<p>I always hear people saying that you should leave the question blank if you don't know the answer. However, this doesn't make much sense to me: you have more to gain than you have to lose.</p>
<p>For correct answers, you gain 1 point. For incorrect answers, you lose 1/4 point. For blank answers, your score stays the same.</p>
<p>Let's assume you have 50 correct answers and have 10 questions you aren't sure about.</p>
<p>If you leave all 10 blank, your score stays the same: 50.</p>
<p>Now, let's imagine you guessed on all of them:
You have a 1/5 probability of getting it right. 10<em>(1/5)= 2
You also have a 4/5 probably of getting it wrong, for which you lose 1/4 point. So 10</em>(4/5)*(-1/4)=-2.</p>
<p>So, you would actually break even. Realistically, you could probably get most questions down to 3 potential answers. With this, your odds get even better.</p>
<p>10<em>(1/3)=3 1/3
10</em>(2/3)*(-1/4)=-1 2/3</p>
<p>So you'd net a total of 1 and 2/3 more points than if you had left the questions blank.</p>
<p>Is there something wrong with this math? What am I missing?</p>
<p>You are right. The score deduction is constructed such that it just (on average) eliminates all the potential advantages from blind guesswork while not penalizing you for blindly guessing either - and will therefore reward you (again, on average) if you can eliminate one or more options from the list before picking one of the remaining answers.</p>
<p>There's been multiple threads about this. :]
When you are in the 700's the penalty point can take a chunk away (usually 20-30)
So some people play it safe :]</p>
<p>anhtimmy: Haha, guess it's one of those reoccurring topics... I probably should have searched first though.</p>
<p>I'm still not sure that leaving anything blank is really playing it safe. You have more to gain than you have to lose. But I suppose if you only have 1 or 2 questions you're unsure about then the chance of error increases. (The fewer questions, the less likely the empirical data will match the theoretical analysis.)</p>
<p>Generally, if you can eliminate and narrow down the choices, then it is to your advantage probabilistically to answer one of the remaining choices. Most prep book agrees that you should go ahead and answer in this situation. However, if you cannot eliminate any choices like when time runs out at the end of the test and you don't have time to read the remaining questions, it is probably better to leave them blank.</p>
<p>Let's say you have 2 questions left at the end of the section when time runs out. You have 1/25 chance of getting 2 points by guessing and 16/25 chance of loosing 0.5 points. It all evens out with large sample of questions but that 0.5 points off can be 20-30 points on the scaled score like previous poster said.</p>
<p>ttparent: Yea, I guess it's all about the scaled curve. If you're in the very upper reaches, you're probably better off with blanks. But otherwise, guesses seem to be worth their value.</p>
<p>Again, guesses are worth something only when you can eliminate, otherwise, zero sum game.</p>