Last visit was: 24 Apr 2024, 22:37 It is currently 24 Apr 2024, 22:37

Close
GMAT Club Daily Prep
Thank you for using the timer - this advanced tool can estimate your performance and suggest more practice questions. We have subscribed you to Daily Prep Questions via email.

Customized
for You

we will pick new questions that match your level based on your Timer History

Track
Your Progress

every week, we’ll send you an estimated GMAT score based on your performance

Practice
Pays

we will pick new questions that match your level based on your Timer History
Not interested in getting valuable practice questions and articles delivered to your email? No problem, unsubscribe here.
Close
Request Expert Reply
Confirm Cancel
SORT BY:
Date

Is the GMAT ESR feature useful?

You may select 1 option
Manager
Manager
Joined: 18 Jan 2012
Posts: 135
Own Kudos [?]: 130 [15]
Given Kudos: 41
Send PM
Founder
Founder
Joined: 04 Dec 2002
Posts: 37305
Own Kudos [?]: 72873 [1]
Given Kudos: 18863
Location: United States (WA)
GMAT 1: 750 Q49 V42
GPA: 3
Send PM
User avatar
Intern
Intern
Joined: 08 Sep 2014
Posts: 38
Own Kudos [?]: 234 [0]
Given Kudos: 12
Concentration: Economics, Entrepreneurship
Schools: NUS '18
GMAT Date: 03-17-2015
Send PM
Founder
Founder
Joined: 04 Dec 2002
Posts: 37305
Own Kudos [?]: 72873 [1]
Given Kudos: 18863
Location: United States (WA)
GMAT 1: 750 Q49 V42
GPA: 3
Send PM
Re: 7 reason why GMAT ESR isn't useful - Detailed Analysis [#permalink]
1
Bookmarks
Expert Reply
rajthakkar wrote:
I am not able to understand what my data tells to me..there was a noise prob at center and I have got another retake from Pearson

My stats are
CR 28th percentile(21 scaled score)
Rc 95th percentile ( 51 scaled score)
SC 57TH percentile (30 scaled score)

Overall 31 scaled score.

Does it mean I got almost all questions correct in RC section?

Posted from my mobile device



I don't think it means you have answered all questions correctly... but it does look like you have managed to score on RC quite well... Is it your strong suit or just how luck worked out?

P.S. The RC passages/questions tend to have non-adaptive scoring and perhaps this reflects it since individual RC questions are supposedly not adaptive.
User avatar
Intern
Intern
Joined: 08 Sep 2014
Posts: 38
Own Kudos [?]: 234 [0]
Given Kudos: 12
Concentration: Economics, Entrepreneurship
Schools: NUS '18
GMAT Date: 03-17-2015
Send PM
Re: 7 reason why GMAT ESR isn't useful - Detailed Analysis [#permalink]
Actually there was so much disturbance so I took longer to read passages.One passage at the middle was quite tricky and 4-5 para long with densed info. Generally I have higher accuracy in SC rather than in RC and generally get 3-4 questions wrong in the RC section MGMAT tests.

It seems ESR doesn't help much.I had to rush through 4-5 questions at the end which contained SC and CR questions.Getting them all wrong could have caused individual CR and SC scores/percentiles to drop.

Posted from my mobile device
SVP
SVP
Joined: 20 Mar 2014
Posts: 2362
Own Kudos [?]: 3626 [0]
Given Kudos: 816
Concentration: Finance, Strategy
GMAT 1: 750 Q49 V44
GPA: 3.7
WE:Engineering (Aerospace and Defense)
Send PM
Re: 7 reason why GMAT ESR isn't useful - Detailed Analysis [#permalink]
bb wrote:
rajthakkar wrote:
I am not able to understand what my data tells to me..there was a noise prob at center and I have got another retake from Pearson

My stats are
CR 28th percentile(21 scaled score)
Rc 95th percentile ( 51 scaled score)
SC 57TH percentile (30 scaled score)

Overall 31 scaled score.

Does it mean I got almost all questions correct in RC section?

Posted from my mobile device



I don't think it means you have answered all questions correctly... but it does look like you have managed to score on RC quite well... Is it your strong suit or just how luck worked out?

