Hacker News new | past | comments | ask | show | jobs | submit login
GRE Scores By Discipline (arisbe.com)
50 points by Jebdm on Jan 6, 2009 | hide | past | web | favorite | 72 comments

Verbal and Quantitative have vastly different distributions. Simply adding them together favors those in quantitative fields.

I think that the smaller variance in verbal scores should have tipped him off.

Renormalized with equal weights. The only major change looks like philosophy jumped from 9th to 3rd.

  1	1.19
  2	1
  9	0.92
  4	0.89
  3	0.87
  5	0.7
  6	0.67
  7	0.43
  8	0.39
  11	0.36
  10	0.32
  15	0.27
  18	0.16
  14	0.14
  16	0.12
  17	0.08
  19	0
  20	-0.03
  13	-0.16
  12	-0.2
  21	-0.59
  23	-0.61
  22	-0.67
  24	-0.86
  25	-1
  26	-1.12
  27	-1.43
  28	-1.83

Yeah. This is true, and it's the first intelligent thing anyone has said in this thread. The verbal section is much more challenging. And it is not word memorization, although that is part of it.

I talked to the dept head in aero at MIT during the 90s and he told me that the verbal section of the GRE was the only correlation he'd seen to success.

Funny, I heard a rumor that the only people who got 800s on the verbal section were foreigners. (Read: people who can't speak English.) The explanation was that they were the only ones willing to invest enough time to form-fit the ETS's method of verbal testing.

In my own experience the verbal section of the GRE was entirely nonsensical. The vocabulary was difficult and exactly the type of vocabulary I would use if it was my goal not to be understood, but at least I could see how it might be a useful and reliable differentiator.

Nobody uses these words often enough for them have any statistically meaningful shades of meaning. If the goal of communication is having other people understand you, why would you deal in words that are notable precisely because they are difficult to understand?

The reading comprehension made less sense. I don't think it's possible to select a passage that has an unambiguous meaning and yet will be misunderstood by a statistically meaningful number of ambitious, college-educated people. So the question-making process must necessarily be corrupt.

I recall many questions which wasted much of my time. Several possible responses, none of which are obviously wrong. I know which one I feel is right---or maybe I do. Which one does the GRE feel is right? Even on sample tests, where I could see the supposed answer, I would read and reread the passages, and understand little about what the correctness or incorrectness of the responses had to do with the meaning of the passage.

I don't think you can presume to have an objective and clear-cut A or B answer as you might on the SAT. The granularity of college-level material is simply too fine. The only thing you can measure is how well you have matched yourself to the GRE's way of thinking, and that is not a particularly valuable expenditure of time.

"If the goal of communication is having other people understand you, why would you deal in words that are notable precisely because they are difficult to understand?"

One important value words have lies in their discriminatory capabilities; the more exact and fine-toothed your toolset is, the more precisely you can express yourself. Sure, it doesn't matter if you're wearing mittens on your vocabulary for most exchanges -- but it sure is nice to be able to take them off once in a while, and say exactly what you mean.

"If the goal of communication is having other people understand you, why would you deal in words that are notable precisely because they are difficult to understand?"

Once in a while you meet someone else like yourself and experience pleasure communicating at high baud.

Really? I take your point to mean that two smart people can communicate better using obscure vocabulary when they are communicating only between themselves. High baud necessarily implies higher average information throughput, so there would have to be a significant number of these words in the communication.

Could you point to some examples of letters between two famous people using this method?

Most notable people seem to prefer simple vocabulary, and in fact, many of them go so far as to recommend it. The people who prefer complex vocabulary (cultural studies journals, intellectuals) do not seem to accomplish anything of value, and indeed they are routinely accused of being obscurantists.

(Ground rule: scientific jargon excepted, since it is not a part of the vocabulary. But, feel free to look at letters by Newton, Einstein, Godel, Hilbert, anyone really.)

It has nothing to do with a preference for complex vocabulary.

