Last visit was: 24 Apr 2024, 01:29 It is currently 24 Apr 2024, 01:29

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
Tutor
Joined: 22 Oct 2012
Status:Private GMAT Tutor
Posts: 364
Own Kudos [?]: 2330 [39]
Given Kudos: 135
Location: India
Concentration: Economics, Finance
Schools: IIMA (A)
GMAT Focus 1:
735 Q90 V85 DI85
GMAT Focus 2:
735 Q90 V85 DI85
GMAT 1: 780 Q51 V47
GRE 1: Q170 V168
Send PM
Tutor
Joined: 22 Oct 2012
Status:Private GMAT Tutor
Posts: 364
Own Kudos [?]: 2330 [10]
Given Kudos: 135
Location: India
Concentration: Economics, Finance
Schools: IIMA (A)
GMAT Focus 1:
735 Q90 V85 DI85
GMAT Focus 2:
735 Q90 V85 DI85
GMAT 1: 780 Q51 V47
GRE 1: Q170 V168
Send PM
Intern
Intern
Joined: 02 Jun 2013
Posts: 34
Own Kudos [?]: 33 [1]
Given Kudos: 142
Location: United States
Schools: IIMA PGPX'22
GMAT 1: 650 Q49 V30
Send PM
Tutor
Joined: 22 Oct 2012
Status:Private GMAT Tutor
Posts: 364
Own Kudos [?]: 2330 [0]
Given Kudos: 135
Location: India
Concentration: Economics, Finance
Schools: IIMA (A)
GMAT Focus 1:
735 Q90 V85 DI85
GMAT Focus 2:
735 Q90 V85 DI85
GMAT 1: 780 Q51 V47
GRE 1: Q170 V168
Send PM
Re: Data-Driven Answers to Questions about GMAT [#permalink]
Expert Reply
Top Contributor
rahul2013 wrote:
What an article Sir...

Thank you very much :)


Thank you for dropping in a comment! Comments like these go a long way in communicating that the work I do helps others and thus motivate me to continue to produce similar work in the future.
Intern
Intern
Joined: 14 Jul 2019
Posts: 35
Own Kudos [?]: 15 [0]
Given Kudos: 72
Location: India
Schools: IMD '21
Send PM
Re: Data-Driven Answers to Questions about GMAT [#permalink]
ChiranjeevSingh wrote:
What is the maximum Verbal or Quant score can I reach if I get a certain number of questions wrong?

Eye opener CJ. Many thanks for taking the efforts to come upon with this. Would surely apply the takeaways into my overall strategy :)

Here’s the data for Verbal:

Attachment:
ESR Verbal 2.jpg


As you can see, I don’t have a lot of data points (no. of students) for the smallest (2-6 incorrect questions) and the largest (>15 incorrect questions) number of incorrect questions entries. However, I have a decent number of data points for the entries in-between. One thing very clearly comes out is that for the same number of incorrect questions, you can get widely different scores. For example, with 12 questions wrong, you can get the lowest of V22 (i.e. 30 %ile) and the highest of V34 (i.e. 71 %ile), a very significant difference indeed.

Besides, it is also possible that a person with 15 questions wrong ends up getting a higher score than a person with just 8 questions wrong. There was indeed one such case in my data in which a person with 15 questions wrong had a V33 and a person with 8 questions wrong had a V32. You may be wondering how this is possible. How could a person get almost half the number of questions wrong compared to another person and still end up at a lower score? Did the person with V32 get a lot of initial questions wrong? We’ll look at these questions in the later sections of this article.

Here’s the data for Quant:

Attachment:
ESR Quant 2.jpg


We can observe in the above data that three students scored a Q50 with 4 questions wrong and three students who scored 5 questions wrong got a Q49 each. There seems to be some consistency at this end of the table. However, as we go down the table (6-13 questions wrong), we see that there are wide ranges of scores for different number of incorrect questions. For example, one student had a Q32 after getting 10 questions wrong and another who got 10 questions wrong had a Q48, a sea difference in the scores without any difference in the number of questions marked incorrect.

Clearly, the GMAT algorithm is driven a lot by the kind of question a person gets wrong. Logically, a person marking an easy question wrong should be punished more severely than a person getting a difficult question wrong. By getting an easy question wrong, a person is indicating that he cannot be depended on even for correctly solving simpler problems consistently. If you cannot depend on a person for consistently solving easier problems correctly, logically he deserves a low score. On the other hand, on difficult problems in life, a lot of us falter in some way or the other. So, getting a difficult question wrong shouldn’t be as significant a problem as getting an easier question wrong should be.

