Is it dishonest to guess on multiple choice exams?

Question

I have never seen a thing, until recently, like someone not marking a cross on a multiple choice test just because he didn't know the answer.

I wonder whether this is a East Asian right-thing-to-do mentality, or just a quirk of some student.

Should maybe we ask all students to do this? Indeed, on a multiple choice with 5 choices per question, the expectancy of a student employing pure guessing would be a 20% score. But is it cheating to mark something when you don't know the answer? At least, you are getting sometimes some point for nothing.

Answer 1

Should maybe we ask all students to do this [refrain from guessing on multiple choice exams]?

Some multiple-choice exams use negative scoring, where points are deducted for wrong answers. This discourages guessing, but introduces other problems.

One problem is that students who know the material vary a great deal in their confidence in that knowledge. If students are actively discouraged from answering questions unless they are sure they know the answer, students who lack confidence in their knowledge will be at a disadvantage. (I'm assuming you want the exam to measure student's knowledge, and not their confidence in that knowledge.)

But is it cheating to mark something when you don't know the answer?

Not unless you're told "Don't mark something if you don't know the answer." Cheating implies deceit - there is nothing dishonest about guessing, unless it's forbidden and you do it anyways.

Answer 2

My (european) students would probably be either very perplexed or laughing at me when I told them that they are not allowed to "guess" if they do not know the answer. This is not only impossible to enforce, but also much less well-defined than you seem to think (as ff524 already indicates). If I have a good idea what the answer is but I am not sure, am I allowed to answer? If I have in principle no clue, but from the way the question is phrased I can guess that the answer will be (c), am I allowed to answer?

More importantly, if there is such an obvious flaw in your testing system (guessing being generally +EV, positive expected value, in multiple choice tests with no point subtractions for wrong answers), it seems lazy to shift the burden of not exploiting this hole to the student versus designing your test in a smarter way (or living with the fact that even a random selector will get X% of all points on your test, which may be ok for you).

But is it cheating to mark something when you don't know the answer? At least, you are getting sometimes some point for nothing.

In any exam a student can get points for getting lucky. Assume a student has learned only 30% of a given chapter, and the question deals with the part that he learned. Is it cheating to answer in this case?

Moreover, and maybe more controversially, I think you'll need a rather wide definition of cheating so that guessing a random answer falls into it, even if you told them before not to do it. Sure, if the rules are that you cannot do it, and you do it anyway, you are breaking the rules. However, if you know there is a pretty big hole in your system and you do not take even easy steps to fix it, I feel you can hardly fault a student for exploiting the hole.

Answer 3

Is it dishonest to guess on multiple choice exams?

No. The next step of logic might be to not write an answer to anything unless it is not falsifiable, for fear of being proven wrong. Which defeats the purpose of science. In most every multiple choice question, there are some answers that are clearly more probable than others - most guesses will still be educated.

It's not dishonest to be wrong. It's only dishonest to know you're wrong but tell people that you're right.

Answer 4

In some sense, I do not understand the context. Many kids, myself included, have the dubious capacity to infer from the wording of the question and the answers what a reasonable answer would be, thus quite successfully gaming the system.

Indeed, a rational person would exclude implausible answers, and look at the plausible, and if those can easily be distinguished, we're done.

That is, a multiple-choice test cannot possibly compel the examinees to really "work the problems out".

(I've done some experiments in which by-me-designed multiple-choice quizzes on sophisticated material were better done by smart English major friends of the (smart-enough, for sure) math grad students, due to reading nuances of questions and answers.

That convinced me ... not so much to not do multiple-choice, but that the constraints of the multiple-choice "pipe" are too narrow, and do not address what we want. The same is surely true at more elementary levels.

(The pseudo-economy of machine-grading and so on is somewhat of a false economy if one wants to avoid rewarding clever-gaming-of-system... duh.)

Answer 5

My innovative solution: allow students to "bet" on their answers.

1. Add two options to each question. "I am very sure this is the right answer." and "I am very unsure this is the right answer."
2. Students can mark either value or none at all for any question.
3. Double the value of all questions marked extremely sure.
4. Halve the value of all questions marked unsure.

And then curve accordingly.

Pros

• Provides a lot of great feedback about what your students are learning and not learning.
• Allows students to self-assess for follow up work.
• Reduces the value of guessing.

Since the ultimate goal of all this testing is to see what your students are learning, and guessing muddies the water, so to speak, you might as well let the students tell you when they're guessing.

Answer 6

On my university on some exams it is solved by giving negative points for the wrong answer, but only after giving some number of wrong answers.

For example, every wrong answer is -0.5 points, but you won't get negative points before you give 3 wrong answers, after third wrong answer, you will get -1.5 points, and for every next wrong answer additional -0.5.

This way, students are discouraged from total random answers, but also students who aren't too confident in their knowledge have some "space" for wrong answers, so they don't have to fear negative points so much.

Answer 7

Here's a tough one: German driving license theory test http://fahrschule.freenet.de

The rules: It's multiple choice. You are not given the number of correct answers (there can be multiple correct ones, or no correct ones, or one correct answer). You get points for each incorrect choice. Checking one wrong answer instead of the correct one is two incorrect answers. Checking only one of two correct answers is one incorrect choice. So there is no guessing "what's the best answer". And you are allowed two or three incorrect answers out of forty questions. They don't bother counting correct answers because you must have almost all correct.

Basically, if you haven't learned your questions and know at least 95% of the answers, there's no chance. On the positive side, you can officially buy all the possible questions and answers and practice as much as you like.

Answer 8

In the university I've studied, the expected value of choosing randomly is almost ever 0, by decreasing your result if you make a wrong choice. If for some reason there's no penalty for missing, I'd assume it is intended to give you value for "guesses" or the test itself is not that important. I don't know what should be the emotional approach, but I'm pretty sure the logical approach is that you should mark every question if the expected value is above 0, unless you know you already have enough points and don't want to risk overall failure.

Answer 9

I suppose on some level if you were purely guessing at an answer, then it would be a form of cheating in that you would get credit for something you did not legitimately know. However, if you are able to work out an educated guess, then you certainly deserve credit because it demonstrates an ability to deduce an answer that you did not know upon reading the question. That demonstrates critical thinking and logic skills above simply knowing the right formula off hand.

Answer 10

Indeed, on a multiple choice with 5 choices per question, it is a given that any student, no matter whether he is as thick as brick, would at least get a 20% grade

Not, he will get not, except when very lucky.

You can get 20% points on 5-choices per question test by guessing (on average) on single-choice test (one answer correct).

Assuming the correct answer on multiple choice is only then, when you have selected all and only correct answers, guessing is an extremally ineffective strategy on such tests. Multiple choice tests are already designed to prevent guessing!

And even 20% is much below typical exam passing range. In my country it's typical to set the threshold to 50%. Everything below that means a failed exam.

If your exam criteria allow passing on 20% of correct answers, you have a problem in completely other place.

Answer 11

I had a bit of a different situation during my Uni education than any mentioned here... I'm not sure if it is the best solution to evaluating the students, but it is definitely implemented and has been used for several generations (in Croatia, Computer Science).

The utterly strange this is, we had multiple choice exams for mathematical problems and in general, other types of subjects that require you to solve exercises, tasks or problems and get a number as an answer.

The main points of how it worked:

• for each question, you had N different answers (where N would be the same for the whole exam). (N ~ 4 or 5, typically)

• the answers offered were usually:

  • 1 correct answer
  • 1 correct answer * 10^x (e.g. if the right answer was 23.5 you might have 2.35 offered)

  • 2 answers you could get by making a low-level calculus mistake in the typical solution-process (1 direct and 1 with a *10^x shift, as with the correct answer)

    (e.g. if there's a standardized procedure for solving the problem, it is the answer you would get if you flip a sign somewhere by accident)

  • 1 completely wrong answer
• the marks were distributed as:
  • +1 for the correct answer
  • 0 for unanswered
  • a bit more than -1/N for wrong answer (e.g. if N = 5, then the wrong answer might be -0.25 instead of -0.20)

• additionally, for the correct answer to be accepted, you had to enclose a paper with your full solution to that question (where you did your calculations).

  They weren't checking this very diligently for every answer, but you were always aware that they could.

---

This tried to encourage the educated guesses (as you have your calculations, and maybe you're unsure between two answers because you're unsure if you used the right procedure). On the other hand, the penalty also discouraged complete guesses, as the potential negative marks were higher than the potential gain.

Of course, such a system still has plenty shortcomings. A typical problem situation was the student would having the right solution in his calculation, but transferred the wrong answer to the multiple choice answer sheet. The policies varied with different courses, although commonly, such an answer was not accepted, to prevent students for purposefully offering two answers and having a better chance of scoring.

---

I just realized I haven't actually offered my direct answer to the question. Here it goes:

In the system I describe, I would say it is unethical to try and guess in multiple choice exams. But, unlike all the other multiple choice setups, the one described here has a way to verify weather you guessed or not (if you have not submitted the full procedure for your answer, you must have guessed) and because of that, it stops depending solely on the students "honor"