velua

Fair enough. I kind of doubt program coordinators would say anything useful (pretty much any program FAQ will say "we don't pre-evaluate applicants"). Professors could possibly be helpful at your target institution if you make a good impression first, and your LOR writers would probably say "yeah, you have a chance with your current scores," almost regardless of the truth.

...or make the decision yourself, and don't trust the opinions of others. If they're wrong, it's not their career that takes a tumble. If your LORs are as strong as your stats, another $195 is nothing in the long run.

Scores look good. I see no reason to retake the GRE. If you're going to take an exam, just take the GMAT. I have no idea about business school, but people say the GMAT is harder on math, so I'd imagine taking the GMAT would look better (assuming you can do well).

You can get a pretty decent idea of GPA, GRE, research, LORs, and so on by looking at the results page or even the yearly application thread on here.

Yeah. All your other stats look very impressive, it would be a shame not to get in because of that. Try to at least hit 150V (155+ if you can), and increase to 4.5A if you can.

In math (and possibly other fields), your first order of business is to pass the quals. You then start attending seminars, looking for advisors, and starting research. It is possible to go to seminars and start some small research projects before then in many cases, but it's not your priority. This might take 1 or 2 years to get to that stage. Keep in mind first-year classes are often intended to prepare you for quals, so getting good grades and passing the quals tend to go hand in hand. Eigen: I was in a math Ph.D. program, my interests changed, now I'm getting ready to apply for CS M.S.

GPA of 3+epsilon (usually 3 is the minimum) is really not ideal. Yes, it's a bad idea to stress out about getting a 4.0 to the detriment of your research. But if you keep a 3.1, then you have very little room for error in case you have a bad semester. Academia is a nasty game, and if you go on probation, they'll try to get you to leave with a Master's if they don't really like you. It's not hard for a professor to give you an A- instead of an A or a B+ instead of an A- if that's what it takes to keep you from getting above the 3.0. If they do like you, they'll find ways to keep you around anyway. There are other reasons to keep a high GPA as well: you don't want to apply to fellowships with a 3.3 grad GPA for example. Summer support is very competitive and TAs are often chosen on factors such as GPA and reasonable TA ratings. Bottom line: try to stay above 3.7, at the very least above 3.5. It's not worth the risk or stress of dropping into the 3.0-3.3 range.

Everything is relative. In STEM, < 165 might be "bad" for Stanford or Berkeley or MIT, while < 160 might be bad for schools like ASU or Utah. If Stanford receives 1500 applications for 100 spots, and let's say 1000 of them have strong LORs and research experience, the easiest thing to get it down to a reasonable number is to filter by GPA or GRE. I think passing the GRE filter is even more important at top schools. Why take a promising candidate with bad scores when they can take a promising candidate with good scores? That's why I think having a "decent" GRE score is necessary (but obviously not sufficient).

It's usually standard policy to kick people out for being on probation too long and not raising their grades. I think I learned a lot from college, not just how to take exams. But when I was TAing at a large public university, I saw how everything was about exams and not learning. This was forced by having so many exams and quizzes, and the students often had poor attitudes too (if I had some extra time, I might mention something interesting that was extra, and the most common question was "is this going to be on the exam?"). It's a sad truth. Many of my friends from college who went on to grad school elsewhere told me basically the same things.

I don't think being good at tests should be a necessary criterion for incoming graduate students either, and it's not that big a deal in general if you get a "bad" Q score (whatever that means), but it is alarming if you get a "bad" Q score as a student in math, CS, physics, engineering, etc. Presumably all of these students have taken a calculus sequence, which is a pretty computationally intensive subject area, and all of these students have the opportunity to review the material and practice. I know the GRE doesn't test calculus, but knowing your way around a triangle and being able to solve linear equations or inequalities is a pre-requisite for calculus. As you say, standardization is somewhat important (especially if you come from an unknown program), and tests can provide that. Subject GREs are obviously better for this than the general GRE, but not all students take them. Exams are not the point of graduate school, but students who are dishonest with themselves about their strengths use "bad test taker" as an excuse, and it gets old. I hear it with the GRE, and I hear it from people who do poorly on the math subject GRE as well ("oh, I was busy proving things, I can't believe I'm supposed to know calculus"). I'm sure people also complain about the other subject GREs. That excuse won't fly if you mess up exams in first-year grad classes and get placed on probation for a poor GPA, and it won't fly if you don't pass your quals and get downgraded to a Master's student. My views are also geared toward the more selective schools (let's say top 50 or so). Bad GREs and GPAs are more forgivable at less prestigious schools. If we get rid of tests, how should applications be evaluated at, say, Stanford, where there are so many applicants with strong LORs and research experience? Should they just throw away applications without a LOR from someone well-known?