If you compare the Quant and Verbal tables above, you’ll see that while it is possible to get a Q45 with 13 questions wrong, in Verbal, the highest you get with 13 questions wrong is V34. It is quite clear that getting a V45 is nowhere as easy as getting a Q45. This brings us to our next question.

Why has GMAC made it much more difficult to score high on Verbal than on Quant?

At least a part of the answer seems to lie in their (the GMAC guys’) perception of the difficulty of the questions. Quant questions that are quite easy for an average Indian test taker are considered by GMAC more difficult than Verbal questions that are quite difficult for an average Indian test taker. In other words, a quant question of Medium difficulty level per GMAC is generally an easy question for an Indian test taker, whereas a Verbal question of Medium difficulty level per GMAC is generally a hard question for an Indian test taker.

How do I know this?

From the ESR reports.

What would be the level of difficulty of Quant questions a person who has scored a Q45 must have faced in the exam?

Most Indians would call such questions Easy-to-Medium.

GMAC doesn’t agree.

Here’s the difficulty level of the Quant questions that people who got Q45 faced:

Attachment:
Quant Diff.jpg


As you can see, almost all the Quant questions that these people with Q45 faced were Medium-High. Now, let’s compare it with the level of difficulty of Verbal questions that people with V45 faced (I had only two such data points with me).

Attachment:
Verbal Diff.jpg


Per GMAC, the guys with V45 faced mainly medium-level Verbal questions, and medium-high questions only in the second quarter.

According to you, what would be the difficulty level of the Verbal questions a person with V45 would face?

Would you call that difficulty level ‘medium’?

I believe most of us, non-native speakers, will say no. At V45 level, we’d expect very difficult questions.

However, that is our perception, not GMAC’s perception.

Thus, even if you solve a lot of these questions correctly, you deserve a V45, not V51.

I think it’s easier to understand why there is such a difference in the perceptions of the difficulty level of questions. Indians are generally good at Quant and weak in Verbal. On the other hand, Americans are quite the opposite. Thus, GMAT, a test conducted by a US-based organization, is expected to reflect the perceptions it currently reflects.

To answer the question, GMAC hasn’t deliberately made it difficult to score high on Verbal than on Quant. We feel that GMAC has done so because with Verbal, our comfort is much less than the comfort of an average test-taker in America, in which GMAC is based. The Verbal questions that are pretty hard for us are manageable for them, and the quant questions that are manageable for us are pretty hard for them.

That’s it for this article. In the next article, we’ll answer the following questions:

  • Are the first few questions more important than the remaining ones?
  • Does the GMAT algorithm follow a fixed pattern i.e. giving similar difficulty level of questions for similar performance?

You can find the next and last part of this article series on this link: https://gmatwithcj.com/articles/data-dri ... at-part-3/
Intern
Intern
Joined: 28 Apr 2020
Posts: 19
Own Kudos [?]: 2 [0]
Given Kudos: 56
Send PM
Re: Data-Driven Answers to Questions about GMAT [#permalink]
ChiranjeevSingh wrote:
What is the maximum Verbal or Quant score can I reach if I get a certain number of questions wrong?

Here’s the data for Verbal:

Attachment:
ESR Verbal 2.jpg


As you can see, I don’t have a lot of data points (no. of students) for the smallest (2-6 incorrect questions) and the largest (>15 incorrect questions) number of incorrect questions entries. However, I have a decent number of data points for the entries in-between. One thing very clearly comes out is that for the same number of incorrect questions, you can get widely different scores. For example, with 12 questions wrong, you can get the lowest of V22 (i.e. 30 %ile) and the highest of V34 (i.e. 71 %ile), a very significant difference indeed.

Besides, it is also possible that a person with 15 questions wrong ends up getting a higher score than a person with just 8 questions wrong. There was indeed one such case in my data in which a person with 15 questions wrong had a V33 and a person with 8 questions wrong had a V32. You may be wondering how this is possible. How could a person get almost half the number of questions wrong compared to another person and still end up at a lower score? Did the person with V32 get a lot of initial questions wrong? We’ll look at these questions in the later sections of this article.

Here’s the data for Quant:

Attachment:
ESR Quant 2.jpg


We can observe in the above data that three students scored a Q50 with 4 questions wrong and three students who scored 5 questions wrong got a Q49 each. There seems to be some consistency at this end of the table. However, as we go down the table (6-13 questions wrong), we see that there are wide ranges of scores for different number of incorrect questions. For example, one student had a Q32 after getting 10 questions wrong and another who got 10 questions wrong had a Q48, a sea difference in the scores without any difference in the number of questions marked incorrect.

