Individual Question Value

[LegionofPhoenix]

Hi there. I have 2 questions actually.

1. Does getting a question (or x number of questions) wrong, subtract from your overall score? I believe I once read that for every 4 questions you get wrong, it subtracts a question you got correct.

Also, read that the questions depreciate in value, and that the initial questions are worth more than the questions later on. However, I've also read that each question is worth the same amount.

What I'm trying to figure out, is if I encounter a problem that either will take me a long time to solve, or a problem which I won't be able to figure out, do I skip it or guess? Because it seems like an educated guess might be worth it.

2. Can you retake the test once a month, or every 30 days? For instance, can you take it on the 30th, and then 2 days later on the 1st? Or do you have to wait 30 full days?

Thanks for the help in advance.

[Vacuum]

I believe the first 10-15 or so questions hold the most point value then all the questions following those tweak your score to a smaller degree. As just a rough example it would be something like, the first 10-15 questions determine what 'hundred' you are at and the following questions will determine what 'ten' you are (this is obviously not accurate since you can't get 1500 on one section, but just an example of the different level of points).

I'm not 100% sure how wrong answers work, other than it drops you down in the 'hardness' of questions. The easier the question, the less points it is worth. So essentially, if you get a lot wrong in a row, you will be getting a lower score since the question you do get right would be worth way less points.

I've heard people say that you should spend extra time on the first questions which are worth the higher point value and then less time on the later questions (if you run into trouble solving). I ran out of time on the quantitative all together, but I'd say I spent equal time on everything for the questions I actually did complete. I only took the GRE once, but I think if I were to take it again my pacing would improve. If you are taking the practice exams, make sure you work on pacing.

I remember asking the guy at the testing center about the dates you can retake, since I took my exam at the very end of the month and he said it was once every calendar month and then said I could basically take it the next week. So to me that is you could take it July 25 and take it again Aug 1st. I'd double check on that though because the guy at ETS that I asked didn't seem too confident in his answer!

[communications13]

The CURRENT (the test chances Aug 1st!) test on a computer is adaptive. That means it adjusts the level of difficulty as you continue the test and considers difficulty of the questions in the score. This means that if two people get the same number of questions right, their score will be different because the computer adapts to their skill levels. So working for a while on a harder question might be to your advantage because it can be worth more points, but it is probably best to not worry about it. There is no meter telling you what level difficulty the question is. Focus your energy on doing your best on each question, not the difficulty of the questions.

Also, unlike the SAT, the GRE does not penalize for wrong anwsers so guessing is to your advantage. If you can rule anything out, then take a guess.

The ets website specifies that you can take it once per calender month, which implies for example anytime in Jan and anytime in Feb, however, you can only take it 5 times in a 12-month period.

[Eigen]

I can't speak for the new versions, but the old ones worked like this:

they start you out at a 500 point (average) score, and give you a question. If you get it right, you go up 200 points, and get a 700 point question. If you get it wrong, you go down 200 points and get a 300 point question. The point swings slowly drop (200, to 150, to 100, to 50, etc), zeroing in on the "point range" where you can consistently answer questions.

For each question you get right, you move up in difficulty, for each question you get wrong you move down in difficulty. The highest level question you can consistently answer correctly determines your score (hence the adaptive part).

The earlier questions have a much larger impact on your score- if you get the first question right and the second one wrong, you would be sitting at around a 600, whereas if you get the first question wrong and the second right you'd be sitting at around a 400. You're better off spending the time on the earlier questions as opposed to the later questions. You can still end up with a high score getting the first question wrong, it just takes much longer strings of consecutive answers.

The point values/question numbers here are approximate- they don't release exactly how the algorithm works, but they give you a general idea. It's not a test like the ACT or SAT where there are sliding scales for questions answered right/wrong and scores- it relates to the difficulty of questions you answer correctly- each question has a score value attached as relates to its difficulty level.

It also lets you feel how your performance is going... If you're getting lots of hard questions (or the questions seem to be getting harder) that's a good thing! And usually, getting the easy questions right is more important than getting the really hard ones right.

Edited February 13, 2011 by Eigen

[Sarsathome]

Agreed with the previous posts. Although the test doesn't explicitly penalize you for answering wrong, the difficulty of the next question will drop.

I've been told the best strategy is to really focus on getting the first 5 to 10 correct, because they determine the majority of your score, say 400 vs. 700. After that it finds the ten point interval you fit in. So 750 vs 780.

It's always best to never leave a question unanswered. If you're running out of time and have many questions left, it's best to intelligently guess the most likely answer to make sure you answer all question.

[DrFaustus666]

Agreed with the previous posts. Although the test doesn't explicitly penalize you for answering wrong, the difficulty of the next question will drop.

I've been told the best strategy is to really focus on getting the first 5 to 10 correct, because they determine the majority of your score, say 400 vs. 700. After that it finds the ten point interval you fit in. So 750 vs 780.

It's always best to never leave a question unanswered. If you're running out of time and have many questions left, it's best to intelligently guess the most likely answer to make sure you answer all question.

I've done very extensive but informal research on this subject (the scoring algorithm of the computer adaptive test), using the two ETS-supplied Poweprep (OLD) tests, together with lists of the right and wrong answers. I've experimented, deliberately answering a question or two wrong, then answering every succeeding question right, just to see what happens.

The short answer to all of this is: to get an 800 (or high 700s) on both Verbal and Quantitative, respectively, you must

EITHER:

(1) answer all of the first 20 questions correctly, then make no more than two or three mistakes in questions 21-end (30 or 28, respectively)

OR

(2) if you make even one mistake in the first eight questions or so, you must answer all succeeding questions correctly to score above about 720.

Also,

(1) answering all questions randomly yeilds a score in the low 300s at best

(2) answering the first 15 or so questions correctly, plus totally random guesses for the remaining questions yeilds scores in the low 600 range

(3) if you make more than one mistake in the first five questions, it is next-to-impossible to score above 750

All of this is with the OLD POWERPREP program, and of course, ETS has probably improved their scoring algorithm, such that it's not possible any longer to take tons of time on the first 15-20 questions, then rapid-fire guess at the rest.

Hope this helps.

John

Edited February 14, 2011 by DrFaustus666

[DrFaustus666]

I've done very extensive but informal research on this subject (the scoring algorithm of the computer adaptive test), using the two ETS-supplied Poweprep (OLD) tests, together with lists of the right and wrong answers. I've experimented, deliberately answering a question or two wrong, then answering every succeeding question right, just to see what happens.

The short answer to all of this is: to get an 800 (or high 700s) on both Verbal and Quantitative, respectively, you must

EITHER:

(1) answer all of the first 20 questions correctly, then make no more than two or three mistakes in questions 21-end (30 or 28, respectively)

OR

(2) if you make even one mistake in the first eight questions or so, you must answer all succeeding questions correctly to score above about 720.

Also,

(1) answering all questions randomly yeilds a score in the low 300s at best

(2) answering the first 15 or so questions correctly, plus totally random guesses for the remaining questions yeilds scores in the low 600 range

(3) if you make more than one mistake in the first five questions, it is next-to-impossible to score above 750

All of this is with the OLD POWERPREP program, and of course, ETS has probably improved their scoring algorithm, such that it's not possible any longer to take tons of time on the first 15-20 questions, then rapid-fire guess at the rest.

Hope this helps.

John

[warpspeed]

you can take the est once every CALENDAR month. so yes, you could take it on the 30th and the 1st of the next month.

[LegionofPhoenix]

Thank You Everyone. You have been incredibly helpful. I really appreciate the detailed answers.

[Ennue]

Hi there. I have 2 questions actually.

1. Does getting a question (or x number of questions) wrong, subtract from your overall score? I believe I once read that for every 4 questions you get wrong, it subtracts a question you got correct.

Also, read that the questions depreciate in value, and that the initial questions are worth more than the questions later on. However, I've also read that each question is worth the same amount.

I understand your confusion, because both things are correct. For the GRE general test (with Quant and Verbal) the questions decrease in value and the penalty for giving a wrong answer are as the other posters decribed. But for the GRE subject tests, every question is equal in value, it is possible to skip questions (which is not possible for the GRE general as far as I know - not sure about that though) and for every four questions you get wrong, a right answer is substracted. So it depends on which test you're taking (which is probably the GRE general test)!

[fenderpete]

The short answer to all of this is: to get an 800 (or high 700s) on both Verbal and Quantitative, respectively, you must

EITHER:

(1) answer all of the first 20 questions correctly, then make no more than two or three mistakes in questions 21-end (30 or 28, respectively)

What happens if you get the first 20 right, then get the rest wrong? Just interested as I'm currently working out a pacing strategy for the GRE in April. I need to break 700-720 if possible, but would love to be in with a chance of getting higher if I can.

Thanks for a great and very informative post!

Edited March 6, 2011 by fenderpete

[Tommy Wallach]

Hey Everyone,

Wow, great discussion! Unfortunately, a lot of misinformation has been spread both in this forum and in the greater testing world, so I want to dispel some myths.

THE BEGINNING IS NOT MORE IMPORTANT! Okay...started there. Moving on:

First of all, computer adaptive tests do technically penalize you more for a blank answer than a wrong answer. However, because you can't leave questions blank mid test, this will only happen if you run out of time. So don't run out of time!

Secondly, the beginning of the test is not more important than the rest of it. This would not make the test very good, as it wouldn't make ANY logical sense to have the beginning actually be more important. What's really going on is far more complicated. In reality, the first few questions do MOVE the difficulty more violently, because the test is trying to work out who you are. Think of it this way: If you were an 800 level student, and you start at a 500-level question #1 (As all students do), and the test only went up by 20 points if you got a question right, it would take it 15 questions to start feeding you the 800 level questions you deserve. Instead, it swings more violently at the beginning to try to zero in on who you are. This DOES NOT make the test more important at the beginning. In truth, all the questions are equally weighted, and just as important for your score. For more info on how this works, feel free to read this article on item response theory.

Thirdly, the test does not SUBTRACT a right answer based on wrong answers. It doesn't work that way at all. Like I said, it's far more complicated.

Fourthly, while extensive messing about with Powerprep does make the beginning of the test look more important, Powerprep IS NOT A REAL TEST. It doesn't have nearly enough questions in its pool to simulate a real GRE, and the scoring algorithm is compromised as a result. Read up on what ETS has said about the testing algorithm, but don't pay attention to what others say, and don't trust experimentation on the Powerprep tests. As this forum makes clear, this is a subject that confuses a lot of people. Do not trust the idea that you have to get every question right after the first 10, or any mumbo-jumbo like that. It's more complicated than that. But to repeat, loudly:

THE BEGINNING OF THIS TEST IS NOT NOT NOT SIGNIFICANTLY MORE IMPORTANT THAN THE END!

Feel free to follow-up with more questions!

[cherubie]

THE BEGINNING OF THIS TEST IS NOT NOT NOT SIGNIFICANTLY MORE IMPORTANT THAN THE END!

Feel free to follow-up with more questions!

Yea...that pretty much sums it all up. I got one of the first 10 questions, then like 2 or 3 more randomly littered questions throughout the whole Quant section (one question I spent 7 minutes on and got it wrong lol) and still racked up an 800Q. So no, you don't need to get the first 20 correct, and if you get 1 of them wrong, it's not doomsday for the GRE. As far as easy/hard questions, my diagnostic shows those were all over the place. I jumped from a level 4 (or 5, whatever's the highest) to a level 2 then continue with a couple 2-3 then to 5 again. It probably varies on the person. But in GENERAL, I feel that if you start seeing statistics questions [not just probability, I mean stats 101 stuff], you're doing pretty well.