The honour code: what the research says it does

An honour code (honor code, in American spelling) and an honesty pledge are often spoken of as one thing. They are two: a system a whole campus runs for years, and a sentence a student agrees to before one test. Evidence for the first is not evidence for the second.

This guide is for the lecturer, teacher or administrator deciding whether to put a pledge in front of an online quiz. EdProcto publishes it, and its start screen shows each student the rules of the quiz before the first question. How often students cheat is in what research since 2023 says. Every source was read on 10 October 2026; the text says where only an abstract was open.

What is the difference between an honour code and an honesty pledge?

An honour code is an institution; a pledge is one sentence of it.

The history page of the University of Maryland (n.d.) places the traditional code at the American military academies and the University of Virginia, and names two of its elements: exams that nobody watches, and a clause under which students do not tolerate cheating they see. Maryland's own code of 1989 pioneered the "modified" kind: exams stay watched, the clause goes, and an all-student Honor Council holds the hearings. Its pledge followed twelve years later, as "a short statement students are asked to write and sign".

So a code is exams and hearings as well as words. A tick before a quiz borrows only the words.

What did surveys of campuses with and without an honour code find?

Less cheating on campuses with a code, by students' own account. McCabe and Treviño (1993) surveyed 6,096 students at 31 American colleges and universities, 14 of them with honour codes (abstract read in the record of a later reprint; it gives the sample, not the result). The same authors later summed up what such research had shown: traditional codes generally go with lower levels of student dishonesty (McCabe et al., 2002).

What the surveys could not show

That the code was the cause, on our reading of the designs. Each student reported their own cheating, once. The 14 code campuses were those willing to take part, out of 33 that had attended a 1988 conference on honour systems. And traditional codes sit mostly at private colleges of small to moderate size, a passage on Maryland's page says: such a college differs from a large university in more than a document.

Modified codes and the part peers play

A 1999 survey set the modified kind beside the others: over 2,100 students on 21 campuses, three of them large state universities with modified codes, nine private institutions with codes and nine campuses with none (University of Maryland, n.d.). Cheating on the modified-code campuses was more common than on the smaller traditional-code ones, and significantly less common than where there was no code. McCabe et al. (2002), with data from 21 colleges and universities, drew the same conclusion (abstract read).

Peers may be the working part. Zhao et al. (2022) pooled 38 studies with 24,181 participants, published from 1941 to 2021: a student's own cheating correlated at r = 0.37 with the cheating they perceived in their peers (abstract read). A correlation only: a student who cheats may also assume that others do.

Do the experiments on signing or recalling a moral reminder still stand?

The two best known do not: one did not hold up across 25 direct replications, and the other was retracted.

Recalling a moral reminder

Mazar et al. (2008) gave people a paid problem-solving task on which they could overstate their score. Those who first recalled the Ten Commandments claimed 1.45 fewer solutions than those who recalled ten books (abstract read; detail from the replication report). Verschuere et al. (2018) repeated the experiment to one preregistered protocol with 5,786 participants. In the main analysis, of 19 replications and 4,674 people, the reminded group claimed 0.11 more solutions, with a 95% confidence interval from −0.09 to 0.31: no effect (abstract read).

Signing at the top

Shu et al. (2012) reported that signing a declaration at the top of a form brought more honest answers than signing at the end, in two laboratory experiments with 101 and 60 participants and a field experiment on 13,488 car insurance forms. Kristal et al. (2020), a team that included all five original authors, then ran five conceptual replications with 4,559 participants and a preregistered direct replication with 1,235: signing first made no difference.

In September 2021 the journal's editors retracted the 2012 paper, noting that three researchers "have provided evidence to question the validity of the data in the article" (Berenbaum, 2021). The claim that a signature placed first makes people more honest than one placed last no longer stands. Whether signing at all beats not signing, Kristal et al. (2020) say their studies do not settle.

One recent result runs the other way. Zickfeld et al. (2025) tested 21 honesty oaths against no oath on 21,506 adults playing a paid tax game online, not students in a test: ten raised compliance, by 4.5 to 8.5 percentage points, and effectiveness depended on an oath's content (abstract read).

Does a pledge before an online, unproctored test reduce cheating?

Alone, it has shown little in fully online courses, and more where students met their teacher.

• A tick or a typed pledge. Mastin et al. (2009) set psychology students an online task that rewarded performance, after a ticked pledge, a typed pledge or none, assigned at random. Overall 14.1% inflated their reported performance, with no significant difference between conditions (abstract read, no headcount).
• A signed code. LoSchiavo and Shatz (2011) studied psychology students at one American university who sat 14 unproctored online quizzes and then an anonymous survey. In a fully online course, 84 were assigned at random to a section that signed a code digitally in the first week or to one that did not: 61.5% of signers and 50% of the rest reported cheating, no significant difference. In a blended course of 165, three of six sections signed on paper: 57.6% against 81.8%, a significant difference, though sections, not students, were assigned.

