GRE - does raw score or percentile matter more?

[wb3060]

Out of curiosity while I'm anxiously awaiting my results...

I was looking through profile eval boards. Someone posted their stats and while their quantitative score was only 1 point less than mine, the percentile was 27 points lower (I'm also wondering if there is some kind of mistake here, as this seems too wildly different). I did take the GRE back in 2013, however. So I don't know if something about the time difference / test takers is different in the time frame. 

Anyway, what matters more - percentile or raw score? I notice people here typically just put their raw score and not percentiles. 

[ultraultra]

Most times I've talked to faculty about GRE, they've asked my percentiles rather than my score. They seemed to think that matters more.

Websites of schools seem to be a nearly even split between mentioning them both, perhaps with slightly more giving scores to beat.

Edited January 28, 2016 by ultraultra

[TakeruK]

I don't think you can get your raw score on the GRE. Maybe you mean the scaled score (the one that used to be out of 800 and is now out of 170). The raw score is the number of questions correct and this used to be reported on the Subject GRE (but this is no longer reported). On the Subject GRE, the raw score is not very useful because it is not adjusted for test differences from test to test and year to year. On the General GRE, with adaptive testing, the raw score is worse than useless!

The scaled score (out of 170, or 800 if you took it before August 2011) is adjusted for test difficulty. It means that if you got 165 on one test, you should have performed at a similar level to another person who got 165 on a different test. The percentile score is a measure of how your scaled score compares against everyone else's scaled score in the last 3 years. This means the percentile rank corresponding to each score will change every year (but it should not change by much, see note below).

In my opinion, which one matters more depends on what the school is trying to measure. If the school wants to know how applicants compare to the entire population of GRE test takers, then the percentile score is clearly the best way to compare. However, if the school wants to directly compare two candidates, I think comparing the scaled score is better. I believe ETS reports "uncertainties" in the scaled scores, that is, scores within some value are basically equivalent to each other, but they don't report this in the percentile score. So, if it turns out the uncertainty in the score is 3 points, then a score of 160 and 162 are basically equivalent. But, if this was converted to percentiles, there is no metric to determine how similar scores are.

That said, in practice, I don't think it makes a difference. At least, not right now. In the past few years (since the Revised General GRE test), there has been almost no shift in the scaled-to-percentile score tables. This is why people use them so interchangeably. Maybe in the future, when decades have passed, it would no longer be correct to compare someone's scaled score in 2013 vs. someone's scaled score in 2020. But, this is why scores are only valid for 5 years---it's unlikely that the testing population would change so much in 5 years that it would invalidate the comparison of scores.

Finally, for the specific scenario you mention, I think there has been a mistake. The GRE Q score did not change so much that 1 point became 27 percentile points!

[wb3060]

25 minutes ago, TakeruK said:

I don't think you can get your raw score on the GRE. Maybe you mean the scaled score (the one that used to be out of 800 and is now out of 170). The raw score is the number of questions correct and this used to be reported on the Subject GRE (but this is no longer reported). On the Subject GRE, the raw score is not very useful because it is not adjusted for test differences from test to test and year to year. On the General GRE, with adaptive testing, the raw score is worse than useless!

The scaled score (out of 170, or 800 if you took it before August 2011) is adjusted for test difficulty. It means that if you got 165 on one test, you should have performed at a similar level to another person who got 165 on a different test. The percentile score is a measure of how your scaled score compares against everyone else's scaled score in the last 3 years. This means the percentile rank corresponding to each score will change every year (but it should not change by much, see note below).

In my opinion, which one matters more depends on what the school is trying to measure. If the school wants to know how applicants compare to the entire population of GRE test takers, then the percentile score is clearly the best way to compare. However, if the school wants to directly compare two candidates, I think comparing the scaled score is better. I believe ETS reports "uncertainties" in the scaled scores, that is, scores within some value are basically equivalent to each other, but they don't report this in the percentile score. So, if it turns out the uncertainty in the score is 3 points, then a score of 160 and 162 are basically equivalent. But, if this was converted to percentiles, there is no metric to determine how similar scores are.

That said, in practice, I don't think it makes a difference. At least, not right now. In the past few years (since the Revised General GRE test), there has been almost no shift in the scaled-to-percentile score tables. This is why people use them so interchangeably. Maybe in the future, when decades have passed, it would no longer be correct to compare someone's scaled score in 2013 vs. someone's scaled score in 2020. But, this is why scores are only valid for 5 years---it's unlikely that the testing population would change so much in 5 years that it would invalidate the comparison of scores.

