When to NOT answer a PGRE question.

[teroenza]

Hello. I have a quick procedural question about the physics GRE. I know that incorrect answers deduct 1/4 point from your raw score, but I do not understand how questions that are left unanswered affect you.

Basically I am trying to understand if I have absolutely no idea how to answer a question, whether or not I should guess, or leave it blank.

Thank you

 

[TMFKAN64]

Statistically, if you know *nothing* about a question, guessing will cost you as many points as you gain. (Guess 5 times, hit once, 1 - 4 * 1/4 = 0.) However, if you can eliminate an answer or two, guessing will help you. (If you can eliminate one answer, you'll hit 1 in 4 by guessing, and so 1 - 3 * 1/4 = 1/4.)

If you know absolutely nothing, Dirty Harry has the answer... "Do you feel lucky?"

 

[mfb]

For the expectation value: If you can exclude some value, or at least think that one answer is more likely than another: guess.
Guessing will increase the spread of your expected result - that can be positive or negative, depending on the result you want, and the number of correct answers you find.

 

[teroenza]

Thank you both. I think I am confused about the penalty for leaving something blank. I'm used to the standard test where any attempt at an answer is better than a blank, so I don't understand why leaving a question blank would ever be advantageous on the PGRE.

 

[TMFKAN64]

There is no penalty for leaving something blank.

The advantage is... a blank answer isn't wrong.

Statistics says that if you guess at 5 questions with 5 possible answers, you should expect one right on average. However, if you are just guessing randomly, there is a 32.768% chance you'll get them all wrong, and lose 1 1/4 points.

So... do you feel lucky... punk?

 

[teroenza]

This is a stupid question, but I have to ask it. What stops me from answering only one question, that I am absolutely confidant in getting right? Somehow this can't be allowed to happen, so would this process be penalized when the scaled score is computed?

 

[micromass]

This is a stupid question, but I have to ask it. What stops me from answering only one question, that I am absolutely confidant in getting right? Somehow this can't be allowed to happen, so would this process be penalized when the scaled score is computed?

You would get a 1% score. That doesn't seem very good...

 

[HallsofIvy]

This is a stupid question, but I have to ask it. What stops me from answering only one question, that I am absolutely confidant in getting right? Somehow this can't be allowed to happen, so would this process be penalized when the scaled score is computed?

No one is saying you are graded only on the questions you answer. What they are saying is that you get, say, 1 point if you answer correctly, 0 is you do not answer, and -1/4 point if you answer incorrectly.

If there are 100 question, you answer only one correctly, you get get 1 point, the "1%" micromass gives. If you answer that one question correctly, answer the other 99 completely at random, you would expect to get 99/5= 19.4, which I will round down to 19, correct and 80 incorrect Since you got a total of 20 right and 80 wrong, you will get 20-(1/4)(80)= 0 points.

 

[teroenza]

Thank you very much HallsofIvy.