When I talk to someone I know has a large vocabulary, I am free to use the best word that occurs to me, and if they think I'm up to snuff, they are free to do the same thing. This makes for a more pleasurable conversation. (Assuming we're actually trying to say something to each other. But if the motivation is to show off, or engage in ego jockeying, that's boring.)

I guess it's more accurate to say that baud rate surges transiently when a rare, but exceptionally precise word enters this sort of conversation.

Obscure vocabulary could be divided into two, one containing obscure synonyms of frequent words, and another is the chunking of information, of many words, to give one label. People familiar with the topic may find it easier to anchor their thoughts with these chunked words of phrases.

A lot of philosophy seems really ridiculous to me, and yet there must be a need to chunk concepts and give them labels. When you agree on the basket of concepts tagged with "logical positivism" or "paleo-conservatism" isn't it easier to just use that phrase and not repeatedly invoke all or the topically relevant concepts of that set?

Of course, the problem is when these labels are overburdened and/or so multifaceted that you could be talking about subtly different things and not get anywhere. This is why philosophy should be based not in the fuzzy verbal, but in the concisely mathematical, but that is another topic.

I can't point to any letters to showcase the method, but I'd point to political science as it is a place where verbal philosophy and practicality do coincide.

Or, more likely, if you are well versed in the vagaries of obscure vocabulary, chances are you'll be better at putting together the less complex words.

"I heard a rumor"

I'm a counterexample to the rumor.

Me too. Somehow I aced the verbal section, with a moderate amount of study, starting ~6 weeks before the test. I think it might be because I was raised without a TV and so read a lot as a kid. The questions at the end of the test (it's adaptive, computer administered) seemed to focus on the exact nuance of similar words.

And the jury's out on any correlation to my success.

99% percentile on verbal. Didn't really study for it though, just read a lot as a kid and had the fortune to have studied Latin.

Worth mentioning that my life doesn't correlate to success in any meaningful way, except that I keep making decisions which make me happy if poor.

Agreeing with both replies to me, I think the secret of my success on the GRE verbal section (twice) was reading a lot for fun. And agreeing with them about correlations between GRE success and life success, I still pursue a lot of personal intellectual interests but have not become notably wealthy, although any middle-class American is wealthy in worldwide terms.

I have friends who have done very well on both sections. They didn't practice much. They tend to be polymaths and read a lot.

Normalizing the distributions would not change the rankings by much.

The fact that every discipline uses the same test for admissions says more about the quality of academia than the scores.

"The fact that every discipline uses the same test for admissions says more about the quality of academia than the scores."

That is not necessarily true. Its a testament to how education is treated as a commodity by College Board and ETS. Academic programs don't care what your scores are. They only use them to help you get fellowships so they don't have to pay for your salary.

The real shame with the GRE is that it says nothing about the quality of the student. I knew someone who just failed their qualifying exam and is getting kicked out, even though he scored a PERFECT on the GRE and had external funding.

So what is the answer to this problem?

The GRE is a pretty generalized test. Most undergrad programs use either the generic SAT or ACT as their test factor - it's a efficient and useful way to measure a large group of people. The most important prerequisite for most graduate programs is an undergraduate degree in that field. How you performed (as well as the quality of the program you graduated from) is likely the most useful tool, so I can see how the GRE would be downplayed.

In some graduate programs, specifically law and business, what field your undergraduate degree is in isn't that important. I'm guessing this is why the LSAT and GMAT tend to weigh more in admission decisions than the GRE does.

There are eight additional subject specific tests related to specific discliplines. http://en.wikipedia.org/wiki/Graduate_Record_Examination#GRE...

Yes, and let's not forget that a generalized test is still useful at the graduate level, for a variety of reasons. The one that comes to mind when you mention subject tests is that people sometimes switch disciplines. I have several friends who went from math to physics and vice versa. They intended to research problems which are closely related to what they've done in their own department but that just happen to be worked on by professors in the other. Does it make sense to hold a math major up to the standard of a physics major taking the physics subject test? No way---he couldn't possibly compare, even though he may be more ably suited to tackling a problem that's being worked on in the physics department. So having a generalized quantitative/analytical section is helpful in at least this case. If the physics professor looking at your application knows nothing about what his colleague is doing (your intended advisor), at least he has a reasonable indication you aren't a dunce.