Finally, for the specific scenario you mention, I think there has been a mistake. The GRE Q score did not change so much that 1 point became 27 percentile points!

Ahh you're right, I mean the scaled score out of 170. Thanks for all the info! 

And I agree, I can't figure out how there could be such a large difference! 

[elw]

They're highly correlated. 

[alphazeta]

As described, these are basically the same thing, but I'd say the percentile is what matters.  Most of us have not caught up with the GRE score format change.  That is, most people have no idea what a 151 "means" and are purely looking at the percentile to evaluate the score.  The 800-point format was around long enough for people to have a grasp of what the scores meant, but it will take a long time for the new scores to have any intuitive resonance.

[s1994]

Raw score: 330; Percentile: 90%

Anything below that will be a red flag for top 10 programs (according to stats on the survey page). And I do not think faculty will think there is a huge difference between 92% and 94%. However, 99% can make a difference.

In short, I will say, if you can make 99%, then go for it. If not, anything around 95% will work the same. Your GRE scores at that time do not matter (especially because other applicants in the shortlist all probably have similar GRE scores). It's the time when your SoP seriously matters. 

Edited January 28, 2016 by shane94

[IndEnth]

The percentile tells you how you compare to your cohort. If you have 94% that means 94% of people scored worse than you and only 6% were as good or better.

And 1 point in the scaled score can make a big difference: only 17% of testtakers scored 162 or higher in their quantitative test in my cohort, but 20% scored 161 or higher. Because it's calculated in full points (and not like the analytical in .5 steps), more than 1% of testtakers is likely to score that number of points.

(I took the GRE a third time and scored 167 in the quantitative test - giving me 94% and proving that it's always worth retaking the GRE.)

[bhudson]

4 hours ago, TakeruK said:

I don't think you can get your raw score on the GRE. Maybe you mean the scaled score (the one that used to be out of 800 and is now out of 170). The raw score is the number of questions correct and this used to be reported on the Subject GRE (but this is no longer reported). On the Subject GRE, the raw score is not very useful because it is not adjusted for test differences from test to test and year to year. On the General GRE, with adaptive testing, the raw score is worse than useless!

You can see your raw score (number of correct answers/incorrect), the difficulty of questions you received, and time spent on each one in your MyGRE account.

After logging in, go to 'GRE Diagnostic Service' and click through to a specific test to see the breakdown of all sections.

Its pretty cool.

Edited January 28, 2016 by bhudson

[terefere]

4 hours ago, shane94 said:

Raw score: 330; Percentile: 90%

Anything below that will be a red flag for top 10 programs (according to stats on the survey page). And I do not think faculty will think there is a huge difference between 92% and 94%. However, 99% can make a difference.

In short, I will say, if you can make 99%, then go for it. If not, anything around 95% will work the same. Your GRE scores at that time do not matter (especially because other applicants in the shortlist all probably have similar GRE scores). It's the time when your SoP seriously matters. 

That's a bit of an exaggeration. For instance, look at Duke stats: https://gradschool.duke.edu/about/statistics/political-science-phd-admissions-and-enrollment-statistics

The average combined score last year was 323, and that means there must have been people with scores below 320 who got accepted.

I think everybody here is seriously overestimating the importance of GRE. It is simply used as a screening mechanism, nothing more than that.

[Determinedandnervous]

16 minutes ago, terefere said:

That's a bit of an exaggeration. For instance, look at Duke stats: https://gradschool.duke.edu/about/statistics/political-science-phd-admissions-and-enrollment-statistics

The average combined score last year was 323, and that means there must have been people with scores below 320 who got accepted.

I think everybody here is seriously overestimating the importance of GRE. It is simply used as a screening mechanism, nothing more than that.

From the website, it seems unclear whether those scores are median or mean. If they are median, I agree totally. If they are the mean, then it's possible these numbers are skewed by a few outliers in one direction or the other.

[s1994]

1 hour ago, terefere said:

That's a bit of an exaggeration. For instance, look at Duke stats: https://gradschool.duke.edu/about/statistics/political-science-phd-admissions-and-enrollment-statistics

The average combined score last year was 323, and that means there must have been people with scores below 320 who got accepted.

I think everybody here is seriously overestimating the importance of GRE. It is simply used as a screening mechanism, nothing more than that.

1) In a competitive year (i.e. 2014-15), it's 330.

2) Look at the stats of the students who got admitted in the past few years. You RARELY see a person with <325 combined scores getting into a top 10. And almost all people with <320 got rejected. Yes, we have sample bias because people with higher grades might tend to report their grades on the survey page. But hey -- the board is anonymous, the benefit of not posting one's low GRE grades is really small. So I do not think this counter-argument would work against my observation that 330 should be safe as a score.