Being imperfect is fine, I've made my share of mistakes on the board as a TA, but consistently making computational errors when all you do in your recitation is problems on the board is a significant issue. If an experienced instructor who's taken the class and TAed it (possibly) many times can't do these problems without mistakes, then the students won't respect you and won't care about doing the math carefully. One would like to believe that we train our students to learn or think or be independent. Although that notion is romantic, it is far from reality. The fact is, exams take priority over learning in classes, especially when you're talking about the huge public universities, where the basic classes have 3 midterms, a final, and a quiz in every non-exam week. There's just no way for them to leave the exam world and try to learn as one might do without tests. It sounds like you believe getting good at mistake-free computation will not improve your ability to do well on the GRE Q section, and I hope you're joking. Double checking things is great, but when a student asks you to do a problem you didn't have prepared in advance, and you can't do it without messing up the calculations, that is a problem. You keep on saying the GRE doesn't measure what you need for success as a grad student, and I have not disputed this. It's a fairly useless test, except possibly as a weedout for those who really can't do basic computations (which may be important as they're likely also recruiting you for cheap TA labor).

"Your analogy of driving and baseball does not work here because while following directions correctly is an essential skill in navigating and while hitting fastballs is a useful and desirable skill in baseball, answered timed arithmetic questions correctly is not a useful/relevant skill in STEM research." Hitting fastballs is what baseball players (non-pitchers anyway) train to do, just like taking exams is what students train to do. You claim that the GRE is not predictive of success in graduate school, and I agree. Being able to take an exam is necessary, but not sufficient. By the way, arithmetic skills are important. I've met countless grad students TAing calculus who are terrible at calculations and just don't care to improve. Not a good attitude to have. Knowing all the theory in the world doesn't make you a good teacher (which is another thing grad school trains you for) if you can't do problems on the board without making computation mistakes. "If the quant section of the GRE had 10 more minutes, I think 170 would be like the 70th percentile" Isn't it already 40 minutes for 20 questions?

Actually I've been in grad school, taken many grad classes (in multiple fields), and taken quals before. It's true that many grad classes have no exams, many of them have exams that are very different from undergrad classes (like oral exams), but many of the pre-quals classes have exams that are basically the same thing as undergrad exams, perhaps a little bit harder. Quals are more difficult than exams for these classes. I'm not suggesting that these exams are similar to the GRE, or that a strong GRE score is predictive of strong performance on exams in grad school. What I'm saying is that for an easy exam which you can take (almost) as often as you want, it very quickly gets old hearing an excuse of "bad test taker" for someone with a weak GRE score, especially when you've studied the material at the high school and college level. If you can't figure out how to find the angles in a triangle, then I have a hard time believing you can pass a qual in, say, real analysis. The fact is, math education in the U.S. is very weak. There's so much hand-holding and rewarding people for rote memorization and giving out partial credit, even if you don't actually think, understand, and solve the problem. This is especially true in high school and to some degree in college, as well. If you know the formula for the product rule but can't actually compute a derivative, there's something wrong. If you're out driving, you take a wrong turn, and follow the rest of the steps correctly, you shouldn't expect to arrive at the correct destination. Everything you say about Q scores not mattering beyond a certain threshold and V scores allowing one to stand out I agree with. But the one thing we're trained to do as students is take exams, so hearing excuses for poor Q scores (whatever "poor" is defined to be) is about as interesting as hearing excuses for why Ryan Howard can't hit a fastball.

Agree to disagree. We've spent 4 years taking exams in college, 4 years before that doing the same thing in high school, and some number of years before that doing the same thing (maybe they were tests or quizzes, not "exams"). You're absolutely right that doing well on exams doesn't necessarily translate to doing well in research, but you're still going to have classes with exams your first couple years, not to mention quals. Plus, recruiting graduate students in STEM fields, especially math, is a good way to hire cheap labor to teach the huge intro classes to all those freshmen.

3.2 GPA seems a little on the low end, especially for international students. The standards are typically significantly higher for international students.

I'm not saying they're the same, only that you can get a good score on the GRE without too much work if you have the quantitative ability to do well in a calculus course. In fact, I'd say my experience TAing calculus for several years got me very good at doing computations quickly and with very few mistakes (the point being that if you learn the material well -- not just well enough to get a B or even an A, but learn it well enough to explain to others -- it should translate to being able to perform well on the GRE with a little preparation). This translated into high scores for me, both on the math section of the general GRE and on the math subject GRE.

