PhD Qualifying Exams -- Are they being phased out?

[Vanadium 50]

[Mentor Note -- this thread is split off the previous thread: https://www.physicsforums.com/threads/are-physicists-good-in-every-branch-of-physics.1049445/ ]

To get a PhD, you need to pass something called a qualifying exam, which covers all of physics. I do high energy, but there were questions on condensed matter, statistical mechanics, electromagnetism, etc. That's the starting point. (Sadly, a lot of places are getting rid ofr this)

Then depending on your career, you may need to learn about other things. Francis Halen (of Halzen and Martin) needed to become an expert in glaciology - the study of ice - to do the science he was interested in. Roland Winston (Winston cones) had to become an expert in non-imaging optics to do the science he was interested in. And so on.

 

[haushofer]

To get a PhD, you need to pass something called a qualifying exam, which covers all of physics.

Is this in the US? As a European I've never heard of this.

About the OP: some people are also good talkers. During my PhD I've experienced quite some people who pretended to know stuff by using jargon, but after some inquiry really didn't understand it as much as they pretended. As a young PhD candidate it's easy to be impressed and intimidated. Everyone has its own "weaknesses" in knowledge.

 

[Vanadium 50]

Yes, it is in the US (and at least some Canadian universities). But some places are getting rid of it.

 

[StatGuy2000]

To get a PhD, you need to pass something called a qualifying exam, which covers all of physics. I do high energy, but there were questions on condensed matter, statistical mechanics, electromagnetism, etc. That's the starting point. (Sadly, a lot of places are getting rid of this)

@Vanadium 50 , I'm surprised to hear that there are many physics graduate programs that are getting rid of the qualifying exams.

From what I know, qualifying exams are still the norm in graduate programs in statistics (my field of expertise). I believe qualifying exams are also still the norm in math graduate programs (both pure and applied).

 

[Vanadium 50]

This may better belong in its own thread:

There are three factors entering these decisions:

(1) In the past, there was abuse. Universities had the support for N students to do research, but needed 2N to TA freshman physics. How to reconcile this? A qual that half the students passed.

(2) There is competition for students, and students don't like quals. Hard to blame them.

(3) It is considered a barrier to diversity. I don't buy this ("are you really saying that women and minorities are less good at solving physics problems"?) but either other people do, or they are using this as a fig leaf for other reasons.

While the qual itself was a painful ordeal, I think preparing for it was important and useful.

 

[weirdoguy]

At my faculty (Warsaw University in Poland) there are qualifying exams at the end of every "step": bachelor, masters and PhD.

 

[symbolipoint]

This discussion about Qualifying Examinations is interesting. If Qualifying Examinations are eliminated, then what replaces them, as a way to ensure high-enough quality graduate students entering graduate programs?

 

[Office_Shredder]

.
(3) It is considered a barrier to diversity. I don't buy this ("are you really saying that women and minorities are less good at solving physics problems"?) but either other people do, or they are using this as a fig leaf for other reasons.

The standard barriers to diversity are all neutral sounding - sat tests being pretty classic (turns out rich people are better at hiring tutors who just teach them how to pass the test without materially improving the students academic prowess).

I guess they might come too early in the program - you're mixing legacy undergrad admits who have been pursuing STEM for 7+ years actively with someone who picked up physics 3 years ago, and then at some schools kicking out whoever does worse on a test. I can guess which person is favored to win on that test, not all the time, but certainly more than half the time.

 

[PhDeezNutz]

I think it is a terrible idea to get rid of quals. Quals to a large extent ensure that someone has command of “basic” material.

The diversity argument is ridiculous. Socioeconomic status will not get you through quals; people who can tutor you to pass quals are likely experts working in industry or at a university.

At my university they still have Quals.

 

[Office_Shredder]

I think it is a terrible idea to get rid of quals. Quals to a large extent ensure that someone has command of “basic” material.

The diversity argument is ridiculous. Socioeconomic status will not get you through quals; people who can tutor you to pass quals are likely experts working in industry or at a university.

At my university they still have Quals.

I don't think quals need to go away. I thought about this a bit more to try to nail down what my gut sense was trying to tell me.

The problem, I think, is going to lie mostly in schools that use quals to filter. You have a bunch of students enter - some already know 2/3 of the material on the quals, some know 1/3. You then give them a test that deliberately comes fast enough that people don't have enough time to comfortably learn it all (you need to fail half your class after all).

The people who came in knowing 2/3 of the material clearly have an advantage. Those are going to be people who learned more science in high school, so didn't have to work to pay their way through college, who went to fancier schools with a better set of physics courses to begin with.

The disadvantaged student? Obviously they have a chance. But they also have a disadvantage. Statistically they are worse off.

At the end of the day my model of the world is a bit speculative. But If a school says they ran the stats and quals flush out minorities more frequently, it seems weird to think they're wrong.

 

[Vanadium 50]

what replaces them

Two answers:

At some places, nothing. If there was a very high pass rate, why test at all? (The argument goes) In my class, plus the ones on either side, we had one person fail without a score enabling that student to try again, and one drop out before the second attempt. Out of 50 or so. So the pass rate was 96-98%, depending on how you count it.

