True Exam Stories

What I want to know is, if you can see the answer well enough to copy it, can't you see the question well enough to notice that it isn't the same? But I guess there are a lot of stories about how dumb cheaters can be.

I don't have any stories about how to mess with cheating students. At the beginning of this semester the instructor in a math class I'm taking told a story about how another professor *failed* to prevent cheating using calculators, as a sort of explanation for why he wasn't going to try to prevent students from using any calculator and doing anything they wanted with it. The moral of the story for him was that students would find a way to do it anyway, so why bother.

I don't know of any especially good prevention techniques, but the attempt my Computer Science professor made during last night's midterm was pretty lame. His try at deterrence was telling us that if we cheated we would be caught and thrown out of school, thus we'd have to tell everyone we got kicked out of CS 110. That may be embarrassing situation but isn't much of an actual assurance that students won't try, anyway.

I'd expect a student from the "color-coded" class to sue the professor; announcing the existence of four versions and subsequently indicating four versions by color-coding in a way that is inconsistent with student expectations arising from the prior announcement clearly amounts to unfair examination conditions.

We run all variations on that here routinely:
different colour but same test; same colour different test; even different letter coding and colour but same different/test - so blue sheet version A may be same test as green sheet version E, but two blue sheets may have different version A and C.
One faculty made version variations up to H - eight different scrambled tests.
It helps to have automatic scramblers (some great TeX code for doing that) and you of course scramble both the questions and the answers between versions.

It slows down the cheaters, but they are winning the technology race

If there are 20 students and the test has 20 questions with 4 answers each, would it not be enough to pseudorandomly generate 20 different orders of questions, and to pseudorandomly order the answers for each question?

Knowing that his neighbor circled 'A' for question 11 is no help to the cheater unless he knows that maps to 'B' on his question 17.

For computer based exams there are programs available the do exactly what you (Three-Fifty) describe. However, such randomizing can ruin certain types of answer .... like "none of the above".

What I want to know is, if you can see the answer well enough to copy it, can't you see the question well enough to notice that it isn't the same? But I guess there are a lot of stories about how dumb cheaters can be.

I think this was supposed to be a ScanTron test-- one of those bubble sheet deals. So being able to see the answer sheet wouldn't necessarily let you distinguish the questions.

But you'd be amazed at how dumb cheaters can be.

However, such randomizing can ruin certain types of answer .... like "none of the above".

I usually use "There is not enough information to answer" instead of "None of the above." Even better is "There is not enough information to answer" with a blank to provide a short explanation of what additional information you would need to be able to answer.

For the large intro astronomy classes here with multiple choice exams, we randomize both the order the questions come in and the order of the answers for each question, so that each student's exam is very very unique. We still have "none of the above" type questions; they're jsut phrased as "none of the other answers are correct."

I used to teach high school science, and I did an A and B version, where the question order was scrambled, and the ABCD answers were scrambled. I didn't always make a big deal about the differing versions.

I had a handful of students that I am fairly sure were trying to cheat, because they had the right answers to most of the questions of the other test, but the result was that they failed the test, and I didn't have to have to explain to an angry parent how I "knew" that their little angel cheated. (Except for the kid who wrote review answers on his hand...the *outside* of his hand, that one was easy)

I like the different color paper bit, but if anyone ever figures out that the tests are the same, it looses all its deterrent value.

You couldn't do this on a "Scantron®" type test, but one of my instructors over the last couple of years [he taught both virology and microbial genetics] uses a variation of multiple choice that he calls "qualified multiple choice", where you not only must indicate the correct answer, but ALSO give a brief explanation of WHY the other answers are WRONG.
It's one thing to copy a "A" or "B" or "C" or whatever mark, but having to also copy the rest of the answer would be something of a hassle, I think.

Here is a joke.

A teacher gives his class a 1 hour long true-false test.
"Ok in 10 seconds you can start .... 4 3 2 1 GO"
The students turn the quiz over grab their pencils and go at it.
After a while the teacher notices a student in the back of the room repeatedly flipping a coin, catching and looking at it and then making marks on his paper.
For a moment the teacher is about to say something but then decides to let the student have a shot at the tiny chance of passing.
After the hour has passed
"OK class time is up. Leave you quiz on your desks and see you tommorrow."
As the cloud of students begins to thin he notices the same student in the back of the room flipping his coin.
"Hey, I said time is up! Please stop."
"Excuse me sir. I am checking my answers!"

The students get up and start

I've used test generation software to randomize order, answers, and distractors on a per-student basis for classes as large as thirty. Mind you, this was for a mid-term/final for a class of inveterate talkers. A couple of kids just dropped the test on the floor and put their heads on their desks, but, what the hey? I graded them all.

During my brief stint as a high school science teacher I scrambled the order of the problems and used different numbers for a physics final. The classes were the worst about talking during tests and this way they at least had to punch the numbers into their own calculator.

The funniest cheating incident I ran into was in Earth Science (7th graders). There were two teacher's editions in the room. I had to step out of the room and the boys grabbed the extra one. They copied the answers including one that started "Student answers will vary," and then stopped halfway through the next sentence before they actually answered the question. I ROFL when I got those papers. Luckily for them it was a review exercise.

Those tricks with color are as old as the hills. As one writer noted, the high end is to have N different exams for N students. The student enters the code number from the copy they get, and the computer matches the key to the code.

My favorite technique is to include a couple of questions that are based on a graph of a function. An example might be a graph of v(t) with a question asking for the acceleration at a specific time or one asking for the displacement between two times. There are two different graphs, and two questions. I scramble the graph but not the question, so a student who is tempted to copy and assumes that the graph on page 3 of his exam means it is the same problem as the one on page 5 of the other exam is going to get it exactly wrong. There are lots of ways to do this on a physics exam. Cheaters seldom read the question, so they won't even notice if it is worth the same points. Heck, they have trouble copying a solution when it is just homework and not a proctored exam.

I also like proctoring from the back to see who looks nervous, and seating students where I want them to sit if I suspect collaboration among a pair. Busted one kid big time when he had to sit up front where he could not see any other papers.

I don't buy the excuse about calculators, particularly in a math class. Some of our math faculty give a two part exam, one with symbolic problems and no calculator that you complete before you get part where calculator use is allowed and it is assumed the student is fluent with the calculator and has it loaded with whatever it can carry. Other faculty are suckers who don't care if they pass ignorance on to the rest of us. Is it so hard to have a problem where there are symbols given rather than numbers?

My personal preference is to restrict them to the kind of mid-range calculator (the Casio fx-115ms is my favorite) that cannot store "flash cards" like you find in a TI-83+. The alternative is to allow them the use of a computer like a 92 and give zero partial credit and only enough time to solve the exam under the assumption they can use the matrix solver, etc, with a high degree of fluency. They seldom want that kind of test.