Inside the Mind of a GMAT Test Writer
If you have spent any time preparing for the GMAT, you already know that it is a Computer Adaptive Test ("CAT"). CATs operate by basing the difficulty level of each subsequent question on the difficulty level of the previous question and whether you got it right or wrong. So, for example, if you answer a question correctly, your next question will be even harder. If you answer it incorrectly, your next question will be somewhat easier. But how do the test writers at ETS (the organization that administers the GMAT) determine which questions are harder than others?
First, ETS engages in a process called "normalization," wherein all freshly written questions are tested by actual test takers to determine what percentage answer the questions correctly. If too many people answer correctly, the question may need to be toughened up. If too few people answer correctly, the question may need to be dumbed down. ETS is looking to assemble a pool of questions that covers a range of difficulty, from cakewalk to mind-bending. And the test takers help them do so.
How does ETS find these test takers? Easy. Everyone who takes the GMAT will end up answering approximately 10 unscored "experimental" math questions and 10 unscored "experimental" verbal questions. These questions are interspersed with the actual, scored questions with no way to identify them as experimental. So you, the test taker, do ETS's dirty work.
Second, the writers at ETS have a general sense of what makes a "500-level" question, a 600-level, 700-level, etc. (By the way, when we talk about a "500-level" question, for example, we mean that if that question represents the hardest level you reach on the exam, you will score in the 500 range.) Since each test is designed to evaluate proficiency in the same range of topics, the writers have to come up with ways to test the same concepts at different levels of difficulty. For example, let's say the writers want to test your understanding of percentages. Consider the following progression:
What is 12.5% of 48?
If 6 is x% of 48, what is the value of x?
If x/48 = 6.25/50 , what is the value of x?
There are 3 marbles in one bag, 2 blue and 1 red. There are 16 marbles in another bag, 1 yellow, 3 blue, 5 green, and 7 red. If one marble is selected from each bag at random, what is the probability that both marbles are blue?
There are 48 men in a certain rowing club. 5/12 of the men are handsome but not rich. 11/24 of the men are rich. What percentage of the men are neither rich nor handsome?
Notice that the answer to all these questions is basically the same, but the amount of thought needed to solve them varies considerably. All ETS does is add layers of information to the question to hide the core concept being tested. As you progress in difficulty, ETS is less interested in whether you can perform basic calculations and more interested in whether you can peel away the layers to get to the core.
As you prepare for the GMAT, you need to keep this in mind. Every difficult question is, at heart, a puffed-up version of a simpler question. Practice dissecting math questions to see what the underlying concept is, then retrace your steps and see how ETS layered the information to make the question more difficult. If you can master this skill, or become even moderately proficient, your performance should improve accordingly.
Consider the following question, for example:
What is the value of x?
We have no way of knowing the value of x because we have been given no information about it. In Data Sufficiency problems you are given 2 pieces of information (called "statements") and asked to determine whether the statements (either individually or together) provide enough information to answer the question. How do GMAT test-writers come up with difficult Data Sufficiency problems? They take an easy piece of information and obscure it by adding layers of complexity.
In order to answer our question (What is the value of x?), the test-writers could provide you with a very straightforward statement. For example:
X = 2
This would be absurdly easy, so the test writers have to somehow tell you that x = 2 without stating it outright. What if we had the following statement:
A little harder, but not much. Let's try:
This statement can be factored into (x - 2)(x - 2) = 0, which tells us that the value of x must be 2. This is a little tougher to decipher, but still not at an especially high level of GMAT difficulty. What if we were given the following statement:
, where x is prime and y is even.
If y is even, then must be even as well. Since, it must be true that 
is also even. If is even, x itself must be even. Since x is both even and prime, it must be true that x = 2, because 2 is the only even prime.
Compare the statement, where x is prime and y is even, to the statement x = 2. The statements provide the same information, but one is unquestionably more difficult than the other.
In Data Sufficiency, the level of difficulty is not wholly dependent on the difficulty of the concept; it depends in part on the skill with which the test writer conceals the necessary information. As you study, you should note any questions where the information was cleverly hidden and work backwards through the levels to see how the writers were able to mislead you. Many of their tricks appear over and over in questions in the Official Guide. If you learn to spot them, you will have an enormous advantage over other test takers.
