To guess or not to guess: That is the question: that gets asked often

Bottom line: Guessing on the SAT shouldn't hurt or help you. Even completely random guessing. Eliminating even one answer choice, and then randomly guessing should increase your score.

If you randomly guessed on every question on the SAT, and your equally asinine friend left every single question blank, you two should get the exact same score. So there is no reason not to guess and since I subscribe to the philosophy that "you miss 100% of the shots you don't take," I always guess.

Now if you eliminate even one answer choice for every question - and there usually is at least one egregiously incorrect choice - and then randomly guess, your score should go up. If you care about the statistics of the matter, read on below. I have incorporated all of this into the "guessing game" which almost always successfully follows the stats.

Let’s play the guessing game.
This is a game I like to show my classes at the very beginning to teach them how the SAT is scored as well as answer the question, “Should I guess?” 

Write the numbers one through 40 on a sheet of paper.  Then randomly guess for each questions.  Have a test prep book ready to grade your answers.  Statistically speaking, there should be eight of each answer choice in 40 questions (8 As, 8 Bs, 8 Cs...).  This means that if you put all As for the 40 questions, you should get 8 questions correct and 32 incorrect, which will yield a score of 0.

This essentially means random guessing neither hurts nor helps you.  Of course, your random guessing could be lucky or unlucky and so use your discretion when actually deciding to guess.  The statistical answer to “should I guess” is that it doesn’t matter.

But then switch up the game a little bit.  Have a parent or friend “eliminate” one incorrect answer choice by using your test prep book and writing an answer choice which you shouldn’t guess (an answer choice which is incorrect).  For example, if your test prep book says the correct answer for question 1 is A, make sure the parent or friend eliminates an answer choice of B, C, D, or E.  Redo the process from above and randomly guess but now you only have four answer choices for each question.  Now go back and grade both sets of random guessing.  You should see that eliminating one answer choice presents a positive score. 

Again, statistically speaking, if you have 40 questions and now only 4 answer choices for each one (remember you got rid of one), you should get 10 correct and 30 incorrect.  This will yield a positive score.  And as always, your own guessing will vary slightly from this perfect scenario.

This means that elimination of one answer choice and then random guessing should be beneficial to you.  Eliminating one answer choice on each question should be fairly easy and the goal is to eliminate as many as possible to improve your score even more.

So now you know the philosophy of guessing. See how the CollegeBoard tells you to improve.