This reminds me of a particularly nasty case we had many decades ago. There was a cohort of students that handed in exams with essentially the same answers. An investigation noted a few things:
The records of the exam hall showed that they sat at desks in a V formation.
The writing in the exam books was unnaturally large.
Where there were corrections to the answers, all the papers had the same corrections.
Should have been an easy call. But there were complicating factors - which I won’t detail. The students protested their innocence. Same answers? “We all studied together.” Eventually it reached a very high level board, and the complicating factors got the students a free pass. The ring-leader (seated at the head of the V) had faced down vastly more dangerous people than a committee of dopey academics. His job was to get the others through and he was going to get his way.
This was a time when the university was, well, naive about the wicked world that was to come. It took ages to do things like enforce randomised allocated desks for taking exams, and require students to provide photo-ids that were matched to exam books. It does make me sad how teaching has lost its innocence.
Allowing easy cheating dilutes the brand, and does so in a manner that is very hard to recover form.
I was a B and C student in high school largely through my lack of effort. I made the highest SAT score in my fairly large school. Nobody said anything about it except my guidance counselor who said she should pinch my head off for making the grades that I did.
Reminds me of the story of the guy taking a university exam. At the end, the prof says “time’s up. Everybody put down their pens now and hand in your paper.”
While others are shuffling up tot he front and piling up their papers, one student is furiously still writing. At the very last he comes up with his paper. The prof says “I told you to stop writing afew minutes ago. I’m not accepting that paper.”
The student puffs up his chest “Do you know who I am?”
The prof says “I don’t care who you are related to and I don’t know who you are.”
“Good!” says the guy, stuffs his paper in the middle of the pile and runs out.
I assume with an online test it’s one question per screen and then it disappears as soon as you answer. (Most of the tests I’ve done). So its less likely someone was copying, it would take a bit of effort to stay up with the neighbour and be on the same “page”.
One online standardized test I’m familiar with, the MAP (Measure of Academic Performance), if you answer questions too quickly, it’ll assume you’re just blindly guessing, which doesn’t give a good indication of student skill, so they’ll pause your test and make you call the proctor over so they can unpause it for you.
If that is a substantial deviation from typical timeframes (eg: if no one has ever completed, or at least never achieved marginally better than “chance” results, on the test in less than 45 minutes) then that seems like evidence of cheating to me. Not absolute proof, but evidence.
IF it was a case of too many answers matching, then I assume the questions were the same. Also, if A gets one set of questions and B gets a different set, an argumentative applicant could say the results may not be truly comparable… “they got the easier questions, I didn’t!”
Or the multiple-choice answers in a different order. If the answer choices on test form A are
a) Iron
b) Gold
c) Nickel
d) Tin
and on form B, they’re
a) Gold
b) Tin
c) Nickel
d) Iron
, and the person next to me correctly answers gold and I answer tin, and so on down the line, that looks pretty fishy for me.
This was the complaint once upon a time about cutural bias.
“Bob has enclosed his yard with a fence all around except the back edge, and a gate 4 feet wide - the yard is 50 feet wide and 40 feet deep. How long is the fence?”
If this is on an IQ or SAT or whatever and the subject of the test has never seen a yard with a fence, will he be more likely to get the question wrong? A similar point was made about a “read this and answer the questions” problem; there was a reading comprehension paragraph describing the Army ranks and hierarchy, followed by questions about the various ranks and positions. Who is more likely to finish that set of questions faster, boys or girls?
Not all questions are as blatant as that, but it’s easy to see how one set may be considered more challenging than another while those who rank them may not pick up on the bias, or they bias may exist in the subject group used for the scaling. Bad enough the same questions may be biased, it’s worse if they are different questions. Someone could argue - “I got the more difficult questions and the scaling is moderately subjective.”
I would think - same questions, different order, would be the preferred anti-cheating procedure that minimizes bias.
The only other fair process might be - the randomization makes up, say, 50 different tests, so which one anyone gets may be random, but there are only 50 sets so any person’s score is comparable to hundreds of thousands of other test takers in the country who got the identical test. If among tens of thousands of people taking each test, presumably a broad random cross-section of test takers, one particular test version is statistically different from the others, that may indicate more scaling adjustment necessary.