 1st year, 1998, ISL, 1st pass
The quals test the ability of the student to respond to surprise
questions at conferences or to justify his work when talking with
competitors. They are not so much about mastery of the material
as about oral communication of the fundamentals needed for Ph.D.
work.
- 5+ year, 1994, STAR, 1st pass
I believe that luck contributes significantly to whether or not you pass
the EE quals. It depends on which professors you get, whether or not you
studied for the particular areas which the professors question you on, and
so on. Although the current quals process tests your knowledge base, it
does not test your research capabilities, a significant part of
Ph.D. level work. I think the department would do better to emulate other
departments and judge students based on research that they have done. Of
course, I realize that this would mean a drastic overhaul of the EE
graduate program and may not be realistic.
- 3rd year, 1998, ISL, 1st pass
I don't remember if this is done, but the lowest score should be dropped.
Also, there should be a way to protest a given score, since a professor
having a bad day could easily give someone an unjustifiably low score.
Students taking courses during quals should be given special extensions.
While it is commonly known when quals are, I've found that many faculty
are not sympathetic to students' workload during quals.
- 5+ year, 1994, CSL, 2nd pass
I think the format of the quals is great! Seriously. I'd much rather
have them this way then take 1 week of 3 hour exams like Comp Sci.
- 4th year, 1994, 1995, ICL, 2nd pass
More presentation skill than anything else. Uh, gee...let's see if I can
impress Professor X with my apparent knowledge of subject Y by using all
the latest buzzwords of the topic.
- 1st year, 1998, CSL, not yet passed
Stressful Process ... should keep refining the process to minimize stress
on students. Still feel 12 minutes is too much time pressure for
students. (Students: get ready to think on your feet) Should make a
better effort to inform students *before* coming to Stanford of the quals
process to help them decide.
- 3rd year, 1996, ICL, 1st pass
For new graduate students (as opposed to 2nd year, coterm, etc. students),
it is difficult to identify ten, much less 25 professors whom you know
very much about. Many examiners ask questions outside or unrelated to
their subject areas (classic example is Kovacs), and it is difficult to
choose an appropriate top 10 or 15 choices because of this. For almost
everyone I know, the last 15-25 choices were more or less blind guesses
based on name and subject area, with subject area not always being
accurate.
- 2nd year, 1998, STAR, 1st pass
Although I don't think that the Quals are a good determinant of whether or
not a student is capable of PhD level work, I do believe that it is a good
test to indicate whether a student is committed to doing a PhD or not.
I'm glad I passed; I wouldn't want to do it a second time
(especially since I think there's a lot of "luck" involved
too: the professors you're assigned to, the questions they
ask you, etc. I'm pretty sure that with the same knowledge
and same group of competing students I could have faild with
a different committee and/or questions.)
- 2nd year, 1998, ISL, 1st pass
The final scores are plagued with noise. Doing badly with only one
professor can change the outcome easily. Delta 5 points can translate to
delta 20 places in the ranking. The average score needed to be about 6 to
pass. It was previously reported as 5. There is a fair amount of luck
involved; given two different sets of professors, it is likely that there
will be two different outcomes.
- 1st year, 1998, ISL, 1st pass
I found the professors from CSL not very careful at their evaluations.
- 3rd year, 1996, ISL, 1st pass
The quals process, it seems to me, gives an advantage to those who are
more comfortable speaking in public, and those who perform better under
time pressure
- neither of which seems particularly relevant to getting a
Ph.D. degree, which is a slow, steady process. As such, it gives a
disadvantage to foreign students, those who are shy, etc. Old fashioned
ways of testing engineering knowledge
-- a pencil, paper, and about 3
hours of testing time
-- seem more reasonable.
- 5+ year, 1994, STAR, 2nd pass
I think one thing that is often hidden/ignored is that the Quals process is a head-to-head competition between grad students. The question is not whether a given student is good enough to do PhD. level work but, rather, where that student ranks compared to other students.
- 3rd year, 1996, STAR, 1st pass
I view the quals as a necessary evil given the way the department is
structured. An alternative would be to allow fewer people into EE grad and
not make them suffer through this process. But then, of course, the
university wouldn't make enormous amounts of money from the masters
program...
- 5+ year, 1994, ICL, 1st pass
Oral communication and thinking on your feet are essential skills in a
research environment. Quals test this. They should not, and I believe
are not, the only requirements for entering the PhD program
- 5+ year, 1994, ISL, 1st pass
As an international student, I felt miserable when English really did
matter! My thought on Qual is in-between yes and no. If you were ranked
very high, say top 10(or 10objection on it. However, students ranked 30th and 100th can never be
differentiated in terms of their RESEARCH ABILITY since some students with
a great research ability did actually fail in the exam and vice versa. It
is just once in a year. Some other factors should be considerd
also(project? written test? GPA?), or just don't admit that many M.S
students with Ph.D in their mind.
- 1st year, 1998, ISL, not yet passed
A PhD qualifying exam should be more like what they do in other
departments at Stanford, where the student prepare a thesis proposal and
has to defend it to a commitee. This, in my opinion is a much better gage
of the research potential of a student. It also gives feedback early in
the process as it forces the student with his/her advisor to focus on
selecting a feasible thesis topic early on.
- 5+ year, 1994, ICL, 2nd pass
I'm glad that "weighting" has been removed from the 1998 Quals... why should some students get bonus points because they were lucky and guessed their weightings properly?
When people ask me how hard it is to pass quals, I ask them, "How hard is it to get heads on a coin toss?" Everybody should know that Stanford EE just randomly picks half of the students to continue with the Ph.D. It's an arbitrary, guilt-free way of cutting down the class size.
