Where do we draw the line between suspicion and evidence of cheating on an exam?

For example, on an online timed essay exam, in a student's essay, 99% of the sentences are identical to sentences found in an online essay (not written by the student). That's evidence of cheating, and warrants consequences, right? What if it's, say, 12%? That's (highly) suspicious, and does not warrant consequences, right? Where do we draw the line between suspicion, and evidence, of cheating?

I know there are academic honesty committees that are designated to deal with this question: so where do they draw the line?

I am looking for an answer that is:

  • justifiable
  • as non-subjective as possible
  • practical

This is a follow-up to Is it fair to have follow-up interviews after a test following suspicions of cheating?.

  • 13
    Would 12% (pick a number) be "highly suspicious" or just indicate the writer had a good memory for phrases and didn't try to reinvent the wheel during an exam ? There's a difference between that and cheating. Nov 29, 2022 at 14:53
  • 2
    @StephenG-HelpUkraine Yes, that is pretty much what I'm asking: how can we tell if it's cheating, or not cheating"?
    – Dan
    Nov 29, 2022 at 15:00
  • 4
    @chepner Depending on the field, certain sentences and phrases are used extremely frequently in literature. Fragments like "Thus we can say without loss of generality, that..." and "Flintstone and Rubble (300,000BC) claim that..." and "the Health Sciences Center at the University of Northern Iowa" can easily account for 12% of an article/essay.
    – shoover
    Nov 29, 2022 at 17:46
  • 3
    Do you permit them to do "research" when writing an essay? Have you specifically taught them about plagiarism, what it is and what it isn't? Or is this an entrapment mechanism?
    – Buffy
    Nov 29, 2022 at 18:15
  • 6
    @chepner That's nonsensical. It's an exam, and not writing a paper. If I remember a paragraph-long definition from a textbook and render it exactly, word for word, in response to a test question, that's not plagiarism, that's "giving the right answer".
    – tbrookside
    Nov 29, 2022 at 19:35

2 Answers 2


There are basically three possible burdens of proof:

  • Preponderance of the evidence. This is what is used by the University of California. It needs to be "more likely than not" that the misconduct occurred.
  • Clear and convincing evidence. This is what is used by New Mexico State University. It needs to be reasonably clear that that the misconduct occurred -- i.e., not just more likely than not, but substantially likely.
  • Beyond any reasonable doubt. This is required for criminal convictions in the US, but from a quick search, I cannot find any universities that use this standard.

So, either of the first two seem to be justifiable. Personally, I prefer the second, but it probably varies by school; some schools have so much cheating that more drastic measures have to be taken.

The key point, though, is that the burden of proof and the investigation procedure is to be worked out in advance by the institution as a whole. In my answer to your linked post, I advised you against single-handedly inventing an investigatory procedure after an exam had taken place. My objection to your scheme was not that you were being "too unfair"; rather, my objection was that individual instructors should not be inventing schemes on-the-fly. Your institution's policies should provide guidelines for how to approach such cases. For example, under the University of California's guidelines, it looks like you would be allowed to subject your suspected cheaters to a technical interview, but you would have to offload your findings to a third-party committee for a determination of guilt, rather than playing judge and jury yourself.

  • 16
    I'd be curious the history of these choices. 'Preponderance of the evidence' is a frighteningly low standard for something that could destroy your future.
    – Chuu
    Nov 29, 2022 at 16:05
  • 1
    Actually, UCLA's policy (linked above) states that even in the most serious cases, the burden of proof is merely preponderance of the evidence. Naturally, the dean will use their discretion, but still. The "history" of this choice is probably from US Civil Law -- and it makes sense in that context, since the two parties in a civil case are on roughly equal footing (in theory). But cheating investigations are closer to a criminal matter than a civil one, so personally I feel that a higher standard of proof (such as clear and convincing evidence) is more appropriate.
    – cag51
    Nov 29, 2022 at 18:34
  • 2
    More likely than not is quite ridiculous, actually. Just have ppl. literally dice roll (d6) their grades and arrest all of them for cheating as it was more likely not to roll that number.
    – DonQuiKong
    Nov 29, 2022 at 21:31
  • 3
    @DonQuiKong It is unfortunate that the preponderance of the evidence standard is often described as "more likely than not", that is a completely incorrect summary of what the standard actually means. But you're right that if it really meant that, it would be ridiculous. For example, if there were a university where more people cheated than didn't, simply showing that statistic would be sufficient to prevail if the description as "more likely than not" were actually correct. See Smith v. Rapid Transit for more. Nov 30, 2022 at 15:01
  • 3
    @Chuu The classic example: A person is hit by a bus. All they know is that it was definitely a blue bus. They sue someone who operates 70% of the blue busses in the county. This does not prove that the defendant's bus hit the person by the preponderance of the evidence standard because it cannot possibly make someone believe that the defendant's bus did in fact hit the person regardless of how much they credit the evidence. This would never even get to a jury because there must be proof to subject to the preponderance standard so there's no input to produce output from. Nov 30, 2022 at 17:04

If a professor strongly suspects a student of cheating but doesn't have solid proof, I think it's better to do more investigation and follow up instead of just reaching a "yes" or "no" conclusion:

  • The professor can set up a meeting and ask the student a few questions about the assignment. Like if it's an exam, to further explain some of their answers. If the student then has no idea what they wrote it is likely they are cheating. Downside: see the related post, Is it fair to have follow-up interviews after a test following suspicions of cheating?

  • When multiple students are suspected, some professors send out an email to elicit confessions: something along the lines of "we know that some of you have cheated. If you confess, you will get a 0 on this assignment, otherwise we will give you a 0 for the class". Downside: not all students who've cheated are known, not all will confess, and those who are unknown and haven't confessed will get away with it. IME a lot of students are well aware of this and won't take the bait

Personally I don't like those approaches. What I recommend is, if you strongly believe a student has cheated but don't have solid proof, let them get away with it, but ensure they are monitored much more closely next time. The reason being, students who cheat rarely only do so once, and if a student is cheating to keep up in your class, they likely won't keep up once the cheating is no longer possible. Especially if the assignment in question is not really important in the final grade. Downside: there's still the possibility that the student shapes up for future assignments and just gets away with it. But the fact you suspect cheating but can't confirm it means that, besides potentially prosecuting an innocent student, you don't have many good alternatives.

  • 16
    I very strongly advise against "sending out emails to elicit confessions" as laid out above: It punishes the honest ones, rewards the dishonest and undermindes any future action you take. Student A who has cheated confesses and gets a 0, while Student B who also cheated but does not confess receives full grades. You made being honest a bad choice furthermore, students might "compare notes" and notice the pattern and thus your bluff - and thus be even less inclined to be honest in the future.
    – CharonX
    Nov 29, 2022 at 15:28
  • I do not think the situation in the first bullet is the same as in the linked question, because in the linked question the professor wants to test the students on just any material related to the course, in contrast with just ask them to explain their test answer.
    – Yanko
    Nov 30, 2022 at 2:23
  • @tarzh, Your answer mixes good and bad advice (even you yourself say that you don't like half of what you suggest here). For a more useful answer, I suggest that you either delete the parts that you don't like (and probably many voters don't like either), or if you want to leave it, then split this answer into two answers: with one answer only having the parts that you like (and that would probably get more upvotes). You are allowed to give more than one answer to the same question.
    – Tripartio
    Nov 30, 2022 at 22:37

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .