Am I missing something? I thought the GRE was pretty easy.

Am I missing something? I thought the GRE was pretty easy.

That's only if you take it with your brain.

These are students who take it with their hearts.

DB

Derek Jeter dove into the leftfield stands to take the GRE

When I was a grad school at Yale, finding a research assistant was easy. If there were there, they were probably good enough. ... My arguments vary - some students are the first in their family to go to college, others have worked full time, others are raising a family, and still others are products of poor high schools that haven't given them a solid understanding of basic mathematics or writing mechanics.

This guy attended Yale, the fine college which gave us George W. Bush?! How many grammatical mistakes is a college professor allowed to make in one sentence before I can say his heart alone just isn't good enough for him to be a teacher?

Here is what Prof. Heart would have written had he studied at a better university than Yale:

When I was in grad school at Yale, finding research assistants was easy. If they were there, they were probably good enough. ... My arguments vary - some students are the first in their families to go to college, others work full time, others are raising families, and still others are products of poor high schools that didn't given them a solid understanding of basic mathematics or writing mechanics.

It is not all that hard to keep singular and plural concepts straight and to use the correct verb tense. I guess at Yale he could get away with not being able to write.

I knew a James Kaufman in college, but I thought the guy in the picture looked about 20 years older than me. According to Wikipedia, it is the same guy I knew. I think we had a freshman writing class together.

Joe Posnaski - an outstanding on-line baseball writer, one of the very best

I think he's trying to pay him a compliment here, but this comes off as damning with faint praise... like, "There's this obscure blogger I've discovered who writes some interesting things about baseball. I think you'd like him!"

Poz is an "outstanding online baseball writer," yes, but that's just something he does in his spare time... for free. He's also an award-winning newspaper columnist, a bestselling author, and a rising star at Sports Illustrated (magazine and online).

Pshaw. Psychologists think they can measure anything. If you pretend you can measure "self-esteem", "sense of belonging", and 300 different personality types, then why not "heart"?

Poz is an "outstanding online baseball writer" in the same sense that, I dunno, Ronald Reagan will go down in history as "a former governor of California".

This guy attended Yale, the fine college which gave us George W. Bush?! How many grammatical mistakes is a college professor allowed to make in one sentence before I can say his heart alone just isn't good enough for him to be a teacher?

You're picking nits, but yes, that wasn't the best series of constructions.

My problem with this blog/essay is that it doesn't appear to have a conclusion that relates to its thesis. I'm not really sure what he's getting at. He decides that he likes stats, but he also likes heart, and that Poz has been unmanned but that he also is kind of right because the blogger agrees with him a little and perhaps stats aren't useful in that they don't take into account something as the stats themselves, like GRE, occasionally miss stuff and there's no reason to forget a person's internal conflict such as they mama got sick before they had to write the most important presentation in the world and here's the dilemma one faces when writing a scouting report aka a letter of recommendation and so on.

I don't think this guy understands what Poz was talking about.

It is not all that hard to keep singular and plural concepts straight and to use the correct verb tense.

And yet:

When I was in grad school at Yale, finding research assistants was easy. If they were there, they were probably good enough. ... My arguments vary - some students are the first in their families to go to college, others work full time, others are raising families, and still others are products of poor high schools that didn't given them a solid understanding of basic mathematics or writing mechanics.

Oops.

Fixed.

some students are the first in their family to go to college

some students are the first in their families to go to college,

Surely the first is correct. Each student has but a single family.

Each student has but a single family.

You could have an adopted family and a biological family.

You could have a family so torn apart by divorce that they never talk to each other, but they talk to you.

“There’s no stat yet that measures heart." Or any test, either.

EKG?

Pshaw. Psychologists think they can measure anything. If you pretend you can measure "self-esteem", "sense of belonging", and 300 different personality types, then why not "heart"?

Walt, as you are one of the most statistically aware posters on this site, I will call bull**** on this. Unlike baseball statistics, psychological tests are held to high standards regarding reliability and validity. The next new-fangled baseball statistic to have to be evaluated as thoroughly will be the first.

I'm not from the US so the GRE was the first test of its kind I've ever taken. I was struck by how easy it was, and how it didn't seem to be testing me on anything except for my ability to study for the GRE.

Do American schools really take the GRE seriously?

I'm not from the US so the GRE was the first test of its kind I've ever taken. I was struck by how easy it was, and how it didn't seem to be testing me on anything except for my ability to study for the GRE.

Do American schools really take the GRE seriously?

Not sure how seriously schools take it. But, the ability to study is a real skill.

What standardized tests do well, IMHO, is screen for the basic skills required to succeed in higher education. Advanced literacy (ability to read and understand in a limited time period), numeracy, ability to think logically, and ability to study and prepare. They are not really meant to be tests of knowledge, rather, tests of critical skill sets.

View them as a scouting report focusing on tools, rather than the minor league stats that are GPAs.

You can't measure heart, but you sure can measure things like response rate, omission, and comission error rates

some students are the first in their families to go to college,

Can't you just add the word "respective" in there to make it work?

Free "grammar" opinion: indefinite "they" in the sense of "one" is good idiomatic English, becoming more acceptable in formal usage all the time because it's the most natural way to avoid "he," which is sexist, "s/he" which is unpronounceable, and "one" which sounds like Eleanor Roosevelt.

Using "they" when you mean "he" or "she" is wrong and annoying. If you object to using "he," then say "he or she."

flournoy, I fully understand "annoying" (which is in the ear of the annoyed, and not really arguable). But why "wrong"? Strictly speaking, it's "wrong" to say "you" when you mean "thou" or "thee," but who makes laws for pronouns? Pronoun systems change every few centuries, and we're living through one of those changes.

This guy attended Yale, the fine college which gave us George W. Bush?!

To be fair, Yale also gave us Dave Longstreth, who is a freaking genius.

What standardized tests do well, IMHO, is screen for the basic skills required to succeed in higher education.

They also measure the ability to not completely freak out in what for some people is a very high stress situation. And a tip of my cap to those people, as their colossal failures in test situations have made people like me (a good test-taker) look much better by comparison over the years.

Rich, colleges and universities no longer "do" grammar.


Do American schools really take the GRE seriously?

Yep, but in different ways. Most chemistry programs I know look very carefully at the quantitative part of the GRE and don't really care about the verbal. And for any of these tests, you probably can't differentiate between two students who score 80th and 70th percentile. But you sure as hell can between those two and the 20th percentile kids.

Do American schools really take the GRE seriously?

From my experience, it's seen as a "minimum competency" requirement in science and engineering programs. The GRE is extremely easy, and if you don't get at least score X, I don't think any level of scientific intuition and engineering curiosity is going to get you through a graduate program that requires an original research contribution.

I spent all of yesterday watching very carefully a machine that measured my baby's heart. It simultaneously measured my wife's heart.

This is a lot of nonsense.

What Elisabeth says in 28. Doing well on the GRE is not a guarantee that one will do well in grad school. But doing really crappy is a very good sign that one won't.

I spent all of yesterday watching very carefully a machine that measured my baby's heart.

Congrats!

I don't agree that studying makes that much difference. Practice can improve your scores a little but not that much.

Also how easy the GRE is doesn't matter since your score is evaluated relative to your peers. Most graduate programs use them to filter out applications. Usually you have to get over 600 in all three facets to even get your application read but after that they don't matter. In the spirit of the test shall we compare GRE scores. Other than the analytical which no longer is part of the test my scores were only ok but still good enough to get the rest of the package opened.

I'll go first:

Analytical: 800 (99 percentile)
Verbal: 650 (90 something percentile)
Quant: 720 (80 something percentile)

Dear god, I don't even remember my GRE scores.

Dear god, I don't even remember my GRE scores.

Me either. Which is funny, b/c I remember my PSATs, SATs, LSATs and GMATs (yeah i was confused about my career choice).

As of 2006, they eliminated the Analytical Reasoning section of the GRE. It was replaced with a writing section.

Aw, that was my favorite.
[/#### swinging]

My advice is never unnecessarily put yourself in a position where taking the GRE® or any test administered by ETS is a consideration.

Using "they" when you mean "he" or "she" is wrong and annoying. If you object to using "he," then say "he or she."

Oh, come on, how many times do we have to go over this? It is perfectly acceptable to do this. And has been for quite some time. And is getting more and more acceptable due to what's stated in #23 and #25 above.

I actually did very little studying for the GRE; I spent half an hour on the internet looking up free sample questions a few days before I took it.
But because I always enjoy discussions of standardized tests, and online penis-measuring contests, I'll go next.

Verbal: 800 (99 percentile)
Quantitative: 800 (93 percentile) (which is interesting, really)
Writing: 6.0 (out of 6) (something like 96 percentile, I don't vividly remember.)

So, in short, I win! (or at least tie)

35: when I took the test in 2003, I believe there was only quant, verb, and writing.

Two stories:
1. I had an excellent score on the verbal section of the SAT, and a good but not great math score. This bothered me, since I wanted to study engineering. So I took the test again, and I got the exact damn score on both sections. I picked engineering anyway.
2. The biggest trap that people warned me about on the GRE was going too quickly. Since it was a computer adaptive test, it was very important not to make early mistakes since you'd have to dig yourself out of a hole. But, in my panic, I whizzed through the math section and found myself with 5 minutes left for the final question. The question was something like, "How many integers between 1 and 1000 have "three" as one of their digits." Since I had the time, I decided to write out every damn number and just count. I'm fairly certain that is not how they wanted you to figure it out.

For my science program in 2004 when I applied, getting a 750 or higher on the GRE quant was pretty much a requirement for admission as a white male, and you were strongly encouraged to shoot for an 800 (which any talented quant person should be able to do, something like 10% of the testtakers on the GRE used to get an 800 in math).

I remember my SAT and LSAT scores (but not the percentiles for the SAT), but the SAT is all messed up now that they combined it with the Achievement tests. Also, don't ever take a foreign language as the subject test when taking the SAT2 or whatever they're called now. Being the idiot that I am, it never dawned on me that I'd be competing for percentiles with a bunch of native German speakers.

Also, I've never understood why the LSAT is scaled between 120 and 180. If you can't get worse than 120, and better than 180, why not just scale it between 0 and 60?

And someone mentioned the 20th percentilers. I taught the LSAT for the Princeton Review for a very brief period of time. I found it incredibly difficult. Much more difficult than the test itself. As someone who was self taught (i.e., I read books and took practice tests, but didn't take a prep class), it was very hard getting dumb people to understand why their answers were wrong.

And speaking of the 20th percenters, I taught the LSAT for the Princeton Review for a very brief period of time. I found it incredibly difficult. As someone who was self taught (i.e., I read books and took practice tests, but didn't take a prep class), it was very hard getting dumb people to understand why their answers were wrong.

Weird. I taught LSAT for a few years at Testmasters and I found it very easy to improve people's scores. Most of the people I tutored went up 7-15 points, one girl 20. I got a couple of tearful thank you calls.

Part of that is probably that the Testmasters program is very good and the P.R. one is crappy.

As someone who was self taught (i.e., I read books and took practice tests, but didn't take a prep class), it was very hard getting dumb people to understand why their answers were wrong.

My guess is that this is akin to the whole superstar makes bad coach phenomena. It's those not naturally gifted at something who are better at teaching someone how to master a particular skill. Those who can do it well innately can't explain it to those who aren't so gifted.

I would imagine that an unwillingness to suffer fools lightly may have been at play as well.

My guess is that this is akin to the whole superstar makes bad coach phenomena. It's those not naturally gifted at something who are better at teaching someone how to master a particular skill. Those who can do it well innately can't explain it to those who aren't so gifted.

I would imagine that an unwillingness to suffer fools lightly may have been at play as well.

No, really, it couldn't be further from the truth! I am bizzarely gifted at standardized tests, always was. When I started teaching, I sat down and tried to break down what I was doing innately. When I tutored, I realized that many people were just approaching the test in a totally different way from a cognitive point of view than I was. See, in life, there are many ways to skin a cat and they're all effective. Someone could have an approach to problem solving that worked for them in "the real world". But for tests, my way was dramatically better than all others. What I would do is take my students and, literally, break them- engineer practice problems or quasi-socratic tutoring such that they would totally fall apart-- and at that point, and only then, would I be able to reengineer their approach to standardized tests. And when they made that change, which takes a shitton of practice and such, ZOOOM their scores would fly up. I took people who had already been through courses, tutored them privately for 3 months, raised their scores dramatically. A girl from a 148 to a 163. Another girl from a 158 to a 169. A guy from a 163 to a 172. Those were life-changing type type results- those kind of gains really affect what school you get into.

The problem with standardized tests is that people who do badly at them are defensive about them. "It doesn't measure intelligence". "I'm smarter than it says I am". "Its unfair".

I would tell my students whenever I heard that (and lord knows, I heard it from every one of them) that (a) it does measure intelligence, albeit a narrow slice of it (b) it doesn't matter about its objective fairness, because to get what YOU want ( to go to the law school of your choice) you have to beat it at its game. So come to terms with the fact that it's an aptitude test, a pretty good one, the fact that you're not scoring well on it suggests that you may not be naturally smart at that specific slice of intelligence...and now, take a few days, get over that, come back, and learn how you can game the test and kick its ass.

Sure, one student never came back after that little speech. But the rest came back and ####### destroyed that ############.

I am bizzarely gifted at standardized tests, always was.

Same here. That, or I'm a super-genius*.

Hmmmm ...



*Especially since I, ahem, know how "bizarrely" is spelled.

The only thing I remember about my GRE is that I was 99th percentile on the verbal. I did not, however, score in the 99th percentile on the SAT verbal. The inescapable conclusion is that high school students as a group are much smarter than college graduates. Or perhaps that's too strong, and it's just a matter of college destroying innate verbal ability.

my GRE is that I was 99th percentile on the verbal. I did not, however, score in the 99th percentile on the SAT verbal

Unlocking my memories, which can't perhaps be trusted after 30+ years, I think I scored 720V 650Q on the SAT and then 790V 580Q on the GRE. Which may be a natural outcome of studying way more humanities than math & science in college. Use it or lose it, that kind of effect.

1. I had an excellent score on the verbal section of the SAT, and a good but not great math score. This bothered me, since I wanted to study engineering. So I took the test again, and I got the exact damn score on both sections. I picked engineering anyway.

Same here. I took the GRE about five years after graduating. Dominated the verbal section (90th percentile), viciously crashed on the quantitative (50th percentile). Took a year and added a psych major to my undergrad degree, figuring getting back to school and doing some math again would bring my quant score up. Took the test again and got the exact damn scores. I learned a lot about stats and probability, but didn't review any algebra. But now having taken psych for a year, I could curse the GRE's test-retest reliability.

Using "they" when you mean "he" or "she" is wrong and annoying. If you object to using "he," then say "he or she."

Having to read "he or she" 46 times on a single page is infinitely more annoying. And there is no way, you can get away with single "he" in academia these days.

1. I had an excellent score on the verbal section of the SAT, and a good but not great math score. This bothered me, since I wanted to study engineering. So I took the test again, and I got the exact damn score on both sections. I picked engineering anyway.

I was the opposite. I did very well on the math section, and horribly on the verbal. I actual got worse on the verbal between my junior and senior years (but better on math). I had absolutely no interest in anything mathematical in college. I took the one math class I was required to take, and that was it. Liberal arts all the way. I even took extra science classes to avoid taking more math, at least to the extent that Bees and Beekeeping, Spring Wildflowers, Earthquake Country, and Natural History of Insects can be considered science classes.

Also, I've never understood why the LSAT is scaled between 120 and 180. If you can't get worse than 120, and better than 180, why not just scale it between 0 and 60?

Subtract 100 and multiply by ten. It's actually scaled the same a a single SAT or achievement test, 200-800.

A 178 is a 780.

I teach for Kaplan, and in order to pick up a new exam (currently teaching LSAT and SAT), you have to score 90th percentile or above. I am currently score-qualified to teach GMAT and GRE as well. As a natural test-taker, it's only the MCAT or similar exams requiring technical knowledge that I need any study time to score-qualify for.

Score-qualifying for the GRE felt hardest, because of the very tight timing on the verbal section, although I was something like 98th and 99th percentile. I am a very fast test-taker, but the practice exam I took felt very hurried on the verbal.

Teaching the SAT, GMAT, or GRE relies on three basic things: review stuff the students used to know (basic algebra, geometry), encourage them to practice vocabulary, provide insight to structural aspects of the test that can be gamed. All three are content-based tests, and they are very, very easy to teach to. The hard part is motivating the students to do what must be done.

The LSAT, on the other hand, tests reasoning, not content. Memorization is practically useless. There are few "tricks." That said, I find it a pleasure to teach. For low-scoring students, I can get almost-immediate and dramatic score improvement just by showing them how to draw a useful sketch for a logic game. The hardest part of the test to teach is Reading Comprehension, because quite often the difference between students comes down to how quickly they read.

I agree that the scaling of these exams is really, really dumb. If you think ERA+ is worth tearing into, you can have a field day with standardized test scores. If I tell you my student got 12 more questions right on the LSAT, that represents a dramatically different score improvement if he started with a 125 than if he started with a 165.

When I started teaching, I sat down and tried to break down what I was doing innately. When I tutored, I realized that many people were just approaching the test in a totally different way from a cognitive point of view than I was.

This is most apparent if you try (and I mean really try) to get a software engineer or an insurance underwriter to explain something- they really have no idea (most don't anyway) that their answers sound like gibberish to the uninitiated- so you have to keep asking- but differently-

Q: What does this provision, co-insurance, do? or mean?
Underwriter: It allows the insured to share in the risk.

Most people give up, WTF, they'll smile and nod, and have no conception of how that "answer" relates to the question.

In case you are wondering, what "co insurance" means, in layman's terms, is that the insurance policy only pays a % of the loss- your $10,000 ring is stolen, with a 20% co-insurance provision the insurance company only pays $8,000. Why is it called "co-insurance"? Because in theory the ring is 100% insured, only you are providing insurance for 20% and the insurance company 80%- and when you look at it THAT WAY, the underwriter's answer kind of makes sense- you and the insurance company "share in the risk" of loss- your share is 20% and the insurance company's is 80%...

Of course the underwriter's phrasing seems wonky still- the Insured is "allowed" to share in the risk, "allowed"? Oh gee, thank you for allowing me to share in the monetary risk of having my house burn down Mr. Insurer, it's a comfort to know that if my house burns down you won't pay to replace everything 100%, you'll allow me to pay a % out of my own pocket (assuming I have pocket's left)

According to underwriters, by "allowing" an insured to "share in the risk" they provide an incentive to the insured to, well, not suffer a loss. Nonsense, the only reason an insured WILLINGLY agrees to a co-insurance provision is in exchange for lower premiums.

I only got a 1050 on my GREs when I took in Jan. I was 50% or lose to it in both math and verbal. My excuse is that I'm applying for a Masters in Education, and they don't give a #### about GRE, just that you took the test. Also, I didn't bother to study for the exam. I'm sure if I apply to a History/Political Science/Econ Grad program, I'd redo the exam, and actually try to pass it with a decent score.

According to underwriters, by "allowing" an insured to "share in the risk" they provide an incentive to the insured to, well, not suffer a loss. Nonsense, the only reason an insured WILLINGLY agrees to a co-insurance provision is in exchange for lower premiums.

Well yes, but deductibles and co-pays do affect policy-holder behavior, therefore allowing lower premiums.

Especially with something a easily sold/pawned as a diamond ring, you can see why the insurer wouldn't want to cover 100% of the cost. In fact, the insurance probably should only be payable to a jeweler to replace the ring.

Just like if your house burns down, the insurer pays to replace your house/possessions, they don't cut you a big check to spend at your discretion.

Hi all,
Sorry about the bad grammar (I like to write & post relatively quickly, at the risk of embarrassing my alma mater). But enjoyed reading the comments (and quoted one of them when I fixed some of the grammar).
Cooper, wasn't trying to slight Poz and took out "on-line" in the article. He and Neyer are bookmarked and my lunch reading.
I did get Poz's conclusion, but it was a riff -- my own thoughts. I know that it's not a perfect (or, perhaps, a good) analogy.
Tom Foley, I was balding and graying then -- I just became an old man sooner. Hey, was never carded. I am hoping that it circles around and when I am 70 I can pass for 20.
Larger point -- in psychology (at least), GRE scores and GPA are the #1 and #2 most important determinants of grad school acceptance with other things (letters of rec, personal statements, etc) being WAY WAY down there. I have very mixed feelings about standardized tests (used to work for ETS, saw inner workings both good and bad). But I feel for the students I teach and mentor.
Anyway, if you like bad grammar and poor baseball analogies, you might enjoy Creativity 101 (http://www.springerpub.com/product/9780826106254)
Cheers,
James

I'm starting to sweat the GRE as I am preparing to apply to grad school (Ph.D in political science) next year. My research interests feature a significant quantitative component, but my quant score is comparatively weak (around 580-610). I'll need to raise it to at least 650, ideally to 700, in order to have my application looked at by top schools. Does anyone have advice for how to improve my score?

Also, my overall application is hampered by comparative lack of quantitative background (topped out at introductory statistics and a political science research methods course, and no classes in economics -- yay, thinking I'd be interested in theory!). So, yeah -- how much should I be freaking out about that?

i thought my GREs were terrible, but my grad school advisor told me i had the highest scores in the department in quite a while.

well, it was art school, so take that fwiw.

Walt, as you are one of the most statistically aware posters on this site, I will call bull**** on this. Unlike baseball statistics, psychological tests are held to high standards regarding reliability and validity. The next new-fangled baseball statistic to have to be evaluated as thoroughly will be the first.

Not only am I "statistically aware", my area of specialization is measurement models. Most standard scales are crap. OK, that's not fair. Most of them are in the ballpark but poorly constructed, not well-analyzed, not subjected to sufficient levels of criticism and not well thought-out conceptually. And many of them are 30-40 years old, hatched in the early days of measurement modeling and have an established acceptance that is extremely difficult to overcome.

Almost by definition, baseball statistics are of a MUCH higher standard than psychological scales because baseball statistics have virtually no measurement error in them. And there's really no theoretical debate to be had about what constitutes a "hit" and little/no debate about the direction of causation. New-fangled baseball statistics are generally based on regression-type models which, in general, are much more stable and reliable models than measurement models.

The problem with baseball statistics has nothing to do with measurement. The problem with baseball statistics is that you need to detect extremely small effects which means you need massive sample sizes to reliably estimate them and, of course, those aren't available. Hence, the predictive capability of the statistics is low. But that performance has very little to do with theory and conceptualization, nothing to do with measurement (except around issues of "heart" which no new-fangled stats are trying to measure as far as I know), little/nothing to do with concerns about direction of causality (and whether causality is even testable), and little/nothing to do with issues of validity and reliability. The issue is a very small signal in a sea of very large, _random_ _sampling_ noise.

If psychological tests had to detect differences as small as the difference between the 350 OBP and the 320 OBP players, they'd be absolutely useless.

For a very nice example, take a look at an article by Perrira et al on the CESD-19 depression scale (sorry, don't remember the journal, it's within the last 10 years for sure, maybe the last 5). First they show that the scale doesn't work well across ethnic groups in their sample (the national adolescent health study). Now this scale had been around for a long time already and that shouldn't have been a new result. They then reconfigure the scale greatly, treating something like 10 of the "effect indicators" as "causal indicators" (with theoretical justification which you may or may not agree with), another 4 as "outcomes" and leaving 5 as "effect indicators". This model fits quite well across groups and strongly suggests what you need for this scale are those 5 effect indicators and that the standard model is badly mis-specified thereby creating serious issues.

The CESD-19 has been around forever without anybody questioning the direction of causation between the indicators and the latent construct -- a fundamental question in building any causal model (which all measurement models are). And I bet that almost all applications of the CESD-19 since that article still use the model for which Perreira et al provide strong evidence isn't a good model.

The issues found by Perreira et al are pretty common to psych scales. Classical measurement theory never even considers the question of direction of causation and so most psychologists never consider it either (most of the work on "causal indicators" comes from outside psychology). Once a scale gets established (deservedly or not) people use it without seriously questioning its concepts, construction, question wording, etc. (This is standard human behavior ... as psychologists know.) Near as I can tell, the primary tool remains exploratory factor analysis which is just a horrible statistical approach for "testing" a scale and, I'd argue, not even particularly good for developing one. When these traditional psych scales are subjected to more rigorous models -- confirmatory FA, multi-group analysis, DIF, IRT, categorical CFA, etc. -- they frequently exhibit serious problems. Also, although this has gotten much better, psychologists (and social scientists including economists) generally don't adjust for the survey design effects. (We can also discuss treatment of missing data although CFA was ahead of the curve on that issue at one time.)

Maybe things have improved greatly in the last 10-15 years but the Perreira et al article (which I stumbled across while researching something else) suggested to me that things were pretty much the same.

And I don't even want to think about the "Sports Loyalty Engagement Index". :-)

I'll grant you that my work the last 10+ years has not been in the area of psych scales so I haven't kept much of an eye on the literature. Maybe psych has gotten better about seriously testing the measurement properties of these scales and, more importantly, improving them based on those studies.

People complain all the time about the GREs, but the fact is they very accurately measure your ability to work hard and jump through pointless hoops, which is the primary determinant (and MUCH more important than innate ability) of success in graduate school. Basically, the verbal section is pure memorization-if you want a perfect score you just need to spend a bunch of time learning vocabulary. The quantitative section is a bit trickier in that there's nothing to memorize, but nevertheless any reasonably intelligent person ought to be able to practice enough to score no worse than the upper 700s.

The part I personally hated about the GRE was the stupid experimental section. In particular, as someone who needed a perfect score on the quantitative section (you basically cannot get into a decent economics, science, engineering or mathematics program without a perfect score on this section) I really did not appreciate being stressed out by having to complete two such sections and having no idea which one would count. Indeed I almost canceled my score because I was worried I had missed a question or two on one of the sections, but decided not to and ended up with the score I needed anyway. Still, it seems quite unfair to me to use already stressed out people who paid a lot of money to take the test as guinea pigs.

In particular, as someone who needed a perfect score on the quantitative section (you basically cannot get into a decent economics, science, engineering or mathematics program without a perfect score on this section)

Has this changed? I took the GRE in 1991/92 and I certainly didn't get a perfect score in the quant. section (although think I got 90%). I got into Chemistry depts. at Harvard and Columbia, and Biophysics at UCSF.

My understanding (which would agree with your statement) is that this was not the case in the past, but it has become so more recently. Basically, it appears that top programs in these fields have seen a surge of applications over the past 10-20 years and rather than devote more time to studying the applications in detail, they simply look for automatic disqualifier criteria to reduce the number of applications they look at to a more manageable level.

Even so, my previous statement overstated things a bit-people can (and do) get into top programs in these fields without perfect scores, it just makes it MUCH harder. For example, in many econ PhD programs, those applying with an 800Q get direct consideration from the admission committee, while anyone applying without an 800Q is tossed into a separate pile. Some random grad students volunteer to read this pile and select a small number (maybe 10% or less) of the applications that they like to be put back into consideration by the admission committee but the rest are tossed in the trash. So while you can get admitted without a perfect score, your chances of even being considered for admission by the people who actually make the decisions are much lower.

Are we actually back to comparing ####### scores on standardized tests?

Holy ################ the baseball season needs to start.

Hey Walt,

Thanks for the response. I agree with most of what you have to say, and even though there is no measurement error, the sampling issues are huge. Nonetheless, you are correct in saying that many psych measures are not the strongest psychometrically, primarily because of the amount of error contained in human behavior, collecting via self-report, etc.

But my point was less about the construct, and more about the need for reliability and validity. You address some of the reliability concerns with issues of factor analysis and modeling. Certainly, older scales are no longer valid as society changes, and we need to replace them with new ones. In psychology, there is a consistent push to demonstrate validity in our studies. A quick search of the CESD shows a number of validity studies with various populations. It compares well with other measures of depression, and the items are drawn from previously validated depression scales. I just found the article you reference (well, the abstract anyway). That article states that the CES-D needs some corrections for multiethnic families. There's nothing there that is not valid for non-minority families.

With respect to baseball statistics (which I love, don't get me wrong), we never bother with this step. UZR says Derek Jeter is bad, so he is, even if other measures of examining fielding disagree. Maybe UZR is right, but it's no wonder people struggle to accept it. Mark Texiera was a stud first baseman in 2008, and sucked in 2009. Is that right, or is UZR just an unreliable statistic that is not anchored to any conceptual underpinning? It's no wonder that some of the baseball world rejects UZR, since it flies in the face of everything that has existed before, and seems to be unstable. Perhaps over time, it will be validated.

But for now, I'd feel better about using the CESD 19 to screen for depression that I do with using UZR to meaningfully represent how someone plays defense.

With respect to baseball statistics (which I love, don't get me wrong), we never bother with this step. UZR says Derek Jeter is bad, so he is, even if other measures of examining fielding disagree. Maybe UZR is right, but it's no wonder people struggle to accept it. Mark Texiera was a stud first baseman in 2008, and sucked in 2009. Is that right, or is UZR just an unreliable statistic that is not anchored to any conceptual underpinning?

Are you serious? These things are discussed on this very site all the time! People are constantly examining the various fielding metrics, comparing them, etc, questioning the way things play into them.

And the statistics get changed as a result. See e.g. the long-standing controversy regarding ZR for LF in Fenway.

I got into Chemistry depts. at Harvard and Columbia, and Biophysics at UCSF.


Oh, it's a pissing contest and I wasn't invited...damn.

I got into Chemistry depts. at Harvard and Columbia, and Biophysics at UCSF.

That IS a really impressively sized penis. You should be very proud.

Has this changed? I took the GRE in 1991/92 and I certainly didn't get a perfect score in the quant. section (although think I got 90%). I got into Chemistry depts. at Harvard and Columbia, and Biophysics at UCSF.

My experience when I applied about six years ago was that a 780 or above was basically a requirement for engineering programs at Caltech, Stanford, Princeton, MIT, etc.

I barely graduated high school on time. Your move.

.thgir ot tfel morf etirw ot tegrof semitemos I

.thgir ot tfel morf etirw ot tegrof semitemos I

Is your first language Hebrew, Arabic, or something else?

I'm considering applying to physics-related grad programs, so here's hoping if I do, I find the GRE as easy as you guys did.

smtms frgt t wrt wth vwls.

I'm considering applying to physics-related grad programs, so here's hoping if I do, I find the GRE as easy as you guys did.

In all seriousness, it is very simple, straightforward math (I aced it having gotten a C+ in differential calc in college, which was my one math class in 4 years). Anyone who graduated high school knows all the math they need to get a perfect score; its just a matter of practicing till you can get zero mistakes (1 mistake drops you to a 780). That's why programs use it as a qualifier; it's a good proxy for your willingness to do mindless ##### work until you're really ####### proficient at it, and if my experience is any indication being a whiz at ##### work is possibly the most important skill in a hard science PhD program.

Telling anyone else about my test scores feels a lot like telling people about my fantasy baseball team. I might care, but no one else should. That said, I had a weird experience with my SAT and GRE scores.

I scored 200 points higher on the GRE math than I did the SAT math after taking no math courses in college. I scored 180 points lower on the GRE verbal after four years of writing scores of papers, reading constantly and learning a lot of words I didn't know in high school.

I scored 200 points higher on the GRE math than I did the SAT math after taking no math courses in college. I scored 180 points lower on the GRE verbal after four years of writing scores of papers, reading constantly and learning a lot of words I didn't know in high school.

I had almost the same experience. I have no idea what to make of these tests. I also had to take the GRE in literature. I didn't study for it at all figuring the idea of trying to study the entire history of literature was pointless, I don't remember how I did.

if my experience is any indication being a whiz at ##### work is possibly the most important skill in a hard science PhD program.

Yikes. I'll keep that in mind. It could be worse, though -- I could have waltzed into one of these threads and announced that I'm thinking about law school. That tends to stir up a hornet's nest of "Don't do it!"

Yikes. I'll keep that in mind.

This is why choosing your advisor is more important than choosing your school. Pick a good one, even if they're at a lesser school.

I scored 200 points higher on the GRE math than I did the SAT math after taking no math courses in college. I scored 180 points lower on the GRE verbal after four years of writing scores of papers, reading constantly and learning a lot of words I didn't know in high school.

That's not surprising. The GRE verbal is harder than the SAT verbal while the GRE math is easier than the SAT math. Zop is right. GRE math is somewhere around grade 10 math. It's very easy. The only issue is time.

Yikes. I'll keep that in mind. It could be worse, though -- I could have waltzed into one of these threads and announced that I'm thinking about law school. That tends to stir up a hornet's nest of "Don't do it!"

I switched from a geochemistry program to law school. I'm 3 years out from the switch, and I'm 100000000% happier in law school.

Maybe physics would be different (depends on the type of physics?), but if you're doing chemistry or bio or any other lab intensive science, you could be working for the best advisor in the world and you'll still be putting in some heavy hours in the lab doing mindless work. It's the nature of the beast.

Thank goodness I am done with applying to post-undergraduate schools. A tedious, annoying, nerve-wracking, and painful process.

I also find it somewhat amusing that despite all the hard work that it takes to get into post-undergrad programs, how you do in a half-hour interview has a huge effect on whether or not you get into the program you want.

This is why choosing your advisor is more important than choosing your school. Pick a good one, even if they're at a lesser school.

I'll second this, because I've heard it a lot from my Dad: he went with advisor over school, and was extremely happy with how that worked out.

Telling anyone else about my test scores feels a lot like telling people about my fantasy baseball team. I might care, but no one else should.

This. 100% correct on both points.

I remember the first time my boss started telling me about his fantasy basketball team...I thought to myself, "Holy ####, is this how I sound when I tell people about my fantasy teams?"

Walt Davis, are you the Walter Davis who was in the social psych doc program at U. of Texas in the late 70s? If so, howdy (though you may not remember me -- I started in '78 if that helps). If not, never mind.
Perry Sailor

I got a 790 on the quantitative (92 percentile), a 550 on the verbal (75 percentile), and a 3.5 on the essay (23 percentile). I also got a 710 on the subject GRE in mathematics (67 percentile) which is more important for the grad schools I applied to who asked for subject GREs (all but the University of Florida).

I graduated with a B.S. in mathematics in December, that is 3 years plus an extra quarter after graduating high school and am receiving my M.S. in mathematics in June (I took 29 units as an undergrad that I transferred over to my grad work). My undergraduate GPA was 3.55. I also had two quarters of summer research and experience tutoring and grading. I now have experience teaching as well, but not at the time that I had to submit applications.

So far, I've been rejected by Stanford, Florida, UC Berkley, UCSD, and Duke. Cal Tech guaranteed a response by April 1, so I should get my rejection from them tomorrow or Thursday. I also haven't heard back from UC Davis yet. I thought that I had a strong application, but I guess not.

Americans are funny.

"Walt Davis, are you the Walter Davis who was in the social psych doc program at U. of Texas in the late 70s?"

It can't hurt to ask. However, "there are 2,462 people in the U.S. named Walter Davis."

"I also haven't heard back from UC Davis yet. I thought that I had a strong application, but I guess not."

Good luck with UC Davis. I live in Davis, though I studied at UCSB (undergrad) and UCLA and UCSD (graduate school). I'm not now associated with UCD. I just live here.

It sounds like you are applying to programs in math and you don't know quite why you have not been accepted. If you are physically proximate to any of those schools, I suggest you go by and get an appointment to speak with the dean or someone in his office who knows the criteria for admissions and have them explain to you why they said no. It's possible that this is just a very tough year due to the economy and they are accepting fewer applicants and getting more applications (including from people who would prefer to be working but can't land a job, now). It's also possible that there is something they are looking for which you are lacking. If you don't get in this year, you might be able to improve your resume in that regard and then apply again, later.

Also, my overall application is hampered by comparative lack of quantitative background (topped out at introductory statistics and a political science research methods course, and no classes in economics -- yay, thinking I'd be interested in theory!). So, yeah -- how much should I be freaking out about that?

This isn't anything to worry about. Practically everything you would learn about quantitative comparisons in one of those classes will be worthless for what the GMAT calls quantitative.

The Problem Solving is pretty much just algebra and geometry. All high-school level math. The Data Sufficiency sounds intimidating but it's really just a non-standard way of evaluating basic math. Point being, if you know algebra, geometry, and the few side topics cold (which should not be difficult if you study) and you know the way they are going to ask questions cold (which should not be difficult if you practice), then you should have no serious problems on the section.

I help some of our students who are applying for graduate programs in math. (Our own masters program doesn't ask for the GRE - we look at transcripts.) A math graduate program basically isn't going to learn anything from the math portion of the general GRE - most applicants are in the 700-800 range, and the differences there are uninformative. Rumor has it that some programs do care about the verbal portion of the general GRE - although if they're trying to figure out whether the foreign-born speak enough English to be put in front of a class as a TA, the TOEFL (or whatever it's called these days) also speaks to that. We guess that the GRE subject test (a wholly different beast) matters quite a bit more, especially for applicants coming from a program that the grad school in question might not be familiar with. And somewhere around 60-70 percentile, if that's the main claim of the application, probably isn't all that competitive at a top program. Letters of recommendation matter too, although I've written enough of them in my day to wonder how anyone at the other end can make read them and make any sense of them.

I barely graduated high school on time. Your move.

And that was in Mississippi!!