P.S. The RC passages/questions tend to have non-adaptive scoring and perhaps this reflects it since individual RC questions are supposedly not adaptive.


bb,

I do think that a scaled score of 51 on any of the sections, means that all the questions in that particular sub section were answered correctly. This is based on my personal observation from GMATPREP and I am assuming that there is 0% difference in how the algorithm works (and interpretation of scaled scores) for both GMATPREP and actual GMAT. I got a 51 whenever I had answered all questions correctly in that sub section, be it DS or CR or RC.
GMAT Tutor
Joined: 24 Jun 2008
Posts: 4128
Own Kudos [?]: 9242 [6]
Given Kudos: 91
 Q51  V47
Send PM
Re: 7 reason why GMAT ESR isn't useful - Detailed Analysis [#permalink]
6
Kudos
Expert Reply
There are a couple of misconceptions in this thread.

bb wrote:
P.S. The RC passages/questions tend to have non-adaptive scoring and perhaps this reflects it since individual RC questions are supposedly not adaptive.


I'm not exactly what sure what this means, but RC questions are scored like any other questions. The one thing that makes RC different from CR or SC is that you're given an entire set of questions which, on average, ought to be roughly at your level (though some will be hard for you, some easy). The question sets are still selected adaptively - you'll get harder passages if you're doing well - and are still scored the same way as other questions.

And I think the explanation below misunderstands how the ESR is calculating scores by question type:

arun@crackverbal wrote:

Reason 6: Percentile scores for individual topics with Verbal & Quant


The biggest flaw here is that the GMAT adaptive algorithm works across all the sub-sections. So even if you are really good in SC you might end up not seeing very high %ile scores because some other section (say CR or RC) pulled you down.

For example, each time the adaptive algorithm puts you in a higher bracket because you answered some SC or CR questions correctly, your next RC passage might pull you down. So net-net in all sections you see around the same %ile ballpark though RC is really your problem area.


I gather you're suggesting that by doing well in SC/CR, a test taker gets harder RC questions, which will hurt his or her score. That's not how adaptive scoring works. On an adaptive test, your score will be the same whether you see tons of hard questions or tons of easy questions. Scoring is all based on probabilities: the algorithm knows the probability a test taker of each ability level will get a right answer to a question. If you get a test full of 700-level questions, and answer about 60% of them correctly, you'll get a 700. If you get a test full of 400-level questions, you'd need to get way more than 60% of them right to get a 700. And if you had a test full of 800-level questions, you could get less than 60% right to get a 700.

Difficultly level is already factored into the algorithm, so your performance on SC or CR doesn't affect your score on RC - it will affect your 'hit rate' on RC, but your score is not based on your 'hit rate'.

I agree that the ESR does not provide a lot of useful information, and the timing information is particularly useless for test takers who guess to finish a section. The most interesting information is the breakdown by question type (the scores for SC, CR and RC, for example). But that information is problematic for a different reason, not identified above: the sample size of questions from which the ESR is producing a score is too small to give a precise estimate of ability. We all know that on a full-length GMAT, the standard error is about 30 points - the GMAT doesn't always give a score exactly equal to your ability, and any score should correctly be considered to have a 'margin of error'. When you cut down the number of questions into batches of 10 or 15, the algorithm has a lot less information to go on. The margin of error goes way, way up, and the score you get on RC-only or CR-only might be quite far from your true level.

That's also why those diagnostic reports on company tests that break down your performance by subject and question type are often almost meaningless, especially when those reports divide up the test into very small batches of questions. I've seen some tests that try to judge how good a test taker is at Geometry from a sample of three questions, and that's just an absurd thing to do, mathematically speaking. So if, on an ESR, you have a very big difference in scores on different question types, that difference is probably meaningful. But if scores are fairly close, it may just be luck that caused one score to be higher than another. If you really want to assess your level in different question types, it's much better to assess yourself by attempting larger pools of official questions of similar difficulty - you really can't use a single test to assess yourself that way.
Tutor
Joined: 10 Jul 2015
Status:Expert GMAT, GRE, and LSAT Tutor / Coach
Affiliations: Harvard University, A.B. with honors in Government, 2002
Posts: 1178
Own Kudos [?]: 2412 [1]
Given Kudos: 272
Location: United States (CO)
Age: 44
GMAT 1: 770 Q47 V48
GMAT 2: 730 Q44 V47
GMAT 3: 750 Q50 V42
GMAT 4: 730 Q48 V42 (Online)
GRE 1: Q168 V169

GRE 2: Q170 V170
Send PM
7 reason why GMAT ESR isn't useful - Detailed Analysis [#permalink]
1
Kudos
Expert Reply
Top Contributor
I find GMAT ESRs to be incredibly useful — in fact, I'm rather obsessed with them. From the ESRs we've discovered exactly how many questions are counted on each section (30 on Verbal and 28 on Quant) and each quarter-section, for example, and we have also been able to infer the precise number of experimental questions on the GMAT (formerly 23, now 12). ESRs also provide excellent information on pacing, question difficulty, and leveraging the scoring algorithm.