Edited January 29, 2016 by shane94

[terefere]

These scores are mean. There are always outliers with perfect scores, so I believe it is skewed to the right and the median is actually lower than that.

Also, do keep in mind the correlation. GRE is not particularly hard to master, so people with generally good academic outcomes tend to get better scores. That's the real bias. Let's be honest, if you are academically gifted and generally do very well, how likely are you to get 150-150 on the GRE?

There is simply no good reason why schools should pay very close attention to GRE scores, other than to screen out people with VERY low scores, who are either badly prepared or too lazy to study.

[terefere]

My apologies, actually I was wrong, these are median GRE scores (and median GPAs)

[IndEnth]

From what I've heard from my sources, GREs are really used as a screening mechanism: If you don't score below a certain (probably arbitrarily defined) level that varies by school, they simply won't even bother to open your file. So if you have really low GRE scores, they might have you booted out before you ever even got a chance. But once they open your file, it really doesn't matter anymore - if you made it through the initial screening, what makes or breaks it are your SoP and letters of recommendation, not how competitive your GRE scores are.

[s1994]

12 minutes ago, IndEnth said:

From what I've heard from my sources, GREs are really used as a screening mechanism: If you don't score below a certain (probably arbitrarily defined) level that varies by school, they simply won't even bother to open your file. So if you have really low GRE scores, they might have you booted out before you ever even got a chance. But once they open your file, it really doesn't matter anymore - if you made it through the initial screening, what makes or breaks it are your SoP and letters of recommendation, not how competitive your GRE scores are.

Agree. I believe 155+155 is an extremely reasonable threshold for top schools. Any students who read carefully, think critically, and have a numerical literacy (which means pre-calculus in high school) should be able to get that score, even though he or she does not study *diligently*. 

[TakeruK]

6 hours ago, bhudson said:

You can see your raw score (number of correct answers/incorrect), the difficulty of questions you received, and time spent on each one in your MyGRE account.

After logging in, go to 'GRE Diagnostic Service' and click through to a specific test to see the breakdown of all sections.

Its pretty cool.

Cool!  Good to know!

[Straightoutta]

How many people had to pay for GRE prep services?

[s1994]

14 minutes ago, Straightoutta said:

How many people had to pay for GRE prep services?

You mean prep course? I used Mag**** (and nothing else) for half a month and I got 330. I have another friend using Man****** and OG (and nothing else) for a summer and he got 335. I really do not think a huge spending on prep courses has a causal relationship with a high score. People who do not study effectively or read critically (which are both essential to a graduate degree) will not get a high/decent score even though they spend tons of money on the test. 

Edited January 29, 2016 by shane94

[IndEnth]

7 hours ago, shane94 said:

People who do not study effectively or read critically (which are both essential to a graduate degree) will not get a high/decent score even though they spend tons of money on the test. 

Getting good GRE scores in my opinion has nothing to do with intelligence, critical thinking or your preparedness for graduate school. It's a pure function of how well you're able to study for standardized tests (which in fact is pretty much the opposite of critical thinking, a quality essential for grad school). I used the free prep materials and bought one more book, and then just buckled down and studied. That often times means simply knowing answers by heart, or at least knowing formulas etc that you haven't used since high school and will never need again by heart. I'm a non-native and got 166+167+5.5 in my GRE (granted, not the first attempt - the first time I underestimated how much studying you actually have to do simply to meet the time requirements).

[s1994]

24 minutes ago, IndEnth said:

Getting good GRE scores in my opinion has nothing to do with intelligence, critical thinking...

It's a pure function of how well you're able to study for standardized tests...

Being able to read a few number articles with many details not inside your field (for example, science articles), digest them, and answer the questions have *nothing* to do one's reading and critical thinking skills? 

The key point here, I think, is that we need to let the prospects know that GRE is something that is important -- You do not want to get something below 330 (or 325) if you want to go to a top school. It's going to make your life harder. However, spending another 2 months to study so that you can increase your score from 330 to 335? Maybe they should spend their time reading literature and polishing their SoP.

 

[ultraultra]

9 minutes ago, shane94 said:

Being able to read a few number articles with many details not inside your field (for example, science articles), digest them, and answer the questions have *nothing* to do one's reading and critical thinking skills? 