"The fact that every discipline uses the same test for admissions says more about the quality of academia than the scores."

What discipline is completely without use for reading skill, or for the kind of secondary school mathematics that appears on the GRE quantitative section?

"What discipline is completely without use for reading skill, or for the kind of secondary school mathematics that appears on the GRE quantitative section?"

The GRE doesn't measure reading skill, that's what the NALS is for. The GRE verbal is just a test of how many wank words you've memorized. Same goes for the math test.

"The GRE verbal is just a test of how many wank words you've memorized."

Evidence for this? (By which I mean, do you have any evidence that the GRE verbal section has failed validation studies to which it has been subjected by graduate schools?)

Well, that's probably the leftover liberal arts philosophy that seems to be so common among college administrators.

At good schools, the GRE is just a formality, especially the general test. Research experience and recommendations are more important. Top math and CS departments don't really care about the GRE general at all.

This is absolutely true. Top CS programs don't even care about the CS GRE very much, at least in my experience (e.g. I didn't have time to write it before applying to grad school, and it wasn't a big deal).

So the "best" schools make you take a test they don't even care about? Yeah, they sure sound really good to me.

Well, no, it's not required everywhere. And the GRE isn't meaningless, it's merely one factor among many.

I don't think this is true in neuroscience -- GREs are more than a formality.

Interesting: Philosophy scored higher on Verbal than English/Language.

Makes sense, philosophers are more interested in analyzing the nuances of words than writers. Writers focus more on the effect of words and philosophers focus more on the meaning of words.

My understanding is that English Lang & Lit grad students are not studying to be writers. They're studying to be literary critics and English professors. Writers get MFA's in Creative Writing or Journalism.

Good point. But, don't these grads also focus more on the effect of words than their meaning?

What's the analytic scale mean? I have an analytical writing score but it's on a 0-6 scale.

800 quant, 710 verbal. And 46 years old when I took it. ;-)

In the past (around the turn of the century), the analytical section consisted of logic and reasoning problems. It was probably similar to some of the LSAT questions.

They converted it to a writing section relatively recently

In 2003, in fact, the year after this data was taken. (That was when I took the test, and I had the writing section instead of the analytical, LSAT-like section. Much, much easier.)

Do you know why they changed? Seems like the previous version would be better than the writing section. While most score 5-6, the older scores look like more of a differentiator.

People are moving standardized tests away from being iq tests, for some reason. Why don't they just replace the SAT and GRE with a real IQ test?

I believe a lot of standardized tests have added writing components because its a basic practical skill that people often arrive at university lacking. I may be wrong, but beyond just the score, I think the actual writing is also sent to the schools you apply to.

In that case, the scale should be tougher. It isn't very useful when most are clustered at the top.

the total scores for the scientific disciplines are consistently higher than those for the humanities and social sciences;

Economics scored #4 and it's not a science. That blows away any credibility this might have.

Economics does have equations and an analytical approach, but its empirical foundations and usefulness are quite dubious, compared to say, civil engineering, or medicine. I would think that medicine would require a very high degree of empirical rigor, and it must be put into practice every day.

This isn't a list of what occupations are useful or even intellectually honest. It's ordered by complexity of abstraction.

no, it's ordered by score.

Also, students applying to medical school don't take the GRE. They take the MCAT. If you are interested in medicine and taking the GRE, you already failed, which may explain the disproportionately low score.

GRE is a test of your perseverance, tenacity, ability to stay focused and work towards a goal more than it is the test of how well you speak/understand/write English.

Ego boost or what?!

I study physics, math, and philosophy. They each take a top in category award. Know what? I'm going to treat myself to a coffee.

[I'd like to see the same done for the individual branches of philosophy. I have a feeling the moral philosophers drag us down. What? Who said that?]

How is it that the average verbal scores are so much lower than math? A math major averages 714 on quantitative whereas an English Lit major averages 573. Is the math that much easier, or do we not speek so goode nemore?