Could the ESRs be more specific? Of course. Breaking down each section into quarters and only providing average question difficulty levels only tells us so much. In a perfect world, I would love to actually be able to actually see all the questions! But let's not forget that pre-ESRs, we had nothing to analyze other than a score report and our vague memory of the test, so beggars can't be choosers.

Originally posted by mcelroytutoring on 25 Jan 2018, 20:54.
Last edited by mcelroytutoring on 29 Jun 2022, 12:15, edited 1 time in total.
Manager
Manager
Joined: 11 Sep 2019
Posts: 96
Own Kudos [?]: 56 [0]
Given Kudos: 55
Location: India
Schools: ISB'22
Send PM
Re: 7 reason why GMAT ESR isn't useful - Detailed Analysis [#permalink]
arun@crackverbal

Hi Arun, this is the post from 2015. Is this still a valid discussion. My GMAT 1st attempt is 490 and I am going to retake. Before buying ESR I tried to search about ESR and found this article.
I was already not sure about ESR, and your article helped me, but I think the latest ESRs have more observations/data points than the ESR presented by you here.

Could you please help me figure it out. Is it possible for you to show up for Video or Written Consultation?

My story of 490 is right here - https://gmatclub.com/forum/poorest-gmat-score-in-first-attempt-seeking-help-in-prep-next-attempt-314050.html#p2437126
Director
Director
Joined: 05 Jan 2020
Status:Admitted to IIM Shillong (PGPEx 2023-24)
Affiliations: CFA Institute; ICAI; BCAS
Posts: 588
Own Kudos [?]: 554 [0]
Given Kudos: 694
Location: India
WE 2: EA to CFO (Consumer Products)
GPA: 3.78
WE:Corporate Finance (Commercial Banking)
Send PM
Re: 7 reason why GMAT ESR isn't useful - Detailed Analysis [#permalink]
IanStewart wrote:
There are a couple of misconceptions in this thread.

bb wrote:
P.S. The RC passages/questions tend to have non-adaptive scoring and perhaps this reflects it since individual RC questions are supposedly not adaptive.


I'm not exactly what sure what this means, but RC questions are scored like any other questions. The one thing that makes RC different from CR or SC is that you're given an entire set of questions which, on average, ought to be roughly at your level (though some will be hard for you, some easy). The question sets are still selected adaptively - you'll get harder passages if you're doing well - and are still scored the same way as other questions.

bb

I agree with IanStewart that RC passages/questions do have adaptive scoring as evident from no. of ques asked vs available on GMAT Prep passages.
3-4 ques are actually asked for any passage but available no. of ques are mostly more than that.
Example 1- 3 ques were asked in my mock test but there are 5 ques available
Example 2- 4 ques were asked in my mock test but there are 6 ques available

It is possible that not all passages may have more available ques than actually asked, so they may not be adaptive at ques level; but many do have.

Also, please refer this post of ChiranjeevSingh

KarishmaB ExpertsGlobal5 AjiteshArun GMATNinja AndrewN MartyTargetTestPrep egmat EMPOWERgmatVerbal souvik101990 Sajjad1994 RonPurewal arun@crackverbal

What's your take on this?
Volunteer Expert
Joined: 16 May 2019
Posts: 3512
Own Kudos [?]: 6858 [1]
Given Kudos: 500
Re: 7 reason why GMAT ESR isn't useful - Detailed Analysis [#permalink]
1
Kudos
Expert Reply
Kushchokhani wrote:
IanStewart wrote:
There are a couple of misconceptions in this thread.

bb wrote:
P.S. The RC passages/questions tend to have non-adaptive scoring and perhaps this reflects it since individual RC questions are supposedly not adaptive.