At others. they make the argument that the final exam in the core classes serves the same purpose as the qual, and thus the qual is redundant and unnecessary..

 

[Vanadium 50]

The standard barriers to diversity are all neutral sounding

Sure, but you need to compare them to the alternative. When schools dumped the PGRE in response to Covid, the story was "we're worried about implicit bias in the test" - and then selected classes from Stanvard and snooty New England SLACs. Did that improve diversity? I don't think we have numbers, but it's hard to see where it would.

If you want to know if your cohort can solve physics problems, giving them physics problems to solve is probably the best you are going to do. It's sure going to be better than looking at Mummy and Daddy's tax returns.

The oral component is another matter. The written exam is uniform, but professors can ask whatever they want in the oral.

"Mr. Jones, please write down Ohm's Law"
"Miss Smith, define the universe. Give three examples."

 

[PhDeezNutz]

I will agree that quals hit you hard and fast. At my school you have to complete them within one year of being accepted to the PhD program. One year to solidify 4 years of material is a tall order. Especially when you have other things on your plate.

 

[Office_Shredder]

"Miss Smith, define the universe. Give three examples."

Our universe, miss universe, and the marvel cinematic universe. You can just mail my degree to me, dm me for the address.

 

[George Jones]

The oral component is another matter. The written exam is uniform, but professors can ask whatever they want in the oral.

At my university, there was only an oral component. Each exam was set by the Qualify Exam Committee (QEC), and consisted of 5 problems, one each from: classical mechanics; electromagnetism; condensed matter/solid state; thermodynamics/statistical mechanics; quantum mechanics.

Format. The student only had to attempt 4 of the 5 problems. To start, the student was left alone in the room (without books, cheat sheets, etc.) for 20 minutes. After this the, QEC committee came in, and the student started solving the problems on the board. If the student adequately solved a problem, then they could move on to another problem. If the student floundered on a problem, then the QEC would ask all kinds of questions about the subject area. After finishing, the student waited in the hall while the committee deliberated on the outcome

Three possible outcomes: 1) outright pass; 2) pass with weakness in 1 area; 3) fail. If 2), then, then the student had to compete and submit a report that sytheized the weak area. If 3), then the student got 1 more chance before being booted. Frequency of outcomes, from greatest to smallest: 2); 1); 3).

 

[Vanadium 50]

Mine was two days of written exam, orals on one of the two following days (where the first question was always "I see you missed problem X on the written - solve it now" and late in the afternoon on the fifth day the results were placed in our mailboxes,

Results could be pass, fail, or try again next year. Usually a student got only one additional chance. To be fair, to be close to the line and not able to bring it up that last bit must be really rare.

 

[Dr Transport]

My niece had to take her qualifying exam for her program in Plasma Physics at Princeton, which involves the Physics qualifier....

She passed.....

 

[hutchphd]

The standard barriers to diversity are all neutral sounding:

Of course they are (how could they be otherwise?). But that says absolutely nothing about the qualification exam: they obviously don't call it the descrimination intervention. But It is clearly a measure of capability in at least part of physics.
I found it useful to take a month and study as hard as I could to catalog the newfound knowledge in my head. It was a useful motivator for me, and the most compressed studying I ever did. I actually had physics dreams (very wierd abstract interludes!.

 

[Vanadium 50]

I found it useful to take a month and study as hard as I could to catalog the newfound knowledge in my head.

I spent a summer doing that and working every single problem in Halliday and Resnick. Every single one.

My strategy was not to lose points on the easy stuff, and a side effect was learning to be fast and accurate on the easy parts as well. Then I could focus on the tough bits.

 

[martinbn]

I read a biography book about John Nash. The PhD program in math at Princeton at the time had only one requirement, to produce a thesis. All else was optional. Nash didn't take any courses. This my not have been typical, but it suggests that all coursework, prelims, quals, candidacy exams, and what not were introduced later.

 

[Hornbein]

I read a biography book about John Nash. The PhD program in math at Princeton at the time had only one requirement, to produce a thesis. All else was optional. Nash didn't take any courses. This my not have been typical, but it suggests that all coursework, prelims, quals, candidacy exams, and what not were introduced later.

When I was a math grad student at Santa Cruz there were another student who was obviously too good to be there. After a year he transferred to Princeton. He said there was no course work there. If you need instruction then you are not admitted to Princeton's math graduate school. Indeed he appeared to know everything already. He said he had been the victim of a poor letter of recommendation.

 

[Vanadium 50]

Go Banana Slugs!

I think we all have WTH stories. My favorite is the story, possibly legendary, that Rudy Mossbauer was not allowed to present his work on the Mossbauer Effect for a PhD at Heidelberg. Just a Masters.

Standards, you know.

 

[gwnorth]

This discussion about Qualifying Examinations is interesting. If Qualifying Examinations are eliminated, then what replaces them, as a way to ensure high-enough quality graduate students entering graduate programs?