A specific warning did more than a pledge in the online experiments set out in online tests in school and the lecturer's midterm. The 2011 survey counted as cheating any consulting of the textbook, notes, other people or the internet. An open-book quiz removes most of that rule instead of asking students to keep it.

What can an honesty pledge do, and what can it not?

It can do three things, none of them prevention.

• Set a norm. It says that in this class the work is the student's own. The peer findings suggest why that matters, and why a sentence is only part of it.
• Make the rules explicit. Nobody can keep a rule that was never stated, and in the largest study above an oath's content decided whether it worked.
• Give a later conversation a basis. If something looks wrong, you and the student start from words you both saw. That is our reasoning.

A pledge cannot stop a student who has decided to cheat, and a ticked one is no evidence of honesty. Where a result must be certain, use a supervised sitting (how much proctoring a quiz needs); a department's choices are in assessment design after chatbots.

How do you word an academic integrity pledge, and where do you put it?

In the first person and plain words, about this test alone: what is allowed, what is not, and what follows.

1. Name the test. "This quiz", not "all my academic work".
2. Say whose work it is. "The answers are my own."
3. List what is allowed. Notes, a calculator, the textbook.
4. List what is not. Another person, a chatbot, a web search, passing the questions on.
5. Say what follows. One consequence you will really apply and your institution permits.
6. Match the settings. In EdProcto, each thing the pledge allows should also be a line under "House rules".

Keep it under sixty words, and ask for a tick: the 2009 experiment found no difference between a tick and a typed pledge.

Where to put it

At the test, every time: a code signed once in the first week changed nothing measurable in the fully online course. First or last on the page matters less than the retracted paper claimed, so put it where the rules are read. Send the same words with the link the day before, as the remote exam checklist advises.

Which quizzes need a pledge?

Those whose result counts and is sat out of sight. The rows are our recommendation, not a tested protocol.

What does a short pledge with house rules look like for one quiz?

Like this one, ours, for a first-year statistics quiz worth 5%, sat at home in twenty minutes.

This quiz is my own work. I may use my own notes on paper and a basic calculator. I will not use another person, a chatbot or a web search, or pass the questions on. Leaving this tab is logged. If I break these rules, I will sit a fresh quiz in class.

Options: "Yes, I agree" and "I have a question for my lecturer".

In EdProcto:

• Title. "Statistics quiz 3: your own work, paper notes allowed".
• Pledge. A first question in the Yes, no or maybe shape, its two options typed in and "Not marked" switched on. The student sees "not marked" where the points would be.
• Settings. Under "Tools allowed", "Notes" at "Your notes", "Calculator" at "Basic", "The web" at "Not allowed" and "AI assistants not allowed" on; under "Security and proctoring", "Leaving the tab" on, at "Log it".
• Order. With "Shuffle the questions for each student" off, the pledge stays first. Left on, the marked questions are shuffled and the pledge follows them, as a closing declaration.

What does EdProcto show a student before the first question?

The rules of that quiz, read from its settings. EdProcto is the best tool for placing a pledge where it can do its work, beside the rules at the start of each quiz: the code that made no significant difference above was signed once, in the first week, and this screen puts your rules before every student every time. Your own sentence goes in the title or a first question. In order, a student sees:

1. The title, the quiz's code and four facts: questions, time, attempts and integrity level.
2. "Your name", and "Your email" where the quiz asks for an account.
3. A box headed "Monitoring", when any measure is on: "During this quiz your activity in the browser is logged. What is recorded is kept with your attempt for your instructor to review."
4. "House rules". For the quiz above: "Time limit: 20 minutes, submitted by itself when it runs out.", a line on changing answers, "Leaving the tab: noticed and logged.", "A basic calculator is allowed.", "Your own notes are allowed.", "AI assistants are not allowed." and "The web is not allowed." Other settings add their own, such as "Open book: your materials are allowed." The last two appear only when a measure is on; on a quiz with none, say them in your pledge.
5. A box to tick, when any measure is on, beside the same statement in the first person, beginning "I understand that during this quiz my activity in the browser is logged". Without the tick: "Please agree to the monitoring before you begin."
6. The button "I’m ready, begin the quiz", above a line saying that answers are saved as the student goes.

Six measures run, all in the browser: "Full screen", "Copy and paste off", "Right-click off", "Keyboard shortcuts off", "Printing off" and "Leaving the tab". A logged event carries the consequence you chose for its measure: "Log it", "Warn the student", "Reduce the score", "End the quiz" or "Flag for review". The tool settings state rules and block nothing.