I'm not exactly what sure what this means, but RC questions are scored like any other questions. The one thing that makes RC different from CR or SC is that you're given an entire set of questions which, on average, ought to be roughly at your level (though some will be hard for you, some easy). The question sets are still selected adaptively - you'll get harder passages if you're doing well - and are still scored the same way as other questions.

bb

I agree with IanStewart that RC passages/questions do have adaptive scoring as evident from no. of ques asked vs available on GMAT Prep passages.
3-4 ques are actually asked for any passage but available no. of ques are mostly more than that.
Example 1- 3 ques were asked in my mock test but there are 5 ques available
Example 2- 4 ques were asked in my mock test but there are 6 ques available

It is possible that not all passages may have more available ques than actually asked, so they may not be adaptive at ques level; but many do have.

Also, please refer this post of ChiranjeevSingh

What's your take on this?

Hello, Kushchokhani. That is quite a panel of Experts you have called on. I suspect that RC passages are adaptive in the sense that a given set of questions will be pre-selected based on test-taker performance. In a sense, then—if my hunch is grounded in reality—bb and IanStewart could both be correct. An incorrect answer to a given question within an RC set (say, question two) might not affect the next question that pops up on the screen for that passage, but it probably would affect the difficulty of subsequent questions that would appear in that section of the exam. Otherwise, it would not be an adaptive test, and RC questions could be thought of as free passes. I agree with you that there could be certain passages that only ever contained 3-4 questions, and that these question sets, after thorough statistical analysis on the part of GMAC™, might show up as is for a test-taker demonstrating an ability that fell in line with the statistics for that passage and set of questions.

Several tutors and test prep professionals who have been in the industry for some time have attended presentations that GMAC™ has given in which the exam is discussed in more detail, and some of them have also read through and cited reports that GMAC™ has released over the years. I have neither attended such presentations nor read the literature, so I would defer to someone who has.

For as much talk as there is on the scoring algorithm, I still feel compelled to point out that understanding exactly how it works will not make someone any better of a test-taker. You have to answer questions correctly, plain and simple, to earn the best score you can.

Thank you for thinking to ask for my view on the matter, even if I think I am less qualified to offer an informed opinion.

- Andrew
GMAT Club Legend
GMAT Club Legend
Joined: 15 Jul 2015
Posts: 5181
Own Kudos [?]: 4653 [3]
Given Kudos: 631
Location: India
GMAT Focus 1:
715 Q83 V90 DI83
GMAT 1: 780 Q50 V51
GRE 1: Q170 V169
Send PM
7 reason why GMAT ESR isn't useful - Detailed Analysis [#permalink]
3
Kudos
Expert Reply
@Kushchokhani wrote:
I agree with IanStewart that RC passages/questions do have adaptive scoring as evident from no. of ques asked vs available on GMAT Prep passages.

Hi Kushchokhani,

As AndrewN and IanStewart have mentioned, RC is not question-level adaptive.

@Kushchokhani wrote:
Also, please refer this post of ChiranjeevSingh

Are you referring to this part of that post?
ChiranjeevSingh wrote:
Thus, I believe it is possible for RC questions to be adaptive even within a passage.

I respect ChiranjeevSingh's attempt to get answers from available data, but that's not how it works on the GMAT exam. It is a fact, confirmed by GMAC, that GMAT RC is not question-level adaptive.
Director
Director
Joined: 05 Jan 2020
Status:Admitted to IIM Shillong (PGPEx 2023-24)
Affiliations: CFA Institute; ICAI; BCAS
Posts: 588
Own Kudos [?]: 554 [0]
Given Kudos: 694
Location: India
WE 2: EA to CFO (Consumer Products)
GPA: 3.78
WE:Corporate Finance (Commercial Banking)
Send PM
Re: 7 reason why GMAT ESR isn't useful - Detailed Analysis [#permalink]
AndrewN AjiteshArun IanStewart

Thank you so much for your response.

Just to be clear about your explanations, if a passage has say 6 ques, then 4 ques will be pre-determined basis our level of difficulty at which the passage is being asked. So, if we are at say 650 difficulty level when RC appears, then 4 out of those 6 ques will be around that level and the other 2 outlier difficulty level ques will not be asked (pre-determined at the time of first RC ques). Or I may have misinterpreted your explanations which may mean that all those 4 out of 6 ques will be random and just the passage itself will be 650-level?
Volunteer Expert
Joined: 16 May 2019
Posts: 3512
Own Kudos [?]: 6858 [2]
Given Kudos: 500
Re: 7 reason why GMAT ESR isn't useful - Detailed Analysis [#permalink]
2
Kudos
Expert Reply
Kushchokhani wrote:
AndrewN AjiteshArun IanStewart

Thank you so much for your response.