My understanding is that qualifying exams are typically taken after completing the requisite coursework of either a master's or integrated PhD, so they don't screen for students entering grad programs. That's based on performance and experience gained at the bachelor's level.

As to what would replace them aren't they somewhat redundant if you've already passed all the requisite coursework? Why do you need another level of assessment?

 

[gwnorth]

When schools dumped the PGRE in response to Covid, the story was "we're worried about implicit bias in the test" - and then selected classes from Stanvard and snooty New England SLACs.

Is there actual data showing that a higher percentage than the historical average of admits have come from these top tier undergraduate programs since the GRE/PGRE was suspended?

Btw love the term Stanvard :)

 

[Office_Shredder]

Is there actual data showing that a higher percentage than the historical average of admits have come from these top tier undergraduate programs since the GRE/PGRE was suspended?

Btw love the term Stanvard :)

https://mitadmissions.org/blogs/ent...act-requirement-for-future-admissions-cycles/

They didn't like, publish their data, but this is probably the best you're going to get.

 

[Vanadium 50]

Is there actual data

I don't think we have number

I don't think anyone collects this, so even getting a baseline will be tough. Juat anecdotes.

 

[gwnorth]

https://mitadmissions.org/blogs/ent...act-requirement-for-future-admissions-cycles/

They didn't like, publish their data, but this is probably the best you're going to get.

But that refers to the SAT/ACT for undergraduate admissions not the GRE/PGRE for graduate admissions.

 

[Vanadium 50]

Yes it does.
I don't think statistics are routinely kept - it's whatever AIP decided to do whenever they decide to do it.

Whenever I have been involved in the admissions process, what other schools decision process might be seldom is even mentioned, so it's hard to tell. All we have are anecdotes around a coffee table.

 

[symbolipoint]

As to what would replace them aren't they somewhat redundant if you've already passed all the requisite coursework? Why do you need another level of assessment?

Passing requisite courses does not always mean being able to pass tests for the Qualifying Examinations.

 

[PhDeezNutz]

Passing requisite courses does not always mean being able to pass tests for the Qualifying Examinations.

I agree. There are many posers who weasel their way throughout grad school examinations because they get old tests and know what to expect.

quals are much less predictable.

 

[Vanadium 50]

Passing requisite courses does not always mean being able to pass tests for the Qualifying Examinations.

I think this is an important point.

The qual is in a different format than finals - usually you have more involved written problems, more time to do them, and there is also an oral component.

Also, you can ask questions on the qual that probe combing what is taught in multiple classes.

You also don't want to pin the class requirements too closely to the qual., I didn't have a grad classical mechanics class, but I had Goldstein as an undergrad. I would argue a PhD physicist should know some classical mechanics - how do you check this?

As an example, I would say to compute the energy levels of the hydrogen atom in a weak electric field E in the z-direction is a fair qual question. It's not a good QM final question because it takes too long and requires more math than most hydrogen atom questions,

 

[Dr Transport]

Yes it does.
I don't think statistics are routinely kept - it's whatever AIP decided to do whenever they decide to do it.

Whenever I have been involved in the admissions process, what other schools decision process might be seldom is even mentioned, so it's hard to tell. All we have are anecdotes around a coffee table.

It used to be published in the AIP/APS book of graduate programs. I'm not even sure that exists any longer,. The closest is the AIP gradschool shopper, https://gradschoolshopper.com/browse/. In the few minutes I looked, I didn't any mention of average test scores, but then again, I only cursory looked at the featured programs. I did find a digitized version of what one
https://www.google.com/books/editio...s_in_Physics_Astron/v6hYAAAAMAAJ?hl=en&gbpv=0

I will say that departments keep track of where their students come from and if there is a trend amongst students from a particular school, they notice and use that info later, good or bad.

 

[gwnorth]

If it's true, that doesn't bode well for international applicants.

 

[andresB]

I will agree that quals hit you hard and fast. At my school you have to complete them within one year of being accepted to the PhD program. One year to solidify 4 years of material is a tall order. Especially when you have other things on your plate.

One year seems like a reasonable term for quals, though that depends on other considerations. I think it works better for a 5-year-long Ph.D. For the European 3-year-long Ph.D I don't think doing quals is good at all.

In my University the topics were the same for everyone independently of their research topic: stat mech, classical mech, quantum mech, and EM. Those topics were covered in two graduate-level courses each semester, so the students should be ready for the exams after one year.

 

[Dr Transport]

I will agree that quals hit you hard and fast. At my school you have to complete them within one year of being accepted to the PhD program. One year to solidify 4 years of material is a tall order. Especially when you have other things on your plate.

Where I went to grad school, you were accepted into the Masters program. After passing the quals, you were awarded your masters and accepted into the PhD program.

Ours had 4 sections, mechanics and stat mech, electromagnetics and special relativity, QM and a spectialty topic of your choice from solid state, optics, plasma or astrophysics. Two sections on Wednesday, the other two on Friday, 9-noon. Three questions of 4 for each section, so you had about half an hour per question. I studied for a year for them.