GRE score percentile

[Jigsaw123]

Does ETS use the same percentile system as mentioned in the workbook which they sent us in calculating percentiles of actual test or is it different. For example 80 percentile starts at 95-96 questions in the work does this mean in the actual test the same scale is used for calculating the percentile?

[asaprocky]

It's different. Based on the score I got, I expected one percentile based on the workbook, but received a different one when I got my scores. Not too far off, just like 2-3 percentile ranks, but it is still different. Don't fret though, work hard and do well and you'll get the score you'll need 

[TakeruK]

ETS publishes the conversion from scaled score to percentile ranking online. This conversion table is updated every year (on July 1) so that your percentile rank is your scaled score ranking in comparison to a rolling window of the past 3 years of test takers. That is, there should be a date on the table in the workbook and it's probably out of date at this point. 

 

In addition, this means that your percentile ranking might change from year to year as the pool of the "past 3 years" changes!

 

Here is the ETS webpage about this practice: http://www.ets.org/gre/subject/scores/understand/

 

and here is the PDF with the latest conversion table (linked from that page too): http://www.ets.org/s/gre/pdf/gre_guide_table2.pdf

[Jigsaw123]

It's different. Based on the score I got, I expected one percentile based on the workbook, but received a different one when I got my scores. Not too far off, just like 2-3 percentile ranks, but it is still different. Don't fret though, work hard and do well and you'll get the score you'll need 

 

I couldn't understand you point clearly. do u mean that more or less the percentiles will be the same as on the workbook? For example instead of 80 percentile starting from 95-96 it can start from 93-94 or from 92-91? Am I right?

 

ETS publishes the conversion from scaled score to percentile ranking online. This conversion table is updated every year (on July 1) so that your percentile rank is your scaled score ranking in comparison to a rolling window of the past 3 years of test takers. That is, there should be a date on the table in the workbook and it's probably out of date at this point. 

 

In addition, this means that your percentile ranking might change from year to year as the pool of the "past 3 years" changes!

 

Here is the ETS webpage about this practice: http://www.ets.org/gre/subject/scores/understand/

 

and here is the PDF with the latest conversion table (linked from that page too): http://www.ets.org/s/gre/pdf/gre_guide_table2.pdf

 

The sources you included were quiet helpful but is it possible to to know the number of questions which guarantees these percentiles? 

[asaprocky]

I couldn't understand you point clearly. do u mean that more or less the percentiles will be the same as on the workbook? For example instead of 80 percentile starting from 95-96 it can start from 93-94 or from 92-91? Am I right?

 

 

The sources you included were quiet helpful but is it possible to to know the number of questions which guarantees these percentiles? 

Yes the ranges can be either more or less than what you initially saw in the workbook. It just all depends on the test taking pool when you take the exam. 

[Jigsaw123]

Yes the ranges can be either more or less than what you initially saw in the workbook. It just all depends on the test taking pool when you take the exam. 

 

Thank You for the information!

[TakeruK]

I couldn't understand you point clearly. do u mean that more or less the percentiles will be the same as on the workbook? For example instead of 80 percentile starting from 95-96 it can start from 93-94 or from 92-91? Am I right?

 

 

The sources you included were quiet helpful but is it possible to to know the number of questions which guarantees these percentiles? 

 

Yes, the workbook will tell accurately tell you the conversion between # of questions correct and the SCALED SCORE. However, the workbook's table for scaled score to percentile might be outdated. So if you want to estimate your scaled score based on the practice test, the workbook is fine. If you want to estimate your percentile rank based on the test, then use the workbook to compute your scaled score, then use the updated online table to go from scaled score to percentile rank. 

 

Does that make sense? In summary, the conversion from # correct to "scaled score" is something that is unique to any particular version of the test (there are more than one version even on the same test date). That will never change and we only ever get to see this conversion for the practice tests that ETS publishes. The scaled score is supposed to account for any differences in test difficulty from one version to another, so that a scaled score of "800" for example, is supposed to mean the same level of skill no matter what test you wrote.

 

The percentile rank varies over time because it shows how you compare to the recent pool of test-takers, so this accounts for students getting better/worse over time.

[brooksja5]

I have a question for you guys... I recently got a 730 scaled score for CHE GRE. With all my practice exams, that scaled to about 70th percentile. But, it is reported as 57% on my scores. Does that seem like something I should check into?

[TakeruK]

I have a question for you guys... I recently got a 730 scaled score for CHE GRE. With all my practice exams, that scaled to about 70th percentile. But, it is reported as 57% on my scores. Does that seem like something I should check into?

 

Your practice exam scaled-score-to-percentile tables are likely very much out of date. The practice booklet I got (for Physics) had a table that was more than 10 years old! Here is the most recent ETS table: http://www.ets.org/s/gre/pdf/gre_guide_table2.pdf

 

On this updated table, for Chemistry, it looks like a scaled score of 730 should be between 52% and 59%, so I think 57% sounds right!

[M4ss5pec]

Out of curiosity, do adcoms pay much attention to the percentile? Is it more of an afterthought compared to the score itself, simply to monitor score fluctuations?

[DropTheBase]

Out of curiosity, do adcoms pay much attention to the percentile? Is it more of an afterthought compared to the score itself, simply to monitor score fluctuations?

 

The percentile gives the score meaning, since it includes the data from a lot of test-takers (nearly all of them). That information is a lot more valuable than a score from a handful of applicants. My understanding is that the score fluctuates every year because more people take the test, meaning more data, meaning a more accurate percentile.

 

DTB

[TakeruK]

The percentile gives the score meaning, since it includes the data from a lot of test-takers (nearly all of them). That information is a lot more valuable than a score from a handful of applicants. My understanding is that the score fluctuates every year because more people take the test, meaning more data, meaning a more accurate percentile.

 

DTB

 

The number of people's scores used in calculating the percentile would probably not change very much over time. As mentioned above, the percentile score is how you rank in comparison to the people who took the test in the past 3 years. It is a rolling window of 3 years. For example, right now, our percentile scores show how we rank compared to the people who took the test between July 1, 2010 and June 30, 2013 (even if you took the test prior to July 1 2010 or after June 30, 2013). Since population grows over time, the number of people taking the subject tests probably increases each year, but only a small amount. It's not like each year, there's an extra new batch of scores to be added (i.e. the oldest set of scores is removed).

 

However, I would say that because they update yearly, the percentile rank is helpful in showing how the student compares with contemporary peers and accounts for the changing nature of the test and student knowledge.