Just to be clear about your explanations, if a passage has say 6 ques, then 4 ques will be pre-determined basis our level of difficulty at which the passage is being asked. So, if we are at say 650 difficulty level when RC appears, then 4 out of those 6 ques will be around that level and the other 2 outlier difficulty level ques will not be asked (pre-determined at the time of first RC ques). Or I may have misinterpreted your explanations which may mean that all those 4 out of 6 ques will be random and just the passage itself will be 650-level?

Hello again, Kushchokhani. Again, I would defer to Ian and Ajitesh, since, from their earlier posts, they appear to be some of those seasoned veterans I mentioned in my earlier response who have looked beyond the OG for information. I think it would be difficult to craft a passage in which 3-4 questions were all of similar difficulty. I suspect the process is a little different. Remember, an ESR reveals just the average difficulty of the questions in a given quarter of a section. A set of questions attached to a passage in RC is likely no different. If we think of a hypothetical passage with a set of 6 questions, only 1 of which would fall within an Easy spectrum and another that might fall within a Hard spectrum, the other 4 questions would fit somewhere in the middle, in Medium territory. (Not every Medium question is of the same difficulty.) Say that two test-takers encountered the same passage with three questions attached. One test-taker who was performing really well might get a mix of M-M-M or M-M-H (in any order), while the other, who might not be performing as well, might also get M-M-M or, perhaps, E-M-M (again, in any order). Depending on whatever the algorithm threw in front of them, their question set could be Medium-Low or Medium-High, or anything in between. There is no such thing, in my mind, as a 650-level passage, since the data are all attached to individual questions. There are also rare occasions in which someone simply is not presented the upper-level questions that would fall in line with their performance, as was the case with one test-taker in 2017 who earned a V50 with zero missed questions. (The score was later upgraded to a 51 after he contacted GMAC™.) The bottom line: just focus on answering questions correctly, without letting other nonsense get in the way.

Thank you for following up.

- Andrew
Tutor
Joined: 16 Oct 2010
Posts: 14817
Own Kudos [?]: 64905 [1]
Given Kudos: 426
Location: Pune, India
Send PM
Re: 7 reason why GMAT ESR isn't useful - Detailed Analysis [#permalink]
1
Kudos
Expert Reply
Kushchokhani wrote:
IanStewart wrote:
There are a couple of misconceptions in this thread.

bb wrote:
P.S. The RC passages/questions tend to have non-adaptive scoring and perhaps this reflects it since individual RC questions are supposedly not adaptive.


I'm not exactly what sure what this means, but RC questions are scored like any other questions. The one thing that makes RC different from CR or SC is that you're given an entire set of questions which, on average, ought to be roughly at your level (though some will be hard for you, some easy). The question sets are still selected adaptively - you'll get harder passages if you're doing well - and are still scored the same way as other questions.

bb

I agree with IanStewart that RC passages/questions do have adaptive scoring as evident from no. of ques asked vs available on GMAT Prep passages.
3-4 ques are actually asked for any passage but available no. of ques are mostly more than that.
Example 1- 3 ques were asked in my mock test but there are 5 ques available
Example 2- 4 ques were asked in my mock test but there are 6 ques available

It is possible that not all passages may have more available ques than actually asked, so they may not be adaptive at ques level; but many do have.

Also, please refer this post of ChiranjeevSingh

KarishmaB ExpertsGlobal5 AjiteshArun GMATNinja AndrewN MartyTargetTestPrep egmat EMPOWERgmatVerbal souvik101990 Sajjad1994 RonPurewal arun@crackverbal

What's your take on this?



Kushchokhani

I think both bb and Ian are saying the same thing - RC passages are not question level adaptive i.e. the test will not choose q. no 2 for you based on your answer to q. no. 1 on the RC passage. The test will pre-select the passage and the 3-4 questions it will give you but this selection will happen based on your previous performance (so in that sense, it is adaptive).
There may be 6 available questions for an RC passage but the test may choose 3 of those for you depending on your previous performance (prior to the RC).
This is my understanding too.
GMAT Club Bot
Re: 7 reason why GMAT ESR isn't useful - Detailed Analysis [#permalink]

Powered by phpBB © phpBB Group | Emoji artwork provided by EmojiOne