I also make it a point to tell prospective students that the quals are evidence that Stanford admits too many grad students in EE, mainly to grab as much money as possible. It lessens the value of a Master's degree when Stanford EE admits people for their money, rather than admitting those students who they are willing and able to support through the Ph.D. Why should anyone be proud of such a non-exclusive degree, one which can just be bought? Why should the degree hold a good reputation in industry? Academic standards are too low for the Stanford MSEE - the Emperor has no clothes!
- 2nd year, 1997, ISL, 1st pass
I think the assignment process allows students to "sneak through" quals if
the choose the proper examiners and get lucky with who they are assigned.
What I am saying is the exam can be much easier or much harder based not
on the students skills, but on who the student happens to have as
examiners. It seems to make strategy a larger part of the test than
actually studying EE stuff,
- 1st year, 1998, SSPL, 1st pass
It is debatable whether the structure and content of all professors'
questions really matches the students with the best ability in reasearch
work. However, there is definetely a correlation (and I am not saying with
a correlation factor of 1.00) between well ranking students and ability to
think and do research. Unfortunately, many other factors come into play,
thus turning the process in potentially a suboptimal one.
- 3rd year, 1996, ICL, 1st pass
I think the Quals procedure should try to evaluate more skills in the
students, I don't think tenminutes with a professor can tell too much
about the mental capabilities needed for a PhD research. For example, how
deep can be any reasoning when you are requested to answer a question in
ten minutes under a lot pressure? Probably that's the the thing the Quals
evaluate the most, basically, how are you able to handle stress.
- 2nd year, 1998, ISL, 1st pass
I found the process incredibly arbitrary. Some of the people best
suited to research in my opinion did not pass.
- 2nd year, 1998, other, not yet passed
There are too many random elements in the process.
a) The choice of professors. No student can possibly know 25 professors, yet we are asked to pick 25.
b) The assignment of professors is totally random, and is not known until the first week of January. Consequently, a lot of preparation work goes waste because one can only realistically prepare well for a select few profs. and there is always a good chance that some of these select few profs. will not be on your commitee
c) after the selection of the committee. the questions are totally random. in the circuits area, one expects circuit questions, not a question about four boxes and their contents. in the software area, professors expected to ask about operating systems asked about some entirely different area
d) the grading criteria is totally random and varies from professor to professor, and question to question. for one question it is speed, for another clarity, for a third explanations, for a fourth just "common sense" and for a fifth just "a general impression that the professor has" based on your answer. it is impossible to tell what you are being graded on.
e) finally, the scoring process introduces more randomness. all scores are normalized based on professor. so professors on whom you score a ten, you could only end up getting a six or seven.the reverse could happen too. score a zero and get pulled up to a three.
f) the questions themselves are in no way capable of testing anything about a student's ability to do a ph.d. many of them are just content questions and if a person has worked in that area extensively or taken several classes in that subject he is sure to do better than someone who has not had an opportunity to take as many classes in that subject.
Some constructive suggestions:
a) Offer the exam twice a year. This is a reasonable request. Taking the exam at the end of the first quarter is quite an unreasonable demand on the students.
b) Separate financial aid offers particularly RAships from the Qualification process. A lot of students take the exam with no interest in pursuing a Ph.D. but only with the interest of increasing their chances of getting funded. Also, professors should have the freedom to pick whom they would like to have as their research students and not be forced to pick only from those who have passed quals.
c) reduce the number of areas to be picked. although we are asked to pick four areas, each of these areas has four sub areas at least. it is impossible to prepare for 4 x 4 = 16 areas throroughly. example: within software systems there is programming languages, operating systems, databases, graphics etc... students should only be asked to pick two areas, the first one the area in which they wish to do research and a second of their choice
d) get rid of the ten minute oral. no professor can make an accurate judgement of a student's ability in ten minutes and no student can give an accurate representation of himself and his capabilities in ten minutes.
f) get rid of the ten person commitee. make it three or five
professors who quiz the student for at least an hour and then
vote by majority on the student's capability to do research
- 1st year, 1998, CSL, not yet passed
Interactive exams are much better for finding out how someone thinks
-not just whether they've memorized the textbook(s).
- 1st year, 1998, SSPL, 1st pass
The normalization process is messed up. If a professor grades everyone
about the same, then doing slightly below average gives you an extremely
low score. Also, there were some professors where I answered everything
they asked correctly and completely, and I didn't do too well. I've
thought about those questions afterwards and I still can't figure out what
more I could have done. Finally, many of the professors seemed to be
measuring your knowledge of a subject area, not your thinking process like
the quals claim to do. For example, professor Cioffi just gave me a list
of stocastic signal processing problems and asked me to solve them. He
determined the score by looking at how many questions you got correct, not
by the thought process that went into the questions. Finally, I would
like to say that I know of people who are much less qualified than me who
passed the quals. I think that this is a very bad system for determining
who would be a good researcher.
- 1st year, 1998, CSL, not yet passed
My expectation of how well I did on a given prof. didn't correlate exactly
with how well I actually did, so I believe there is some
arbitrariness. Also, only 1 question did I believe my ability to do
research was being tested. All the others tested knowledge and I could see
no relevance to research ability.
- 1st year, 1998, CSL, 1st pass
Personally, I don't like Qual. I don't think you can judge people in only
about 10 minutes. I don't understand what we study all year long for if
what you want is 120 minutes with the committee. And about the
announcement of who you'll get for your committee, it is too close to the
exam day. Like I said, it depends a lot on each prof. If you don't know
who you're gonna get then how you're gonna prepare for it. My suggestion
is try finding something else to select students. Qual is like the most
absurd way of judging who is good and who is not.
- 1st year, 1998, CSL, not yet passed
| | | Student Opinions |