Nearly everyone could get a perfect or near-perfect score if there was unlimited time to complete the exam. The reason it is so hard to get a great score (especially on the math) is because you have to recall concepts - some of which are useful for quantitative methods (probability, algebra), some of which are basically useless (geometry, strange operators)  - and apply them in an extremely limited period of time. This also applies for recalling random vocabulary that is never used in your field. Succeeding on the GRE is far more about coping under time pressure than it is about "critical thinking." And while there's time pressure in academia, most notably during comprehensive exams, no one is ever going to hold a gun to your head and demand you address every paper reviewer's comments in 2 minutes or less. 

I understand that the schools are looking for any way to cut through the noise. But, coming from Canada, I personally view the intense emphasis on GRE scores as extremely unhealthy. At the beginning of this cycle, two of my other friends were set to apply for PhDs as well. Both of them - who are extremely intelligent, driven, creative women who will someday make excellent scholars - decided not to apply this cycle because they had too much anxiety and panic about the GRE, which came to represent to them their worth as a scholar and their potential for a successful future, because our professors emphasized its salience in the process. I almost followed suit. The idea that one test -- which itself presents major accessibility issues (from both a disability and a class perspective) -- has the potential to overshadow other salient aspects of your file (such as your letters of reference, writing sample, and research experience), is absurd to me. But maybe all of this is because in Canada we don't have the same grade inflation or subsequent standardized test culture, so all of these problems are newer to us than those who took the SAT.

[IndEnth]

Fine, I give you the text questions. But filling in words in a text that are deliberately so difficult that even natives have to study them in lists and calculating the content of a cylinder to me are just very weird measures of a standardized test.

All I'm saying is that

• you can study for it and do well, even if that stuff doesn't come naturally to you;
• you shouldn't take it too hard if it doesn't work the first time - it doesn't say anything about your intelligence or your fitness for grad school; and
• it's worth retaking the test, because you can learn how to do well on it and it's a first-threshold kinda thing that might make people open your file - or not

Edited January 29, 2016 by IndEnth

[s1994]

15 minutes ago, ultraultra said:

Nearly everyone could get a perfect or near-perfect score if there was unlimited time to complete the exam. The reason it is so hard to get a great score (especially on the math) is because you have to recall concepts - some of which are useful for quantitative methods (probability, algebra), some of which are basically useless (geometry, strange operators)  - and apply them in an extremely limited period of time. This also applies for recalling random vocabulary that is never used in your field. Succeeding on the GRE is far more about coping under time pressure than it is about "critical thinking." And while there's time pressure in academia, most notably during comprehensive exams, no one is ever going to hold a gun to your head and demand you address every paper reviewer's comments in 2 minutes or less. 

I understand that the schools are looking for any way to cut through the noise. But, coming from Canada, I personally view the intense emphasis on GRE scores as extremely unhealthy. At the beginning of this cycle, two of my other friends were set to apply for PhDs as well. Both of them - who are extremely intelligent, driven, creative women who will someday make excellent scholars - decided not to apply this cycle because they had too much anxiety and panic about the GRE, which came to represent to them their worth as a scholar and their potential for a successful future, because our professors emphasized its salience in the process. I almost followed suit. The idea that one test -- which itself presents major accessibility issues (from both a disability and a class perspective) -- has the potential to overshadow other salient aspects of your file (such as your letters of reference, writing sample, and research experience), is absurd to me. But maybe all of this is because in Canada we don't have the same grade inflation or subsequent standardized test culture, so all of these problems are newer to us than those who took the SAT.

Is it good to put emphasis on GRE grades? No. And most schools don't. A lot of them do a "cut-off" -- They do not care that much if you ace the test.

Is it important to study GRE and get a good score? Yes. That's how you cope with the entire application process so that you can get your file reviewed carefully. 

Being deterred from applying to Ph.D. programs because your friends are told that GRE is important? Well, if they have the time to perform excellently in class to get fantastic letters, to do research diligently and write a publishable WS, and to have tons of research experiences, why can't they spend a month to periodically study for GRE and get a "threshold" score (around 330)? 

It occurs to me that the unhealthy part of grad school applications is not school's over-emphasis on GRE (and in fact schools don't emphasize on GRE, at least PSC depts), but people not realizing that GRE is a way to simply get your file reviewed, and nothing more. Study for a month and get a 330, and you are set for standardized test portion.

Edited January 29, 2016 by shane94

[Syas]

My scores were 158V-162Q-5W, and I asked the professors that I knew from the top 10 schools whether they would advice me to retake the test.  One said that my scores were good enough for the committee to review my application, and the other said that they review all applications regardless of the GRE scores. Both said that my scores were sufficient and they saw no reason for me to devote more energy on the test instead of preparing the other parts of my application.