Clearly, the GMAT algorithm is driven a lot by the kind of question a person gets wrong. Logically, a person marking an easy question wrong should be punished more severely than a person getting a difficult question wrong. By getting an easy question wrong, a person is indicating that he cannot be depended on even for correctly solving simpler problems consistently. If you cannot depend on a person for consistently solving easier problems correctly, logically he deserves a low score. On the other hand, on difficult problems in life, a lot of us falter in some way or the other. So, getting a difficult question wrong shouldn’t be as significant a problem as getting an easier question wrong should be.

If you compare the Quant and Verbal tables above, you’ll see that while it is possible to get a Q45 with 13 questions wrong, in Verbal, the highest you get with 13 questions wrong is V34. It is quite clear that getting a V45 is nowhere as easy as getting a Q45. This brings us to our next question.

Why has GMAC made it much more difficult to score high on Verbal than on Quant?

At least a part of the answer seems to lie in their (the GMAC guys’) perception of the difficulty of the questions. Quant questions that are quite easy for an average Indian test taker are considered by GMAC more difficult than Verbal questions that are quite difficult for an average Indian test taker. In other words, a quant question of Medium difficulty level per GMAC is generally an easy question for an Indian test taker, whereas a Verbal question of Medium difficulty level per GMAC is generally a hard question for an Indian test taker.

How do I know this?

From the ESR reports.

What would be the level of difficulty of Quant questions a person who has scored a Q45 must have faced in the exam?

Most Indians would call such questions Easy-to-Medium.

GMAC doesn’t agree.

Here’s the difficulty level of the Quant questions that people who got Q45 faced:

Attachment:
Quant Diff.jpg


As you can see, almost all the Quant questions that these people with Q45 faced were Medium-High. Now, let’s compare it with the level of difficulty of Verbal questions that people with V45 faced (I had only two such data points with me).

Attachment:
Verbal Diff.jpg


Per GMAC, the guys with V45 faced mainly medium-level Verbal questions, and medium-high questions only in the second quarter.

According to you, what would be the difficulty level of the Verbal questions a person with V45 would face?

Would you call that difficulty level ‘medium’?

I believe most of us, non-native speakers, will say no. At V45 level, we’d expect very difficult questions.

However, that is our perception, not GMAC’s perception.

Thus, even if you solve a lot of these questions correctly, you deserve a V45, not V51.

I think it’s easier to understand why there is such a difference in the perceptions of the difficulty level of questions. Indians are generally good at Quant and weak in Verbal. On the other hand, Americans are quite the opposite. Thus, GMAT, a test conducted by a US-based organization, is expected to reflect the perceptions it currently reflects.

To answer the question, GMAC hasn’t deliberately made it difficult to score high on Verbal than on Quant. We feel that GMAC has done so because with Verbal, our comfort is much less than the comfort of an average test-taker in America, in which GMAC is based. The Verbal questions that are pretty hard for us are manageable for them, and the quant questions that are manageable for us are pretty hard for them.

That’s it for this article. In the next article, we’ll answer the following questions:

  • Are the first few questions more important than the remaining ones?
  • Does the GMAT algorithm follow a fixed pattern i.e. giving similar difficulty level of questions for similar performance?

You can find the next and last part of this article series on this link: https://gmatwithcj.com/articles/data-dri ... at-part-3/


Nice article sir . Please share the link for another article which mentions about the remaining questions .
Manager
Manager
Joined: 26 Mar 2021
Posts: 113
Own Kudos [?]: 31 [0]
Given Kudos: 36
Location: India
Concentration: General Management, Strategy
Schools: IIMB EPGP'23
GMAT 1: 640 Q50 V25
Send PM
Re: Data-Driven Answers to Questions about GMAT [#permalink]
Hats Off to the hard work !!! A must-read article. answers many imp. queries.

I have few assumptions as follows, please shed light on them:

1: Answering the first 5 -7 questions right in any section (quant/verbal) increases the chances of securing a high score as it increases the level of difficulty and hence the weights of the questions asked.

2: Since in an RC, questions are not adaptive hence it becomes critical to answer maximum questions right.
Tutor
Joined: 22 Oct 2012
Status:Private GMAT Tutor
Posts: 364
Own Kudos [?]: 2330 [0]
Given Kudos: 135
Location: India
Concentration: Economics, Finance
Schools: IIMA (A)
GMAT Focus 1:
735 Q90 V85 DI85
GMAT Focus 2:
735 Q90 V85 DI85
GMAT 1: 780 Q51 V47
GRE 1: Q170 V168
Send PM
Re: Data-Driven Answers to Questions about GMAT [#permalink]
Expert Reply
bansalgaurav wrote:
Hats Off to the hard work !!! A must-read article. answers many imp. queries.

I have few assumptions as follows, please shed light on them:

1: Answering the first 5 -7 questions right in any section (quant/verbal) increases the chances of securing a high score as it increases the level of difficulty and hence the weights of the questions asked.

2: Since in an RC, questions are not adaptive hence it becomes critical to answer maximum questions right.


Thank you, Gaurav! :)

1. That's true to an extent. The penalty for getting easy and medium questions wrong is higher than that for getting hard questions wrong. Since a test usually starts with medium questions and will offer easy questions if you get medium questions wrong, getting a few questions wrong in the beginning is the same as getting easy and medium questions wrong. Thus, a higher penalty.

2. I don't think RC questions are not adaptive. If you observe RC passages in OG, they regularly have 6-8 questions of varying difficulty levels. Thus, I believe it is possible for RC questions to be adaptive even within a passage. In addition, from an overall perspective, I'd expect you to get harder RC passages if you're doing well on the test.

- CJ
GMAT Club Legend
GMAT Club Legend
Joined: 15 Jul 2015
Posts: 5179
Own Kudos [?]: 4653 [1]
Given Kudos: 629
Location: India
GMAT Focus 1:
715 Q83 V90 DI83
GMAT 1: 780 Q50 V51
GRE 1: Q170 V169
Send PM
Re: Data-Driven Answers to Questions about GMAT [#permalink]
1
Kudos
Expert Reply
ChiranjeevSingh wrote:
2. I don't think RC questions are not adaptive. If you observe RC passages in OG, they regularly have 6-8 questions of varying difficulty levels. Thus, I believe it is possible for RC questions to be adaptive even within a passage. In addition, from an overall perspective, I'd expect you to get harder RC passages if you're doing well on the test.

To add to the discussion, GMAT RC is not question-level adaptive. An RC passage may be presented with different combinations of questions, but every set of questions is fixed. Once the set has been chosen to be presented to a test taker, it can't be changed.
Intern
Intern
Joined: 17 Jan 2018
Posts: 31
Own Kudos [?]: 10 [0]
Given Kudos: 808
Location: India
Send PM
Re: Data-Driven Answers to Questions about GMAT [#permalink]
Great article, thanks for sharing this - definitely deserves more kudos/views!
Tutor
Joined: 22 Oct 2012
Status:Private GMAT Tutor
Posts: 364
Own Kudos [?]: 2330 [0]
Given Kudos: 135
Location: India
Concentration: Economics, Finance
Schools: IIMA (A)
GMAT Focus 1:
735 Q90 V85 DI85
GMAT Focus 2:
735 Q90 V85 DI85
GMAT 1: 780 Q51 V47
GRE 1: Q170 V168
Send PM
Re: Data-Driven Answers to Questions about GMAT [#permalink]
Expert Reply
AjiteshArun wrote:
ChiranjeevSingh wrote:
2. I don't think RC questions are not adaptive. If you observe RC passages in OG, they regularly have 6-8 questions of varying difficulty levels. Thus, I believe it is possible for RC questions to be adaptive even within a passage. In addition, from an overall perspective, I'd expect you to get harder RC passages if you're doing well on the test.

To add to the discussion, GMAT RC is not question-level adaptive. An RC passage may be presented with different combinations of questions, but every set of questions is fixed. Once the set has been chosen to be presented to a test taker, it can't be changed.


Hi Ajitesh,

Would you mind sharing the basis of what you said?

Regards,
Chiranjeev
GMAT Club Legend
GMAT Club Legend
Joined: 15 Jul 2015
Posts: 5179
Own Kudos [?]: 4653 [2]
Given Kudos: 629
Location: India
GMAT Focus 1:
715 Q83 V90 DI83
GMAT 1: 780 Q50 V51
GRE 1: Q170 V169
Send PM
Data-Driven Answers to Questions about GMAT [#permalink]
2
Kudos
Expert Reply
ChiranjeevSingh wrote:
Hi Ajitesh,

Would you mind sharing the basis of what you said?

Regards,
Chiranjeev

Hi Chiranjeev,

This is based on presentations made by and conversations with Lawrence Rudner (ex-VP, R&D, GMAC) and (later) Fanmin Guo (Dr. Rudner's successor, till ~2017).
GMAT Club Bot
Data-Driven Answers to Questions about GMAT [#permalink]

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