For example, let's consider the following sentence:
The dog are friendly.
It does not take much effort to see that this sentence is flawed: the noun ("dog") is singular but the verb ("are") is plural. This would be much too easy for the GMAT, so the test writers must camouflage the error. One simple way to do so is to insert a lot of unnecessary verbiage between the noun and verb. We call this verbiage the "middleman". For example:
The dog, which was one of two puppies rescued from the shelter, are friendly.
The subject-verb flaw is a little harder to see now, but still readily apparent on a first read. If we take out the "middleman" (the intervening clause), we are back to the original sentence ("The dog are friendly"). Notice, however, that the writers have inserted a plural noun ("puppies") in the new clause so that you have plurality on the brain when you read "are friendly." If you are already thinking in plural terms, you are much less likely to spot the error. Even on a visual level, the subject of the sentence ("dog") is so far removed from the verb ("are") that the eye quickly alights on "puppies" as a possible subject for the plural "are". As tricky as this may already seem, the writers can put yet another kink in the rope:
Two puppies were rescued from the shelter, but neither of them are friendly.
The error in this sentence is significantly less apparent than those in the previous examples, though it is of the same basic type: subject-verb disagreement. Here the subject is "neither of them", which is singular (think of it as "neither one of them"). The verb, however, is still plural ("are"). The saga of the mismatched subject and verb goes on. Can the writers make the problem even harder to spot? Sure. Let's take a look at the following example:
Neither of the two puppies that were rescued from the shelter are friendly.
If you compare this sentence with the previous examples, the error seems almost completely camouflaged. We can see that the subject is "neither (one) of the two puppies", which is singular. But the verb "are" is still plural. As was the case with the math questions we looked at earlier in the month, the difficulty is layered. We have gone from "The dog are friendly" to "Neither of the two puppies that were rescued from the shelter are friendly" in a few steps, obscuring the central subject-verb issue along the way. Breaking sentences down into their component parts and analyzing their relationships is the key to success in Sentence Correction.
But each argument must start with a conclusion and at least one piece of evidence that the author thinks will justify the conclusion. Let's take a basic example:
Some people have died after ingesting cyanide.
Therefore, cyanide is deadly.
Is this a persuasive argument? Not really. The fact that some people have died after ingesting cyanide is not, by itself, enough to prove that the cyanide was the cause of their deaths. This gap in logic is readily apparent, so the test writers need to add information to divert your attention away from the faulty logic. For example,
At a certain party, some people drank wine laced with cyanide and some drank unadulterated wine. Those who drank the laced wine subsequently died. Therefore, cyanide is deadly.
Does the new information close the gap in logic? Not really, though it does make the gap less obvious. After all, if the people who drank the cyanide all died, the cyanide must be deadly, right? And the comparison between those who drank the cyanide and those who did not seems compelling as well, at least until we realize that we do not know what happened to those who drank the unadulterated wine. The author clearly wants us to believe that they survived, but for all we know they died, too. So while all the new information seems to support the conclusion, what proof is there that the cyanide was the actual cause of the deaths? The flaw can be further disguised:
History has shown that cyanide can be classed as a deadly toxin. In one famous episode, for example, Lucrezia Borgia, notorious poisoner of sixteenth-century Italy, once served cyanide-laced wine to several of her political rivals at dinner, while serving unadulterated wine to her cronies. Within seconds, those who had ingested the cyanide were all dead in their seats.
Here, a narrative leads you away from the thread of logic. There is still no way to know that cyanide was actually responsible for the deaths. We can conceive of other explanations for their deaths, even if cyanide poisoning seems the most likely. However, the presentation of information makes this flaw recede into the background while the narrative comes to the fore. Putting the conclusion at the beginning of the argument also helps mislead the reader. But if we strip away the narrative, we are left with the original argument:
Some people have died after ingesting cyanide.
Therefore, cyanide is deadly.
As you read CR texts, you must strip away all the distracting layers to get to the core of the argument: conclusion and necessary evidence. Everything else is there to mislead you.
Copyright 2005 , MG Prep, Inc.; May not be reprinted without permission.
