Do you prefer multiple choice or free response exams?

[ftg]

Former Computer Science prof here.

It all … depends. (Helpful, I know.)

As to mutliple choice: I was a whiz at those. Give me a SAT/GRE/PSAT whatever and I breezed thru it easily.

Usually. But some exam folk insist on basically a lot of trick questions. I did the Sun Java Cert. exam and it was full on stuff specifically designed to trick you for no good reason. A horribly designed exam. (Goes with the language, I guess.)

Written tests were more of a challenge but usually not all that bad.

Now, oral exams. Here it depends on the personality of the person asking. Some try to be a bit helpful, some go full on out of their way to mess with you. If I didn’t have too much of the latter, they were pretty nice.

On the other end:

I only did full multiple choice exams for Mickey Mouse courses for non-majors. I didn’t care, they didn’t care, let’s turn it over to the Scantron.

For exams in the majors there were standard written exam questions. For intro courses to a sub-area I’d start of the exam with one question made up of a bunch of easy-ish T/F questions. Put the folk at ease a bit and get them thinking positive about the test. (But still … Once I had a student that went 0 for 10 on the questions for a midterm. Knew they were in trouble and dropped the course without getting the test back.)

But once into the 2nd+ course in an area, no such “gimmes” on a test.

For oral exams, I remembered the style of questions I liked getting when I was a student and went along those lines. No trickery, “gotcha” or anything.

One thing that always baffled me. Eventually I went so far as to do this:
On the first day of class I would tell them they were going to get a question about X on the midterm.
When I got to the topic, I reminded them of the question and here was how to figure out the answer.
The class before the midterm I would remind them about question X.
After the midterm, I’d cover the answers. Point out how almost everyone got it wrong. Tell them it was going to appear again on the final.
Before the final, I’d remind them of question X.
Again, very few got the question right.

I … just … don’t … get … it.

(I told this to other profs and they repeated the experiment in their classes. Same result.)

[GreenElf]

I’m not glib orally or on paper, and I’m an anally slow writer (with nice cursive handwriting :)), so I prefer fill-in-the-dot tests.

[Chronos]

Speaking as a student, it depends on why I’m taking the test. If it’s just a test to get some silly meaningless requirement out of the way (and yes, such tests do exist), then I prefer multiple choice, because it’s quicker and easier for everyone else. If, however, it’s a test that seeks to measure how well I actually understand something, then open response, because that’s very difficult to measure well with multiple choice.

Speaking as a teacher, I will never knowingly give a test just to get a silly meaningless requirement out of the way, and so I will always give open response tests.

Quoth Hari Seldon:

Free response questions are very hard to mark because when the student gets the wrong answer, you have to decide if they just made a silly error and really understood or if they really didn’t understand the material and that meant following the work closely.

Yes, that’s hard to do for an open-response question. But it’s impossible for a multiple-choice question. Hard is better than impossible.

[Volxen]

@Chronos

Speaking as a student, it depends on why I’m taking the test. If it’s just a test to get some silly meaningless requirement out of the way (and yes, such tests do exist), then I prefer multiple choice, because it’s quicker and easier for everyone else. If, however, it’s a test that seeks to measure how well I actually understand something, then open response, because that’s very difficult to measure well with multiple choice.

Speaking as a teacher, I will never knowingly give a test just to get a silly meaningless requirement out of the way, and so I will always give open response tests.
Yes, that’s hard to do for an open-response question. But it’s impossible for a multiple-choice question. Hard is better than impossible.

I think that really depends on how the test is structured. I teach computer science at the community college level, and I use multiple choice tests, but my MC tests are very different from “traditional” MC tests that simply test surface-level information (e.g. whats the definition of X, which of the following sentences best describes X, etc.).

For example, many of my MC questions will provide some code, and the students have to correctly determine the output of the code. Or as another example, each of the options will contain code, and the question will be phrased, “Which of the options below will NOT do x”, meaning that all of the options except for one will do x, and the student’s have to determine which option is wrong (which of the code snippets will not do x). It’s not something that can simply be memorized, as the only way they can arrive at the correct answer to is to actually study the code and work out the solution. And I choose the distractors (wrong answers) in such a way that if they make common mistakes, it will naturally lead them to one of the wrong answers. And starting this semester I’m also going to be including a “none of the above” option for all of the questions, which will make the questions more difficult, but also more accurate in terms of representing what the students really know. I may even implement a penalty for choosing the wrong answer to discourage guessing.

All that being said, my tests are not fully MC – there are always one or two programming problems the students have to do from scratch. Though I have been told, even by some of my better students, that my MC questions are more difficult than my programming problems.

[Chronos]

And those are good questions, as multiple choice goes. But they’re still more limited than open response questions. For instance, suppose that a student makes a mistake that’s not one of the common ones: They’re not going to get any of your answer choices. This might lead them to say “none of the above”, if that’s an option, but in that case you’ve lost all information as to what mistake the student made. Or you can expand your number of answer choices to encompass even more of the possible mistakes, but that tends to lead to the answer choices clustering, and for the correct answer to be the “most typical” of the answer choices, enabling students to answer even without reading the question.

[Thudlow_Boink]

@Chronos

And those are good questions, as multiple choice goes. But they’re still more limited than open response questions. For instance, suppose that a student makes a mistake that’s not one of the common ones: They’re not going to get any of your answer choices. This might lead them to say “none of the above”, if that’s an option, but in that case you’ve lost all information as to what mistake the student made. Or you can expand your number of answer choices to encompass even more of the possible mistakes, but that tends to lead to the answer choices clustering, and for the correct answer to be the “most typical” of the answer choices, enabling students to answer even without reading the question.

Yeah, if it’s the type of multiple choice question where the student is intended to work through a problem, get an answer, and then find that answer on a list of choices, it works better to just make them write that answer down—except that it’s perhaps more time consuming to grade that way.

[Volxen]

@Thudlow_Boink

Yeah, if it’s the type of multiple choice question where the student is intended to work through a problem, get an answer, and then find that answer on a list of choices, it works better to just make them write that answer down—except that it’s perhaps more time consuming to grade that way.

It’s definitely more time consuming to grade that way. Unless you don’t believe in partial credit, that is.

Honestly, I think asking the type of questions I mentioned, with five options to choose from – the correct answer, three really strong/plausible distractors, and a “none of the above” option – is just as reliable as a short answer or fill-in-the blank type of question. The key is that the distractors do need to be really strong/plausible, and the “none of the above” option should be the correct answer for approximately 20% - 25% of the questions. The NOTA option is also important, because it means if students make a mistake that leads to a result not covered by any of the options, then it’s not as simple as choosing the option that is closest to their incorrect result (which is what students will do if there is no NOTA option). And if all of that is not enough, you can deduct a certain amount of points per incorrect answer to discourage blind guessing.

[RadicalPi]

On most multiple tests, there are four answers: the correct answer and three wrong answers that are trying to hide that correct answer. If you figure out which is the one that the others are based on, then you know the answer. Most of the time, you don’t even have to know anything about the subject you’re being tested on. I know I don’t.

This sort of thing goes a long way toward explaining why I think multiple choice tests are pretty much worthless except as a measure of how good you are at taking multiple choice tests.

[BigT]

I say free response test actually are the ones that test something else–they test your ability to write. Which is fine if that’s the point, but it often isn’t. I also find that they tend to have things they want you to say, but you don’t know what they are.

It’s not that I do badly on them. I don’t. They do take a whole lot of time, though, and ultimately, in my opinion, contain a lot less information than they could have gotten from me with multiple choice in the same time period.

And you can still B.S. them.

[Mudshark]

I preferred taking multiple choice tests when I was student.

[Nava]

I dislike MC more than essays/problems; then again, I’m also one of those people who prefer oral exams to written ones.

Most people are very bad at writing questions which don’t glow a spotlight on the “right” answer.
Even when that doesn’t happen, test-taking techniques can get you a high grade on a subject in which you don’t know shit from shinola.
There’s a certain type of teacher (analysis of those I personally know indicates it tends to be people who don’t really comprehend the subject) who wants regurgitation rather than comprehension. Their MCs tend to suck even worse than their essays; just yesterday I had an MC where the “right” answer did not match the text we’d studied from and there was a “none of the above” option.
MCs make for easy correction - lazy correction. I don’t trust a teacher who isn’t willing to put at least some effort into grading.