Methods to track behind cheating on exams using on-line sites?

Question

In the current state of the technologies, what ways can be used to track clear yet directly untraceable cases of academic misconduct specific to students who post questions from and view answers to a take-home exam using on-line (homework) sites?

My case is specific to Chegg, although answers related to other on-line sites would be certainly welcomed.

As a background, I have a list (from Chegg) of the email address, log-in times, university affiliation, and question posted / viewed. I can clearly correlate the posting information to being from my exam during my exam. I can directly correlate some cases because the email address is ... surprisingly ... the student's university email address or their personal email address. What I do not have in some cases is a) a valid or directly representative email address and b) a full name.

I have a roster list of students and have diligently checked their email addresses against the offending emails to no avail. I will pass the roster and list to our administration to review against their (larger) database of email accounts. However, I can only imagine that an email address could be from a neighbor or friend or roommate or ... someone entirely untraceable in the full university email database.

After a formal inquiry to obtain the above list, I have an email response (from Chegg) that tells me that I will get no further help to obtain any other information. I will be glad to hear about other levers that can be pulled to get better information.

I can accept that at some point I may have to conceed that I am simply out of luck to track the real perpetrator. But, before I do so, I have to wonder whether this community has insights to additional resources to help solve my problem.

To help also focus the discussion further, I am not interested to learn about the ways to avoid this problem in the future. I have my own insights and plans to move forward on this front. I am also not interested to spend time trashing Chegg or equivalent sites, deserving or not.

Finally, in addition to the specific question on additional effective methods, I am also interested in one other broader aspect of the problem. I am curious to hear from like-minded individuals who have initiated or are aware of actions being initiated for a larger academic community stand against the problem. I make this latter statement especially as we faculty are all most likely to have to face yet another round of on-line teaching for the coming Fall 2020 semester.

In summary, I hope to hear about approaches that others have used that could also help me solve my problem to track perpetrators of cheating who, up to this point, are essentially directly untraceable.

Answer 1

I can think of a few approaches that might help. I'm not sure pursuing the sites used will be very fruitful. There are always disposable emails they can sign up with, Tor exit nodes they can route traffic through etc. It's a fool's errand.

Oral interviews post-exam

Ask a random sample of students for an interview after they've submitted their answers. With the script in front of both of you, ask them about their thought process when answering the question and evaluate the extent to which they can confidently discuss the subject.

Cons: time-consuming, potential for false positives from students who forget most of the content after taking the exam/have other assessments to study for/don't cope well with the interview pressure.

Issue unique questions per student

If you have control over the online assessment platform, you may be able to issue unique questions per student to catch out blatant copy-pasting of the question text in order to publicly outsource an answer. I am not proposing that a paper is manually put together per student; this seems like something you'd want to automate.

Cons: the ability to do this well will vary from subject to subject. It might be easy to generate lots of linear programming problems of comparable difficulty if your exam e.g. assesses the student's ability to apply the Simplex algorithm. But, explanation questions may be more difficult and lead to a perception that some students unfairly get 'easier' or 'harder' papers.

Answer 2

To handle the current situation, I would have two suggestions:

1. Notify all students that you aware of cheating during the exam and that you are in the process of collating identifying evidence (be as specific or vague about the website, email address as you see fit). State the consequences for those who are found to have cheated but that these may be lessened if the students come forward on their own accord. This is unlikely to faze a determined cheat, but may bring forth others who made a critical mistake in the heat of the moment and importantly may know the owners of the email addresses you have been unable to identify.
2. Send an email to each of the addresses you have found along the lines of 1. letting them know that the address has been implicated. Again, this is really just a scare tactic but might bring a few of the unknowns to light.

I think if you have a well thought-out cheat who used a truly disposable email address and did not communicate their actions to any of their peers then you are unlikely to be able to catch them in retrospect. But, as you have already observed, not all students are that shrewd.

Answer 3

Invest your time in adapting your exam questions rather than searching for people cheating.

Searching for people who are cheating is bound to be a frustrating an time consuming endeavor. As you've already noted there are many methods which could be used (both online and offline), and I can guarantee you a devious student can out-cheat your sleuthing skills.

Think carefully about the purpose of the exam and what you are certifying with a pass/top-grade. For a minimum pass threshold, consider explaining your syllabus ("By the end of this course student should be able to... "), to the next teacher of your class and what you think they'd expect based on that. For a top grade you're probably assessing a more general skill, do they have good written communication, are they able to extract the core information effectively from text.

Depending on the size of your class, how well you know each student, and amount of time you're willing to invest you may want to pursue different exam strategies. In person interviews (via-video call) are often the gold standard, but they are time consuming, difficult to standardise and vulnerable to bias.

Consider 3 categories:

1. Minimum-knowledge questions, which you expect most students to get correct. You should probably be able to scrape a pass answering only these questions.
2. Long-form questions, where the best students can shine
3. Coursework, like an exam, but longer.

Final thought; be clear before the exam what the structure is, particularly if it has changed from previous years. Surprising students and testing exam technique is no fun for anyone involved.

EDIT: Writing good exams is a skill, but like any skill it's something you can get better at if you invest time in learning about it. There are plenty of resources, articles and books written on the topic and even reading a couple of articles will give you some good ideas. I included some examples below, but realised I was basically writing a book so I'll leave expanding this as an exercise to reader.

Some examples:

Statement questions e.g. state X's theorem, define Y
These questions are hard to cheat on, since it's usually just quicker/easier to know the answer. They can often be overlooked in open-book exams as 'you could just look it up in the notes', I would argue that that's fine. A good student will just write the answer down and move on, a poor student will spend time looking it up. In either case they will be able to do so in a real world situation.

Multiple choice These are hard to cheat on for the same reasons and above, guessing is not a valid strategy in non-trivial exams, so no need to worry about that. You can use these questions to quickly test linked knowledge (they have to understand the terms in the question and how the options relate to it). I would avoid "trick" answers and follow-up with an "explain" question.

Short-essay questions For small classes where you know the students these are hard to cheat on, because it should be obvious to you if the style is inconsistent. These are a chance for your best pupils to show their general written communication skills and their ability to apply the course knowledge to a specific problem. Be clear about how much text you expect (~X words) and keep them relatively brief, personally I believe long-essay questions are better suited to coursework.