I didn't really intend either interpretation, and certainly not the second. What I meant was one can train for the test, and do so pretty quickly on the math side, especially with a couple college math courses under one's belt. It may be a different style test than a freshman calculus exam, but if you can get a B in calculus, you can get a 165Q IMO, as long as you're careful and double-check your answers. Getting a 170 may be a little more challenging, but that's not really necessary. 160Q is 78% according to ETS, but I wonder what percentile it is among STEM students. I'm sure it's brought up a bit from all the humanities (and many of the social science) students. Anyway, my views are more geared toward the more selective schools. I'm sure 160Q would be fine to get into Auburn. I agree with this: To answer the question, I think the writing is the least important. Quant is clearly the most important, and in fact, your agency said to get a minimum of 320 = Q+V, not a minimum of 320 and 4A (or 3.5, or 4.5, whatever the number would be). For the schools you're interested in, I think 320 minimum is a bit high (unless you get 170Q).

I don't see why "being bad at tests" is so forgivable for the GRE. That's what we've been doing for 20 years of our lives in school (or at least 10). You learn stuff, prepare for an exam, and take the exam. Getting a 160 requires more than a few mistakes. It's indicative of a deeper problem, or that you should re-take the exam if you really just couldn't focus or whatever. Especially if you come from a weak school where in calculus classes, everyone gives you tons of partial credit for knowing, e.g., the product rule, but being incapable of computing derivatives without making mistakes, these exams can be somewhat of a reality check if you're that bad with numbers (really, subject GREs are better for this purpose). I agree that the Q section really provides no new information, but it's alarming at the very least if a STEM student with a 3.8 GPA (or 3.6, or whatever) only gets 160Q, assuming it's a top 50ish school anyway. And the general GRE has little value, but a 160 is a pretty bad score for anyone who's taken a calculus class.

This is true, but I'm sure one of the GREs could be helpful for many disciplines where it's not required. For example, some elite schools in CS mention it could be helpful to take the subject GRE in math or physics if you have the background. In TakeruK's scenario, I'd go with person C to be honest (assuming all else is equal). The Math part of the general GRE is so basic that anyone in STEM should be able to get a 165+ with a day of preparation to be honest. 80%+ is roughly 159+ on the verbal part, which is far above what STEM programs will care about. They'll likely be happy with 153+, and certainly with 156+. Back when I was applying to Math Ph.D. programs, it was pretty much assumed you should get 800 on the Math section. If you got 780, you were probably still okay (but maybe not for top 20 schools), but if you didn't break 750, you should really retake it. As long as you got 500+ on verbal and 4+ on writing, you were probably in okay shape. Of course it's desirable to score higher, but it's unlikely it'd make much of a difference in your applications. Now for STEM fields outside of Math, there's a bit more room for error on the Math section, but you should still comfortably break 90% IMO. Edit: For the schools mentioned (Florida Tech and Auburn), unless they're highly ranked in some engineering field (which would surprise me), hitting 160Q, 155V, and 4A would probably be safe.

Interesting perspective. I guess you can say Q Is least important as long as it's above a certain number, whether 160 or 165 or something else. But a 145V is more forgivable than a 145Q (for STEM fields). I think V is more important than the writing, but don't think it should be. So much of it depends on vocab, and if an engineer has never seen the word jejune, that shouldn't be such a big problem. In writing, the essay where you explain what's wrong with an argument is not a bad exercise. In math, it could help indicate how good you are at dissecting a proof. Similar for other STEM fields, I'd imagine. The essay where you take a stance is the difficult one for me -- it's sort of hit or miss, depending on whether I can think of lots of examples to back up my opinion. I suppose it's debatable how important this essay should be. I think the most important section is really the subject test -- whether it's math, physics, or something else.

Many schools have January deadlines. Even if it's in December, I doubt they start reviewing applications before the new year. Can you find a school somewhere else in the country for a summer class, or do you have work or family obligations? When I was an undergrad, I took linear algebra over the summer. It was just a couple months so I packed a couple bags and hopped on a plane.

I agree, and physics sounds like a semi-related field, so it's fine. I would be hesitant to get a letter from an English professor for a science program, for example.

I said the Math subject GRE, not the Math section of the regular GRE. Everyone knows the Math on the regular GRE is a complete joke.

I understand. Well I'd just focus on that and try to find a linear algebra class this summer. If not, maybe you can explain why you messed up that class in your SOP or LORs (if there is a good reason). If not, just see what happens. If you took it your first year (or even 2nd year), it may not be that big a deal anyway.

B+ isn't so bad. I got a B+ in Calc3, and also in Diff Eqs and I got into grad school with that. Granted, I got pretty much straight As in tons of upper level stuff, including real analysis honors and differential geometry. I also TAed Calc3 one semester during undergrad. The C+ is a problem. By crush the GRE, do you mean the regular GRE or the subject GRE? If it's convenient, take linear alebra somewhere this summer. It shouldn't be too hard to find a school where it's offered this summer. Don't offer to retake it before the start of grad school. Either take it this summer or don't. The biostats students I've met seemed to have a pretty weak knowledge of abstract algebra, real analysis, topology, etc., and had never taken courses in them. I'd just try to take linear algebra somewhere else this summer and focus on the rest of your app.