The GRE math questions are almost the same difficulty as the SAT.

I would guess that if you're applying to graduate school, you've done a fair amount of reading and writing no matter what your major, so those questions are "tougher".

However, its not the same for math. There is still a general requirement, but the exposure to rigorous math is much more varied. Which reminds me, I kind of remember seeing a "Math for Dancers" class in the catalogue...

I think Math for Dancers would be an interesting class to take, even if you're majoring in a hard science/math.

Not all people think with only their brain, some people need their body to think. It'd be interesting to see how that actually works from the point of view of someone who math comes to naturally (sans-dancing).

They math questions are math questions and the verbal questions are mostly an obscure vocabulary test with a dose of reading comprehension.

Wrong answer. (Though a true statement.)

The tests are scaled differently. That's all there is to it.


You mean the math is intentionally "scaled up"? Or that the math is just easier & there is no effort made to rescale it? Given that these scores are used for comparison so often, you'd think they would normalize the scores so that a 700 means the same thing on both sections.

For what it's worth, this is a trend on other standardized tests as well (including the SAT).

Sounds easy. Why does everyone do so bad?

The vocab is really obscure and even if you know a lot more words than the average person, you still get bitten. The only answer is to study very hard on a nearly completely useless subject.

Does anyone know where I could find data on more recent scores, or scores preceding 2002? I'm just curious to see if there any trends and whether the standings have changed much in the past 10 years or so.

Interesting that Math and Science graduates barely scored lower on the verbal sections. I've always felt that the so called verbal section is really more of a discreet math/type theory test plus vocab.

There are many people who take the GRE for math and science who have English as a second language. This is less true for the humanities and social sciences (besides Anthropology, a major exception.)

Can you explain?

The verbal section consists of antonyms, analogies, sentence completion, and reading comprehension.

I don't know much about discreet math. How does it apply in this case?

The antonyms and analogies are usually a matter of figuring out which of the answers is a valid type- for instance, most of the antonyms will have 2 or 3 words that actually mean the opposite, but only 1 will be the same figure of speech, and usable in the same type of sentance. It's basically like acting as a compiler and finding type mismatch errors. As for the sentence completion, the task is to memorize a set of rules, look at an input, and once again act as a compiler. Reading comprehension is probably the most 'verbal' of the verbal questions, but once again math and CS students have the advantage here: we've been reading obtuse texts looking for the answer to a specific question for years!

" It's basically like acting as a compiler and finding type mismatch errors."

This is exactly how I answered many questions in the Verbal section. (Score: 800 (verbal) 790 (quant) 6.0 (writing)).

as for knowing difficult words, I found that reading a lot helps (vs memorizing word lists). You've seen most words in context before, which helps to distinguish between similar words with different shades of meaning and so on.

Well, I admit I was somewhat skeptical about your claim, but at the risk of engaging in some confirmation bias, that is definitely an accurate description of how I got through my ACT/SAT verbal section. (Never took the GRE.)

I see what you mean.

The GRE verbal section uses very difficult vocabulary. Learning new vocabulary well enough for the GRE was a very daunting task for me.

If you know the words, the test is easy. The trick is knowing the words.

We don't talk much about discreet math.

skreet skreet

And if you want another (false) ego boost, calculate your iq from your gre:


It's hilarious that sites that purport to calculate IQ scores from other test scores generally don't mention error in estimation in either kind of test, nor do they mention regression to the mean.

Yeah, it's quite a sketchy methodology. I know his stats are false because I come out with a 150 IQ, which I definitely do not have. Plus, Mensa doesn't consider the modern SAT or GRE to correlate with IQ.

Is it me, or is education being on the bottom uplifting? I feel like we'd have a serious problem if it were on top.

Ceteris paribus, sure. But if the field of education attracted better talent -- provided better rewards for excellence -- a lot of our other social/economic problems might be lessened. Sadly, it's not our A-team or even B-team educating the multitudes...

Religion > Education ... interesting

That's close for physics and mathematics: 790 quant, 590 verbal for me.

800 quantitative and 660 verbal and I'm an EE.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact