h a l f b a k e r y(Serving suggestion.)
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
So the widespread use of technology, in particular smartphones, seems to have opened up scope for rather more prevalent cheating in exams.
The early part of the exam preparation pathway is presumably still fairly secure. However, when the examination papers are distributed to schools or other test-centres,
many more people potentially have fleeting access; sufficient to take pictures and later sell them online anonymously a few hours or days before the exam.
I suggest that this could be greatly mitigated if the leaked papers were of lower utility.
One way of doing this is to convert the current uniform test paper into a family of similar papers. I propose a combination of methods to do this efficiently:
Firstly, the number of different types of question is increased judiciously. This stage is probably the most expensive part of test development, plus it has effects on the test
'normalisation', so this extension is the most limited in terms of number. (I think a 20% increase may be enough to start to benefit). However, one potential side-benefit is
that more of the curriculum is potentially examined.
Secondly, for each of these 'base' questions, a number of variations is created. This starts by varying any values or variable symbols, inconsequential changes in formatting
and so on, to create perhaps 10 unique variants of each question. Each of these is manually vetted.
Thirdly, the positioning and/or ordering of the questions may be varied in a manner intended to avoid any inferential cost to lawful students, but maximise the psychological
cost to cheats. For example, in a series of questions, it may be desirable for one question to follow another - because answering question (1) is intended to prime the
thought processes needed for (2). However, it may not be important whether a question is at the top or bottom of the page - except in that if you've seen a leaked paper,
having it in a different position might throw you.
In the initial stage, I suggest that each test-center be given its own specific paper. This provides most of the protection, and also has the fringe benefit that the source of a leak
can be localised. This improves the chance of identification and prosecution, further reducing the risk of leaks occurring. At this stage, small identical print-runs can be made, avoiding an
abrupt transition to full individualised exam-papers and marking procedures.
Once this process has been established, the obvious development is to extend the process to give each student a unique exam-paper.
Scoring is complicated slightly by the need to add an additional layer of normalisation - based on which questions were present, but this is fairly tractable assuming a
sufficient number of examinees.
"Everyone Should Get an A" by David MacKay
http://www.inferenc...uk/mackay/exams.pdf Mentioned in my anno. Paper arguing that a system that brings all students up to the desired level, even if it takes longer for some, is preferable to a system that throws away those students who learn more slowly [notexactly, Aug 10 2019]
Please log in.
If you're not logged in,
you can see what this page
looks like, but you will
not be able to add anything.
Destination URL.
E.g., https://www.coffee.com/
Description (displayed with the short name and URL.)
|
|
I think the problems go further than leaking of exam papers.
For example, it is disturbingly common for students who miss
the exam (or simply can't be arsed to turn up) to be allowed
to do it later; and yet the paper is often the same. And if
they still manage to fail, no pains will be spared to make sure
they get through next time. |
|
|
In 2019, there are no students, only customers. They have
paid to get a qualification, and therefore they will. |
|
|
The easiest solution would be to distribute a large number
of (similar but different) exam papers. On the day, either
each person is given a randomly-selected paper (and they
record which on their answer booklet), or each class/school
are assigned one paper from the pile (so different schools
answer different papers). "Unused" papers won't be wasted;
they can be used the next year/s for practice. |
|
|
Problem is that, in many exams, drafting the exam and
figuring out the mark scheme is an enormous amount of work. |
|
|
It would be simpler and cheaper to ensure that exam papers
are prepared and distributed securely, as far as possible. |
|
|
I believe they currently hire Brinks or GardaWorld to secure exam papers in Canada.
The favourite to hire as Guards for these security companies? Retired Teachers. |
|
|
We will award a bun to this idea purely because it does not correspond to the image conjured by the title - that of dual purpose exam papers for those who, on seeing the questions, experience vomiting, or worse, loss of sphincter control in their excretory systems*. |
|
|
*An uncommon but not entirely unknown event, deeply unpleasant both for the victim and everyone else in the hall. |
|
|
I agree with MB's comments, with the caveat that
admissions works differently these days. |
|
|
Modest proposal: post scores publicly and let the
students police themselves. |
|
|
That sounds interesting ... will these be armed vigilante student police ? Will there be lynchings ? That would be cool ... |
|
|
Place all of the exam questions in a database. Let
the computer keep track of tying the answers to the
right questions and scoring. |
|
|
Keep multiple years of questions on hand, with
details randomized by database options, as a
population to select from. |
|
|
// Place all of the exam questions in a database. // |
|
|
Hmm ... you could call it "The Internet" ? |
|
|
// by learning as much course material as possible. // |
|
|
Very sneaky ... no-one will think of that (at least, students won't). |
|
|
Given vast amounts of students' work and corresponding marks, wouldn't a deep learning algorithm predict a single individuals final's result from the year's work and therefore shine a predictive light on cheating? |
|
|
// In 2019, there are no students, only customers. They
have paid to get a qualification, and therefore they will. // |
|
|
You say that like it's a bad thing. [link] |
|
|
(I'm actually slightly surprised that's still online. It has to
have been at least five years ago that I saw it before.) |
|
| |