Is the GRE Physics a sensible test? What does it test for?

[nonequilibrium]

Hello,

I'm currently at the graduated undergrad level concerning physics and I'm considering eventually doing a PhD in America. Usually this requires a Physics GRE test, which is meant to check whether you have absorbed the necessary knowledge from the undergraduate studies. Good. Well, I checked a practice test of Physics GRE and conceptually it all seems pretty straight forward, not hard, but what strikes me is that it requires a lot of speedy calculations, which I don't feel very comfortable with. Why does the emphasis seem to be on being able to calculate things quickly? Sure, you should be able to calculate simple things rather quickly without much effort, but still, the time allotted combined with the number of questions and the type of questions seems to imply you have to be quite quick on all of them. Isn't it more important to understand what's going on and know what you're doing than quickly producing the quantitive result?

I don't want to be as arrogant as to argue with the test, I'm sure lots of people responsible for the test have thought over what they wanted to poll with the test, I just wanted to check whether my impression is correct and why the emphasis lies where it does.

 

[Vanadium 50]

The GRE isn't perfect.
Grades aren't perfect.
Letters of Recommendation aren't perfect.
The sum is better than all of them, but still, not perfect.

 

[nonequilibrium]

I had kind of expected such replies, but they're not really helpful. My question was why the emphasis was on being able to quickly do calculations (and whether my impression of this emphasis was correct in the first place).

 

[Nabeshin]

A good portion of the questions aren't calculations at all, but rather the ability to quickly rule out wrong answers. Things like taking limits, dimensional analysis, etc. are very helpful (but of course there are many questions where these will not narrow it down enough). The format makes a lot of sense somehow, because what they are trying to do is test the entire undergraduate physics curriculum. You can't do that in three hours using the kind of in-depth exams you get in your physics classes. If you want to have that kind of depth, you have to move to something more like the graduate school quals format which can span over many days, and this seems impractical to standardize and administrate for the entire country (and internationally!).

 

[nonequilibrium]

Aha Nabeshin, that makes sense. I admit that it's been a year now since I've looked at the practice test so perhaps my memory is serving me wrong. The ability to deduce the wrong answers makes sense, it's a quick yet sensible way to poll understanding.

 

[TMFKAN64]

I don't think anyone is really happy with the Physics GRE. In many cases, it rewards the wrong behavior... at least in my case, I found if I trained myself to skip questions as soon as I made an error or didn't know how to move forward, my score improved. Thinking bad, moving on good. It's a test that is made artificially hard by the time limit.

That said, it is pointless to complain about it. It is what it is. Life is unfair, accept it, and move on...

 

[nonequilibrium]

Again, I'm not complaining, I'm trying to understand...

 

[Obis]

In mathematics GRE there are some questions that can be solved with long computations, however, if you know some underlying concept you can immediately see the answer, with minimal computations. It might be the case that there's questions like this in physics too.

 

[AlephZero]

In mathematics GRE there are some questions that can be solved with long computations, however, if you know some underlying concept you can immediately see the answer, with minimal computations.

Multiple choice tests don't ask you to find the correct answer from scratch, and you don't get any marks for showing your working or setting out a clear logical argument. They ask you to select the correct answer from a menu of choces.

If you want to do well on any test, learn the skill which is being tested. Whether you "ought" to be tested on a different skill is different issue - but being able to quickly spot wrong answers in real life situations is a useful skill to have.

 

[nonequilibrium]

Thank you both for those comments.

 

[king vitamin]

I like to think of it as going through hoops for the grad schools. I think that anyone intelligent enough to succeed in grad school with enough ambition can use the tools available (practice tests), study hard, and make a good grade (>750) on the PGRE (I suppose excepting things like disabilities/severe anxiety etc). So, if the grad schools tell you they want you to do well on it, and you want to go to grad school, then you'll study hard for it and not bomb it even if it doesn't correlate to the work you'll eventually do. I personally don't know if I could have made a perfect score on it even if I took it 50 times, but I worked hard, made a high score, and it showed on my applications that I took it seriously.

As for specifics on what's helpful, everyone agrees that common sense to get rid of options is a good idea. I found a few times on the practice tests that I did an incredibly long calculation to get the right answer only to find that every single other answer was nonsensical. You can also usually guess if you get down to 2 or 3 answers. I guessed 3 times on my actual PGRE and looked up the answers afterwards - I got one right and two wrong, so guessing netted me and extra half-point than had I not answered.

You also discover tricks for doing quick and dirty pencil and paper math. For instance, it was easier for me to memorize decimal expansions for the reciprocals of the first several primes so that when I had to divide by a number, I could quickly turn it into multiplication (which I could to faster). I imagine this kind of stuff is different for everyone.

 

[Vanadium 50]

The GRE is 100 questions, and covers ~12 courses. So there's 8 questions per course - fewer than one per course week. If you keep the total time constant and ask fewer but longer questions, it becomes more of a crap shoot with respect to the questions: the scores develop a random component. If you keep the question count constant and go to longer questions, the test becomes more expensive to administer and grade.

 

[Choppy]

The way I see it, the GRE has 2 purposes.

The first is to establish a base level of competance in the subject. In this respect, as Vanadium pointed out, you have to ask enough questions to cover the core undergraduate curriculum and you have to do so in a manner that's reasonable to administrate.

The second purpose is to provide a common standard against which graduate school candidates can be ranked. In this respect you probably don't want a test that can be fully and correctly completed by even students who are well above-average. Here the point is to stratify the applicants, and a time constraint is one way to do this. Whether this gives you the stratification you actually want is debatable.

 

[twofish-quant]

Good. Well, I checked a practice test of Physics GRE and conceptually it all seems pretty straight forward, not hard, but what strikes me is that it requires a lot of speedy calculations, which I don't feel very comfortable with.

Be *VERY* careful with practice physics GRE questions. I've often found that practice GRE physics questions in some study guides are quite different from the ones that are in the real physics GRE.

Isn't it more important to understand what's going on and know what you're doing than quickly producing the quantitative result?

Maybe, but tests are a set of compromises. The trouble with tests that test understanding is that those can be tough perhaps impossible to machine grade. Also, you have an limited time to test everything in an undergraduate physics curriculum.

My experience with the physics GRE was that it wasn't a calculation test so much as the "spot the trick" test. Once you've figured out the point of the problem, the arithmetic wasn't bad, and going through a large amount of calculation was one sign that you've got the problem wrong (and that you should skip to the next one).

I don't want to be as arrogant as to argue with the test, I'm sure lots of people responsible for the test have thought over what they wanted to poll with the test

One other thing with standardized testing is that it can be painful to change the test. Even if there are known deficiencies, changing a standardized test is a multi-year project. You have so many systems that depend on test results, that making even minor changes can be painful.

 

[twofish-quant]

I like to think of it as going through hoops for the grad schools. I think that anyone intelligent enough to succeed in grad school with enough ambition can use the tools available (practice tests), study hard, and make a good grade (>750) on the PGRE (I suppose excepting things like disabilities/severe anxiety etc).

I'm not sure that's true, but that's irrelevant. If you happen to be great at graduate school, but can't/won't jump through the hoops to get in, then no one cares.

Something that you have to remember is that the world doesn't revolve around you, and the systems can be extremely unfair and perhaps even silly, but if you don't get accepted and someone else decent does, then no one cares that you didn't get in.

If you stink at the physics GRE's, you might still be a decent physicist, but no one cares about that. The point of the physics GRE's is that if you do well, then the odds are that you won't stink at doing physics.