First, ETS engages in a process called "normalization," wherein all freshly written questions are tested by actual test takers to determine what percentage answer the questions correctly. If too many people answer correctly, the question may need to be toughened up. If too few people answer correctly, the question may need to be dumbed down. ETS is looking to assemble a pool of questions that covers a range of difficulty, from cakewalk to mind-bending. And the test takers help them do so.
How does ETS find these test takers? Easy. Everyone who takes the GMAT will end up answering approximately 10 unscored "experimental" math questions and 10 unscored "experimental" verbal questions. These questions are interspersed with the actual, scored questions with no way to identify them as experimental. So you, the test taker, do ETS's dirty work.
Second, the writers at ETS have a general sense of what makes a "500-level" question, a 600-level, 700-level, etc. (By the way, when we talk about a "500-level" question, for example, we mean that if that question represents the hardest level you reach on the exam, you will score in the 500 range.) Since each test is designed to evaluate proficiency in the same range of topics, the writers have to come up with ways to test the same concepts at different levels of difficulty. For example, let's say the writers want to test your understanding of percentages. Consider the following progression:
What is 12.5% of 48?
If 6 is x% of 48, what is the value of x?
If x/48 = 6.25/50 , what is the value of x?
There are 3 marbles in one bag, 2 blue and 1 red. There are 16 marbles in another bag, 1 yellow, 3 blue, 5 green, and 7 red. If one marble is selected from each bag at random, what is the probability that both marbles are blue?
There are 48 men in a certain rowing club. 5/12 of the men are handsome but not rich. 11/24 of the men are rich. What percentage of the men are neither rich nor handsome?
Notice that the answer to all these questions is basically the same, but the amount of thought needed to solve them varies considerably. All ETS does is add layers of information to the question to hide the core concept being tested. As you progress in difficulty, ETS is less interested in whether you can perform basic calculations and more interested in whether you can peel away the layers to get to the core.
As you prepare for the GMAT, you need to keep this in mind. Every difficult question is, at heart, a puffed-up version of a simpler question. Practice dissecting math questions to see what the underlying concept is, then retrace your steps and see how ETS layered the information to make the question more difficult. If you can master this skill, or become even moderately proficient, your performance should improve accordingly.
Layering in Data Sufficiency
Consider the following question, for example:
What is the value of x?
We have no way of knowing the value of x because we have been given no information about it. In Data Sufficiency problems you are given 2 pieces of information (called "statements") and asked to determine whether the statements (either individually or together) provide enough information to answer the question. How do GMAT test-writers come up with difficult Data Sufficiency problems? They take an easy piece of information and obscure it by adding layers of complexity.
In order to answer our question (What is the value of x?), the test-writers could provide you with a very straightforward statement. For example:
X = 2
This would be absurdly easy, so the test writers have to somehow tell you that x = 2 without stating it outright. What if we had the following statement:
A little harder, but not much. Let's try:
This statement can be factored into (x - 2)(x - 2) = 0, which tells us that the value of x must be 2. This is a little tougher to decipher, but still not at an especially high level of GMAT difficulty. What if we were given the following statement:
, where x is prime and y is even.If y is even, then must be even as well. Since
, it must be true that is also even. If is even, x itself must be even. Since x is both even and prime, it must be true that x = 2, because 2 is the only even prime. Compare the statement
, where x is prime and y is even, to the statement x = 2. The statements provide the same information, but one is unquestionably more difficult than the other. In Data Sufficiency, the level of difficulty is not wholly dependent on the difficulty of the concept; it depends in part on the skill with which the test writer conceals the necessary information. As you study, you should note any questions where the information was cleverly hidden and work backwards through the levels to see how the writers were able to mislead you. Many of their tricks appear over and over in questions in the Official Guide. If you learn to spot them, you will have an enormous advantage over other test takers.
Layering in Sentence Correction:
So far we have looked at the way ETS layers information in math questions. In this section, we will look at the layering effect in Sentence Correction. The same basic principles apply: a simple problem is made increasingly complex by adding information to obscure the core issues. In Sentence Correction questions, you are given a sentence in which a portion has been underlined. Your task is to determine whether the underlined portion is correct as it stands or whether it needs to be replaced with one of the answer choices in order to make the sentence grammatical and clear.For example, let's consider the following sentence:
The dog are friendly.