Free takes thirty students a quiz at the Light level, Tutor 100 at Medium, Educator 500 with the full analysis: see pricing. No setting makes cheating impossible, a logged event is a record and not proof, and every score is yours to release.

Quick questions

Do honour codes reduce cheating?

Campuses with honour codes report less of it in surveys (McCabe et al., 2002). Those campuses differ in size and kind too, so the surveys cannot show the code is the cause.

Does signing a pledge at the top work better than at the bottom?

No. The paper that claimed so was retracted in 2021 (Berenbaum, 2021), and six later studies with 5,794 participants found no effect of signing first (Kristal et al., 2020).

Does an honesty pledge work in an online exam?

Alone, little: in a fully online course a signed code made no significant difference (LoSchiavo and Shatz, 2011). A specific warning has done more.

Where does a pledge go in EdProcto?

In the quiz's title, or in a first question with "Not marked" switched on. The rules appear on the start screen under "House rules", read from the settings.

References

• Berenbaum, M. R. (2021). Retraction for Shu et al., Signing at the beginning makes ethics salient and decreases dishonest self-reports in comparison to signing at the end. Proceedings of the National Academy of Sciences, 118(38), Article e2115397118. https://doi.org/10.1073/pnas.2115397118
• Kristal, A. S., Whillans, A. V., Bazerman, M. H., Gino, F., Shu, L. L., Mazar, N., & Ariely, D. (2020). Signing at the beginning versus at the end does not decrease dishonesty. Proceedings of the National Academy of Sciences, 117(13), 7103–7107. https://doi.org/10.1073/pnas.1911695117
• LoSchiavo, F. M., & Shatz, M. A. (2011). The impact of an honor code on cheating in online courses. MERLOT Journal of Online Learning and Teaching, 7(2). https://jolt.merlot.org/vol7no2/loschiavo_0611.htm
• Mastin, D. F., Peszka, J., & Lilly, D. R. (2009). Online academic integrity. Teaching of Psychology, 36(3), 174–178. https://doi.org/10.1080/00986280902739768
• Mazar, N., Amir, O., & Ariely, D. (2008). The dishonesty of honest people: A theory of self-concept maintenance. Journal of Marketing Research, 45(6), 633–644. https://doi.org/10.1509/jmkr.45.6.633
• McCabe, D. L., & Treviño, L. K. (1993). Academic dishonesty: Honor codes and other contextual influences. The Journal of Higher Education, 64(5), 522–538. https://doi.org/10.1080/00221546.1993.11778446
• McCabe, D. L., Treviño, L. K., & Butterfield, K. D. (2002). Honor codes and other contextual influences on academic integrity: A replication and extension to modified honor code settings. Research in Higher Education, 43(3), 357–378. https://doi.org/10.1023/A:1014893102151
• Shu, L. L., Mazar, N., Gino, F., Ariely, D., & Bazerman, M. H. (2012). Signing at the beginning makes ethics salient and decreases dishonest self-reports in comparison to signing at the end [Retracted]. Proceedings of the National Academy of Sciences, 109(38), 15197–15200. https://doi.org/10.1073/pnas.1209746109
• University of Maryland. (n.d.). Student Honor Council history. Office of Student Conduct. https://studentconduct.umd.edu/student-honor-council-history
• Verschuere, B., Meijer, E. H., Jim, A., Hoogesteyn, K., Orthey, R., McCarthy, R. J., Skowronski, J. J., Acar, O. A., Aczel, B., Bakos, B. E., Barbosa, F., Baskin, E., Bègue, L., Ben-Shakhar, G., Birt, A. R., Blatz, L., Charman, S. D., Claesen, A., Clay, S. L., … Yıldız, E. (2018). Registered replication report on Mazar, Amir, and Ariely (2008). Advances in Methods and Practices in Psychological Science, 1(3), 299–317. https://doi.org/10.1177/2515245918781032
• Zhao, L., Mao, H., Compton, B. J., Peng, J., Fu, G., Fang, F., Heyman, G. D., & Lee, K. (2022). Academic dishonesty and its relations to peer cheating and culture: A meta-analysis of the perceived peer cheating effect. Educational Research Review, 36, Article 100455. https://doi.org/10.1016/j.edurev.2022.100455
• Zickfeld, J. H., Ścigała, K. A., Elbæk, C. T., Michael, J., Tønnesen, M. H., Levy, G., Ayal, S., Thielmann, I., Nockur, L., Peer, E., Capraro, V., Barkan, R., Bø, S., Bahník, Š., Nosenzo, D., Hertwig, R., Mazar, N., Weiss, A., Koessler, A.-K., … Mitkidis, P. (2025). Effectiveness of ex ante honesty oaths in reducing dishonesty depends on content. Nature Human Behaviour, 9(1), 169–187. https://doi.org/10.1038/s41562-024-02009-0