It does not take much effort to see that this sentence is flawed: the noun ("dog") is singular but the verb ("are") is plural. This would be much too easy for the GMAT, so the test writers must camouflage the error. One simple way to do so is to insert a lot of unnecessary verbiage between the noun and verb. We call this verbiage the "middleman". For example:
The dog, which was one of two puppies rescued from the shelter, are friendly.
The subject-verb flaw is a little harder to see now, but still readily apparent on a first read. If we take out the "middleman" (the intervening clause), we are back to the original sentence ("The dog are friendly"). Notice, however, that the writers have inserted a plural noun ("puppies") in the new clause so that you have plurality on the brain when you read "are friendly." If you are already thinking in plural terms, you are much less likely to spot the error. Even on a visual level, the subject of the sentence ("dog") is so far removed from the verb ("are") that the eye quickly alights on "puppies" as a possible subject for the plural "are". As tricky as this may already seem, the writers can put yet another kink in the rope:
Two puppies were rescued from the shelter, but neither of them are friendly.
The error in this sentence is significantly less apparent than those in the previous examples, though it is of the same basic type: subject-verb disagreement. Here the subject is "neither of them", which is singular (think of it as "neither one of them"). The verb, however, is still plural ("are"). The saga of the mismatched subject and verb goes on. Can the writers make the problem even harder to spot? Sure. Let's take a look at the following example:
Neither of the two puppies that were rescued from the shelter are friendly.
If you compare this sentence with the previous examples, the error seems almost completely camouflaged. We can see that the subject is "neither (one) of the two puppies", which is singular. But the verb "are" is still plural. As was the case with the math questions we looked at earlier in the month, the difficulty is layered. We have gone from "The dog are friendly" to "Neither of the two puppies that were rescued from the shelter are friendly" in a few steps, obscuring the central subject-verb issue along the way. Breaking sentences down into their component parts and analyzing their relationships is the key to success in Sentence Correction.
Layering in Critical Reasoning
In CR, you are given an argument and asked to evaluate its logic. This requires an ability to understand how the author supports his main point (the conclusion). Does the author use persuasive evidence? Are there gaps in the logic? The more information you have to sort through, the more difficult it is to determine whether the argument hangs together.But each argument must start with a conclusion and at least one piece of evidence that the author thinks will justify the conclusion. Let's take a basic example:
Some people have died after ingesting cyanide.
Therefore, cyanide is deadly.
Is this a persuasive argument? Not really. The fact that some people have died after ingesting cyanide is not, by itself, enough to prove that the cyanide was the cause of their deaths. This gap in logic is readily apparent, so the test writers need to add information to divert your attention away from the faulty logic. For example,
At a certain party, some people drank wine laced with cyanide and some drank unadulterated wine. Those who drank the laced wine subsequently died. Therefore, cyanide is deadly.
Does the new information close the gap in logic? Not really, though it does make the gap less obvious. After all, if the people who drank the cyanide all died, the cyanide must be deadly, right? And the comparison between those who drank the cyanide and those who did not seems compelling as well, at least until we realize that we do not know what happened to those who drank the unadulterated wine. The author clearly wants us to believe that they survived, but for all we know they died, too. So while all the new information seems to support the conclusion, what proof is there that the cyanide was the actual cause of the deaths? The flaw can be further disguised:
History has shown that cyanide can be classed as a deadly toxin. In one famous episode, for example, Lucrezia Borgia, notorious poisoner of sixteenth-century Italy, once served cyanide-laced wine to several of her political rivals at dinner, while serving unadulterated wine to her cronies. Within seconds, those who had ingested the cyanide were all dead in their seats.
Here, a narrative leads you away from the thread of logic. There is still no way to know that cyanide was actually responsible for the deaths. We can conceive of other explanations for their deaths, even if cyanide poisoning seems the most likely. However, the presentation of information makes this flaw recede into the background while the narrative comes to the fore. Putting the conclusion at the beginning of the argument also helps mislead the reader. But if we strip away the narrative, we are left with the original argument:
Some people have died after ingesting cyanide.
Therefore, cyanide is deadly.
As you read CR texts, you must strip away all the distracting layers to get to the core of the argument: conclusion and necessary evidence. Everything else is there to mislead you.
Copyright 2005 , MG Prep, Inc.; May not be reprinted without permission.
